Adrian - I built many desktop/transportable systems back then (including this board) and also managed this very email system. That board is what we called a 286/386 motherboard. It could come with DRAM sockets and/or SIMM sockets, and a choice of 286 or 386SX CPU. The board would barely handle 33MHz, and failed with a higher FREQ. the 386SX-40 is likely because of availability and later, pricing (40MHz was cheaper than the 33MHz part). The kludge board (early models) near the CPU was for use with a 286 CPU, and is bypassed when using a 386SX by that circuit. If it was a 286CPU, the other socket was empty or similarly bypassed. The PAL/keyboard and BIOS issue is due to differences between how the keyboards connect on 286 and 386. Many failures were "fixed" by replacing that PAL, and not the keyboard BIOS. And yes, changing from Phoenix to AMI would often (in those days) require swapping both, and all three on this 286/386 contortion.
Ah so that means there is little hope for this board as it sounds somewhat bespoke and customized for AT&T/Lucent. Clearly I won't find a better PAL and without a working BIOS there isn't much to be done.
@@adriansdigitalbasement AFAIK it is sometimes possible to read OTP PAL chips, and it may be possible to work around with some effort... but I wouldn't keep my hopes up, probably not worth the effort. Though as someone else mentioned, might worth trying again and saving with ESC instead of F10...
I had a look at the BIOS and it is custom, where the hard drive information is normally stored the BIOS is looking for what appears to be a date and time. The Keyboard protocol seems standard nothing out of the ordinary.
Networking gear is another great source of surprise PC re-usable equipment. An example was about 17 years ago while cleaning up old equipment I noticed an old router oddly had a VGA port, serial ports, and PS2 connector. I cracked it open and found a tiny motherboard sporting 6 network ports, an IDE connector + 2.5 HDD, one DDR slot, and an old Cyrix VIA chip. I connected up a dual connector IDE cable and a CD-ROM and the thing booted off of an XP CD. I had no problem installing XP on the little sucker. XP crawled but it ran. I put in the biggest DDR memory I could find - 512MB - and wound up transferring our old keycard system onto it from an old dying dust packed tower that was sitting under a desk in the basement for the last 5-6 years. I screwed that little system to the wall and it ran like that for another 10+ years before finally a complete security system upgrade.
I ran into a similar situation with a small form factor pc used as a controller for a remote access KVM device in a server room. It was an AMD K-266 with 128mb RAM. I acquired a scrap pass to bring it home and ran it as a firewall/ router running Coyote Linux booted off a floppy.
I’d put it back to the stock config and connect a terminal or emulator to a serial port. At a previous job, We had an AT&T audix voice mail system that was similar, and it ran AT&T Unix and expected a serial console not a vga with keyboard. At my current job we’ve got some modern server running Linux that are configured the same way, the primary console, including post and bios setup, are directed to a com port.
Beat me to it. :) I'd imagine that when an official repairman worked on it, he had some kind of handheld serial monitor that he could plug in to talk to the hardware.
What he said... most of the ATT & Lucent are expecting Unix and not screens. Hook up a terminal via serial with the hard drive attached. As I remember it, it was 9600,N,1 but I probably have that wrong.
That's exactly how it worked. Those two cards plugged into the phone system controller and would have been treated as extensions in the system. From there, the users would access the system through their phones, meanwhile the operator would have a serial console at their desk where they could manage the system as a whole. downloads.avaya.com/css/P8/documents/003677613
Yep telecoms stuff is often RS-232 and a further hint is the lack of a video card. A field tech wouldn't likely be dragging around a video card and monitor. Good luck.
I had to administer a similar voicemail box back in the early 2000s. It was running aDOS based system and was not friendly to configure. I'm so glad those days are gone.
dude, that thing is worth a fortune! us telephone collectors that still use Merlin Classic systems pay good money for these since they are pretty rare! You should offer it up on eBay or something so it can go back into service! These systems were introduced in the mid 90's. The build sticker is likely the date it was rebuilt by Avaya, which im suprised they didn't rebrand it since Lucent was gone by 2000 or so unless this was done right before they spun off into Avaya from lucent. These things were rock solid, and likely the reason it was refurbished is because the hard drive failed. These cost about $10k when they were new. This would wedge into the merlin system using extensions. The system would be programmed to forward a call to a VM port (one of those 2 cards) where the VM system would take over. The end user would call an extension or dial a code (depending on Classic or Merlin Plus system) and the system would prompt for the VM box and passcode. It was a highly advanced system for its day. My Avaya IPOffice 500 from 2008 or so (which is still sold today) does everything this huge box does on a 16gb CF card!
The last analog phone system I supported was a System 75. It was bought in the early 90s and lasted until 2005. I miss the Charlie's Angles speaker phone.
We had one, called Smooth Operator, when I was working in the early 90s; they had specifically configured it so we couldn't use it as any kind of standard PC. I seem to remember it coming up in the conversation with the sales people at the time. So it is implemented as security by obscurity specifically to stop people trying to use it as a normal PC.
A lot of people are mentioning the fact that the "serial port/console" is your way in! I agree. Probably the reason for this weird keyboard setup + the keyboard connector was not accessible from the outside. I wouldn't be surprised that they have created some weird serial keyboard (or special console device) for their technicians
A program on a Toshiba Laptop should do it. (some laptops did not carry the correct pin out or line voltage to work the programming card on phone systems, same could be on this).
@@highpath4776 could be also a special serial port, I work on several devices whose serial port required a special cable, but I would say knowing most probably programed though phone system
Adrian - in the setup.com program you were supposed to hit Escape, F10 was just to exit the program. Escape was reboot, which I assume would apply the settings.
@@NLDXBOX me as well. As another commenter mentioned, I would also like to see what happens when it boots off of the hard drive that was in there. If it even boots at all from it.
Check out the help menu as well for other commands. Looks like it never saved or it's wired to reset it self. Try to change the controller board placement, could be an IRQ conflict thing. Or change the controller and other boards. Try to unpatch the cable under it.
Yeah, and if that fails, I remember that there was also a utility for these PhoenixBIOS chips that could lock the configuration. It would have been a separate utility that would 'toggle' the lock at each run. Back in the 80386 era, I had a computer with the exact same BIOS as this one, only it was accessible through F2, as he showed. You could also use the supplied disk to access setup. The disk also contained this other utility to protect access to setup. You could even set a password for it, but that wasn't required. If protection was enabled, the 'Press F2 for setup' message disappeared at boot. Therefore, it wouldn't save the config if you tried to edit it using setup.com.
We had one of these at the company I work for until about 2012, when we went to IP phones and got rid of our Avaya PBX. Which BTW, was so reliable we never had a single problem with it to my knowledge in the 10 years it was in operation. The "Audix" voicemail PC was identical except it did have a front panel with a floppy drive and an optical minidisk for backup. Once I had to take it apart and reseat everything and as far as I can remember, the interior was identical. We managed the system with either an amber screen terminal or a piece of software that replicated all the features and connected to the PBX's IP. *Edit: Reading the other comments on its value, I really wish I'd rescued it from the recycling once it'd been retired.
Hmm. The original BIOS that replaced the IBM PC BIOS code was Phoenix, it was a "cleanroom" rewrite and not a copy. Back in those days, things were a bit different. The configuration program being built in to the EPROM was not there in the early days. If it has it you can hold down a key during startup and you will get a keyboard error and the option to press F2 to enter setup. The parameters were maintained by battery backed CMOS, not EEPROM or Flash so no battery and you lose configuration. I would guess that little circuit was to maintain the CMOS memory? Those ISA interface cards would have been a small fortune back when that was new.
Terrible how I have forgotten most of what I knew. And it took be ages to learn. And the manufacturers changed the systems and next sales point in so had to learn the new tricks.
Brook Trout was the company that made chips that say "Bt" on them. They were used a lot in TV Tuners and I think DAC chips for video cards. They were purchased or merged with Conexent which was originally part of Rockwell, who also made Modem and Telecom chipsets.
I used to be a CallWare Reseller/Installer. It was a Voicemail system that ran on top of Novell Netware and we used a lot of the Brooktrout boards. Those were very common in-band signaling board (that means the phone system sent a hook flash then the DTMF code for what type of call (ie. Voice Mail Retrieve, No Answer Message, Caller Busy Message, etc) and then it sent the extension number that it was coming from and then another hook flash, and transferred the call. There were all kinds of systems and each brand had is own codes and methods. Some were well documented others we had to use a box called a digigrapper to monitor the line and see what codes it was sending and do a bunch of test calls. There were also systems that used out of band signaling as well like SMDI ( a seperate cable) or TSAPI (A network protocol that signaled across IPX/SPX). You could run the non DSP cards for up to 4 ports but (each one of those RJ11's is wired for two lines), after that you had to get the DSP version to off load the voice processing from the CPU. As we started going to faster processors that became a non-issue. Then VOIP started to take over the world and it all changed.
This thing was running DOS 4.0 and had a single executable and autoexec on it. I remember seeing one of these in use a long time ago and there were nothing but RJ11s between it and the PBX, so I'm assuming the DTMF signalling you mention was used. Another comment said this was available as a 286 as well which was very slow and struggled. Hmm
@@adriansdigitalbasement I am cleaning out my storage unit that I have been backing full since 1994. I am pulling a lot of the boards ram, keeping a few cases and stuff. If there is anything in particular you're looking for let me know. The vast majority of it is Pentium 1 and up. I scrapped the XT/AT stuff years ago, which I am kicking myself over now.
14:36 9506 is the date code on the Dallas RTC, so it's original. Also, the case says "refurbished". It was an old machine from 1996, they replaced the hard drive and the power supply in 2000, then sold it as refurbished. It would have been wise to replace that RTC chip as well...
@@frogz I'm not sure if it's still running. The BIOS utility said that the RTC date is 1989. It might be caused by the utility was unable to read the SRAM/RTC data. But if you ask few hundred dollars for a refurbish job, yes, replacing a 5 year old RTC battery is a minimum. Okay, the expected running time is around 10 years according to the datasheet (they don't say it's guaranteed!), but still. A working RTC in such equipment is quite important.
Those little isolation transformers are very useful to make your own passive ‘DI Box’, useful when your computer’s sound card is picking up any ground loop humming. Your video has very excellent audio, but you never know when a ground isn’t appropriately connected to the local wiring and you start picking up ground loop noise. Reminds me of my first PC, a Pentium 3, 300 MHz. It’s a worthwhile educational toy if you stick with it and poke around and learn to be resourceful. The 8086 book became a favorite of mine. THANKS FOR SHARING!!
It's very likely that the motherboard is setup to run with a serial console... I'd try plugging in a terminal or serial cable into another machine with a terminal emulator... also if I remember correctly the older at&t voicemail systems from the early 90's were unix based and used a serial terminal vs. a vga monitor and keyboard. Looking at the data on the hard drive might give you some clues as well.
Hi Adrian !! This Dallas clock chip have a battery embedded !! Just open this IC, remove the battery and replace it with 2 alcalines AA socket/support !! I did it very very times in the past !!
Brooktrout is one of the few companies that make Analog voice boards for VoiceMail and Fax. Each of the two boards has two Analog ports, as you pointed out. These do not have dialtone but will answer a ringing line, and are just an audio circuit (usually) and can send and receive DTMF, etc. The VM communicated with the phone system over one or more serial ports (RS-232), those handled the signalling of what port to answer,a nd with what, light Message Waiting lights, etc. Common serial settings were 9600 8N1, try hooking a serial console to them! It looks like the original build was 1994-1996, and was factory refurbished in 2000 as indicated by the sticker on the box. I never worked on the Merlin Mail, but AT&T commonly used System V Release 3, 4 or 5 of AT&T Unix, or MS or other DOS as the OS. All of your bios control is probably done through Serial, or hard coded at the factory to ignore the keyboard. You can drop the HDD, and Brooktrouts in any 386 or 486 motherboard and get it running probably, did that on quite a few VMs in an emergency.
Hi, fantastic video so far. I just paused to reminisce. Anybody who was buying PCs in the late '90s will remember the fact that they all came with modems installed, as telephony was being exploited to the max in those days, particularly in the consumer space. And the fax a voicemail systems in those home PCs was starting to rival those of expensive office setups like this, with individual mailboxes for each member of the family, etc. OK, old man wistful thinking ended lol
If these chips can switch between internal battery and getting power from the PC, being in a machine that was turned on 24/7 for decades, there's a good chance there's a few years worth of juice still in there!
I've found several reasonably nice PCs out on the curb waiting for trash pickup. Usually, all I have to do is press the key for the restore utility while it's booting. People mess up the OS and the software and throw it away, but there's nothing wrong with the hardware and they must not realize that there is a built in factory-condition restore. Every now and then, I can tell they were poking around inside especially if the BIOS reset jumper is in the wrong position or some cables are unplugged. Hardly anybody bothers to remove or erase the hard disks.
You were right when you said the cards had analogue trunk lines, that's exactly what those are. They're also known as FXO, where the O stands for office, as in central office, because they act like a telco's central office (AKA telephone exchange), and act like an analogue phone line. The DSPs are analogue to digital and digital to analogue converters (ADCs and DACs), digitising the voice of people leaving messages and playing back recorded messages. There would also be a CODEC chip for encoding/decoding the digitised audio to/from a more compressed format to save space on the hard drive. The chips at the rear of the cards look like SRAM, which would be used to buffer the digitised samples for playback. Basically the CODEC chip takes the audio file and decodes it to the raw digital audio samples and stores them into SRAM, which the DSP would read and convert to an analogue signal. Modern systems don't need hardware CODEC chips, it's all done in software these days, but old systems like this wouldn't have been powerful enough for the CPU to encode and decode the audio in real-time. This only became possible in the Pentium MMX era when Intel introduced the SIMD instructions which allowed CPUs to encode and decode audio faster than real-time. That's why MP3s didn't take off until the mid-late 90's, because before then PCs weren't powerful enough to decode the audio for real-time playback (without hardware assistance anyway).
Brooktrout! Man what a blast from the past. They used to make cards for faxes and voicemail systems. Way back in the 90s I had to maintain a bunch of OS/2 based voicemail systems, and they all used Brooktrout cards. Most of those systems were tied into Mitel PBXs, and had an external interface box to do the trunk line signalling. But the brooktrouts did the actual voice line processing. Pretty cool cards, the ones I dealt with did on board DTMF recognition, and could play back a decent set of PCM sample rates. We would use the Voicemail systems to do all our IVR work as well. I wonder if Asterisk or some other soft PBX still supports them? Also... plug the serial board back in and see if the BIOS does serial redirection... most Telco stuff had service done thru the serial port.
I once trashpicked an analog pbx system from an e-waste bin, ended up selling it minus the case and the power supply for $900 on ebay - best trashpick of all time!
As others have said, these things are usually controlled by the vendor over serial. A standard modem is usually connected to a COM port. They may have done something to allow the keyboard to be controlled over the modem for more control.
That would be my guess. A old dumb terminal like a VT-100. Would also explain the 2 rs-232 ports. I had a older AT&T system that only had 1 rs-232 for a terminal or you could dial into it. If you called the correct port directly and entered the right DTMF code you would get the modem. That could be why one of the cards has a DSP chip on it.
I have a Merlin 820 KSU and 8 Merlin voice terminals (that’s what they called the desk phones) installed in my house and I love the system. It even integrates with VoIP and my mobile phone. Amazing for a system that was designed in 1984!
That's pretty cool. I guess they knew IP phones were coming sooner or later, what with the ISDN buzz, and wanted to be sure they didn't have to make a whole new system just to accommodate it later. What kind of mobile phone integrations does it have?
@@kaitlyn__L This era, some of them would optionally notify your pager you had a desk call when you were away. That was a pricey ($3,000-5,000 IIRC) option. And that was hot back then. I kid you not.
Kaitlyn L The nice thing about the Merlin system is that just about anything can be used as a trunk line. For my mobile phone integration I just have a small unit that connects to my phone via Bluetooth, but also can be wired in to the KSU with just regular twisted pair. I can make and receive calls via my mobile from any Merlin voice terminal. It’s called the “Dock ‘n Talk.”
I mean obviously there are security features to prevent snooping of recorded calls! Also the 386SX is not a fancy 286. It has a virtual 8086 mode like all 386s and much more sophisticated memory management. It has a 16 bit external interface but is 32 bits internally. You should boot the thing with an external serial console (e.g. Putty) connected to COM1 at 9800,8,N,1. That will almost surely bring up an admin interface to the Merlin system. Don't forget the null modem.
If you still remember SEAGATE disk manager, there is an option to write a small boot overlay into boot sector, and override the Type 48, in order to use a large hard drive. Brooktrout was famous of making voice and fax (back in the old days we also have the GammaFax multiline fax card), may be running DOS 5.0 and may have TSR to do background voice handling (from the GammaFax experience I got). The daughter board should be a replacement of oscillator, this board is just a bit newer to use DIMM instead of DIP (pin).
We had a similar Vocemail system at work up until around 2 years ago, and was even older (around 95-95) and ran NT4. This was being used for around 2000 users and we only replaced it when the hard drive decided to start failing (Luckily when it decided to fully fall over it did it on the morning of replacement).
That is possibly a Quadtel BIOS. Phoenix bought Quadtel in 1992, so if this ATT/Lucent product existed before 1992 it would have been a Quadtel BIOS (later rebranded as "Phoenix" after 1992). Try "Ctrl-Alt S" for Setup.
FXS (Foreign Exchange Subscriber) is the port that actually delivers the analog line to the subscriber. In other words, it is the “plug in the wall” that delivers a dial tone, battery current and ring voltage. ... FXO (Foreign Exchange Office) is the port that receives the analog line. you can use them on a POTS lie.....omg i'm dying here
I presume you can also feed a POTS or similar phone into it, for listening / if fax, faxing in /out and also programming mailboxes for voice and fax 2 email or similar
Wow, had to click on this one because my father used to work for At&t and then Lucent. Thought it would just be for lulz, but then I remembered one day going to work with him about 20 years ago (I had the day off school or something...), and I "helped" him upgrade very similar looking machines from their operator pool. I had no idea what we were doing at the time, but we were swapping out EPROM chips in motherboards like this. I realize now they were probably the proprietary bios chips they were using either in machines like these, or their operator terminals. I remember him telling me later they used some proprietary *nix variant. Don't tell Ma Bell, but I kept a handful of the chips we replaced just to play with. Unfortunately, if they didn't get tossed, they are on the opposite side of the country in my parents basement so I cant just check them out... but thanks for the memories!
You CAN get custom BIOSes with custom initial parameters for when the CMOS battery goes on the fritz, but the system won't be crippled. 25 years ago, I worked for a company that churned out big LED matrix displays for advertising and they used a PC with such a BIOS so they could work without a keyboard and a display and even if the CMOS died. (It booted DOS on a floppy that contained the advertisements to be displayed; the floppies were mailed each month to the branches with new advertisements).
I used to work at the Lucent factory/Bell labs in Westminster CO where they assembled these merlin boxes and Definity switching cabinets, I was there 1998-2002 .i worked for IBM supporting Lucent desktop machines and servers. I believe this was a Lowtech security feature where you had to connect a modified keyboard or special keyboard that would allow you to interact with the BIOS. Basically to prevent someone ,other than a technician , just plugging a regular old keyboard and messing with the box. Also, what other said about the rs232, I think you can get console access with a tty terminal and see the exported video, like old school serial over Ethernet, so they could remotely dial into the box and diagnose it. I think all this was to make it like an appliance as much as possible, headless just plug it in and connect to your phone system.
Yep, 386 motherboards in custom-purpose hardware. Kind of like the CD duplicator we had at Future Vision. It was in a very sparse-looking beige metal housing, with nothing more than a row of membrane buttons, and a 2-line LED display. It could run 4 SCSI 4x CD writers at once (each on their own card/cable). Some time later we had to open the system, and found that 386 (or 386SX) motherbard, 5 SCSI-1 cards (one each for the external writers, one for the SCSI HDD you would dump your master copy down to, then read off of to write to the drives). You'd program your choices of what you wanted to copy, how many copies, etc through a labyrinthine menu on that 2-line display.
12:19 That 'thing' and those bodge wires are almost certainly the issue. I've never seen anything like that on standard PC's of the era. I doubt that the BIOS, or keyboard controllers have been modified beyond something about that bodge wire and assembly intervening to probably prevent CMOS changes, as you are able to use the keyboard in DOS. Also, one of those bodge wires runs to a group of 5 pins, which might be related to keyboard or unpopulated PS/2 interfaces that a lot of these motherboards had provisions for, but weren't used. You might want to start with those bodge wires and see what happens if you disconnect them. Or, try to find a schematic for the motherboard, which looks super generic and is probably an unlicensed copy of somebody elses board, which was also super common for boards of this era.
thankfully someone else commented who used to work with these, the bodge wires are for converting the 286 into a 386SX, before later board revisions incorporated the needed changes natively.
I’ve seen other comments here saying what I’m gonna say...drop a 9-pin serial/console cable into COM 1. Spin up a terminal and I’m pretty sure you’ll get output. Try various rates but likely as not 9600 will be the one.
A few observations: - Both Brooktrout cards use a DSP: On the later revision it's the small TMS320BC52PZ on the left side ("U2"). Always suspect a DSP when you see "TMS320" ;-) Some boards are prepared to use different versions (package sizes) of the same chip and this seems to be one of them. - You should check out these EEPROM-like chips on both cards if they are actually EPROMs. Often cheaper, non-reusable EPROMs are used (you can't erase them as they don't have a window like the large EPROM DIP packages). Better remove the sticker and check the IC number. If it's begins with "27C" you have an EPROM (I sometimes had luck with programming them ONCE, though). EEPROMs often have "28C" which work with your MINIPRO (a 29C would indicate a FLASH memory "PEROM" - also good). - You point out the date on the I/O controller chip on the I/O card. Right next to it, on the card itself is also a datecode: 3895 ;-) - You also missed the datecode on the DALLAS clock module: 9506. - Also the 386SX is not "essentially a 286". It's to the 386DX what the 8088 is to the 8086: It can run the same software, but externally supports older hardware (and less memory) - and it was much cheaper. The 286 can actually be faster when clocked the same but it was obsolete the moment 32-bit software became the standard. The 386SX can run these, though (I believe Win 3.11 and later). - As for the "oscillator module": You should look up the number on the TO-92 package. It wonder if it's really a transistor. Never seen such a module. Maybe desolder it and draw a schematic - as tiny as it is? - Also: The long botch wire on the back runs from this module to resistor R16. This seems to go to unused turbo switch connector. You may want to check that out.
The "oscillator module" at 10:45 might be a primitive watchdog. Of course, the CPU needs some GPIO pin to periodically reset the timer (after it is enabled). Since since this box only has 1MB of RAM, I'd repurpose the A20 gate for that (which is famously controlled by the keyboard controller). A PAL (PLA, GAL, CPLD, ULP, ULA, ...) with feedback logic could be used to enable the watchdog when the A20 gate gets asserted for the first time after power up, and from there on, it needs to be periodically reset. Would make sense to an unattended voicemail system to have a watchdog which automatically resets the system if the system hangs/crashes/behaves irregularly. Sorry for mentioning the A20 gate. A truly ugly hack...and Intel CPUs supported it until Q2 2013. Well, officially, Intel just mentions that you should not expect the A20 to work as expected on 64 bit Haswell (and later) architecture CPUs. That makes me wonder...just how does an Intel Core i9-10900X access more than 1MB of memory under DOS 6.22?
Lucent name brings back bad memories of fighting with Winmodems and "added value" software that came with them in the mid to late 90's. Before ditching dialup completely I resorted to using an external hardware modem.
Here's a suggestion on how to get this working: Make a daughter card that plugs into the PAL socket and implements the stock IBM logic (7400 and 7407) to interface to the keyboard, then replace the keyboard controller and the main BIOS with standard parts. This should bypass the PAL tomfoolery and get it to behave like a stock AT board.
Funny you're doing something on this. I run a 30 year old multi conference/chat line that uses old dialogic d/40b cards. Its a intel P2 400 mhz running DOS 5. The cards are from 1988 or there abouts. It has a dialogic bridge board D/80. The program is written MS C+ with some asembler code.
I have an Alcatel-Lucent OmniPCX and it is one of the most custom and locked down things I have ever seen. I was just barely able to get a console working but it does follow the same principle as this device: A slightly custom computer (CPU) board/module and then multiple add-ons for whatever your needs were.
The assembly at 10:56 is an oscillator. It has "OSC2" beside it in the silkscreen. At 14:23 you were wondering about the battery in the clock chip. The box said it is a Refurbished unit. The 07/00 date may be when it was repaired.
I feel like I discovered this video a bit late, but from my memories of obtaining a Merlin Mail system many years ago, and 'going down the same rabbit hole', the boards were configured for a specific set of hardware in certain slots. I managed to use Linux to do a 'dd' image of the hard drive when I had the system, and I want to say it, or at least my system, ran some version of AT&T Unix or BSD Unix. Having maintained the 'big brother' of this system, 'AT&T AUDIX', it wouldn't at all surprise me if the board was, in some way, custom for AT&T/Lucent... I do remember that, on my system, the port labeled COM 1 had the same messages as your VGA port. It now makes me think that the bios, along with the custom PAL chip are modified by the manufacturer to enable use of a serial console for management and maintenance....
Good old Brooktrout cards. I've seen these used in fax servers, though I don't know if the ones you have are the same or not. That BIOS stuff can be a bit of a fiddle. BIOS manufacturers do include a number of "hidden" settings that can be set up using a special utility or something. Unless you happen to get lucky with finding the right one somewhere you probably aren't going to be able to figure out what needs to be changed to re-enable what's been disabled.
Brookcom were heavily into telcom cards in the 90's. My former company was using them until we switched to Voip that was in 2015. The Power supply likely has both a -5 and -12 one of them was removed so modern power systems don't have both. We had a hard time finding power susplies because of that.
I ran an AT&T system 75 for a few years (the Merlin is a small version IIRC). They NEVER crashed and we had DOS based 386 voicemail system for it until 2005. We eventually upgraded when ebay was getting expensive for parts.
@@JesusisJesus That was a long time ago. I am looking for an old Merlin or System 75 though. I am building a retro BBS cluster and thought it would be neat to have a system like back in the day.
Adrian. Brooktrout cards were used a lot for fax services. Particularly a program called Zetafax. You can still get a virtual brooktrout device on relatively recent versions of the software for doing FOIP.
yes those are DRAM spots. 386 motherboards often had a choice where you could have both 30 pin Simms DRAMs ( usually socketed ) because early on there was a huge price difference between the two memory types. ( 8x128kbit +1parity... or maybe x256)
Those are set up to do a serial console. It's 1200 baud, 7 bit. (It was designed to have a modem hooked up, so AT&T could dial in to make changes) The password to get in is probably: p1zza They run Concurrent CP/M!
Years ago, we had a voice mail system at work that ran on a DOS box similar to this, complete with the DSP cards and the cheap IDE hard drive (a Western Digital in our case). The system it went with was a Panasonic; we ditched it all and put the Panasonic up for sale after switching to IP phones and Asterisk.
The board with the resistor diode and capacitor is a Bios lockout reset, Its design to lockout the bios, reset it to default pre programmed configuration after reboot, even if you manage to get in, and here's the kicker if you remove it doesn't boot at all.
Looking at that small board, it looks like it would heve been the spot for the 14MHz oscillator, and they instead removed it, to put in another clock signal, probably a divided down 66MHz clock, so likely either 16.5MHz or 8.25MHz, which is probably used by one of the add on voice cards as a clock synchronous with the processor, as pretty much no other cards would use the 14MHz clock anyway, unless it was an early CGA card generating NTSC video.
At 12:20 the windows logo was because there had been some glitch with Windows and some non Intel CPUs so I guess Bill seized the opportunity to make some logos and sell them! This logo thing went on for a while and was evetually dropped or maybe not because it is kind of the thing you could still see in some laptops today that the sticker is displayed to the owner instead of being buried in the CPU where nobody could see it anyway because of the fan and stuff, starting with 486 DX2-66 CPUs started needing fans officially recommended by Intel. Great video thank you!
i´d put my money on the often mentioned serial console. from a service perspective that makes a lot of sense. technican comes with his laptops hooks it up to the serial port and does his thing. no need to add a vga card or keyboard.
Super Nifty! Those old boards (when working!) are great as "floppy interface" machines for working with older 8-bit machines for floppy disk transfers, especially on computers that don't have flash systems, like old Kaypros, etc. I have a very similar board installed in my "386 Frankenstein" machine - sitting on the bench right now for work with a Commodore 64!
Yeah, I have a IBM PS/1 that i swapped out the 1.2 drive for a 360k and is primarily used to make floppies for my trs-80/kaypro/osborne. Technically it's a nice dos gaming machine if i had any interest in moving past 8 bit machines :)
@@brianv2871 Are there any computer boards for sales and trade other than ebay I would prefer to seek out my stock of 386s to be collected from me I have SIMMs as well, no means of testing them though
Avaya, previously Lucent and formerly AT&T used hacked up PC architecture for their Merlin Mail system. Merlin Mail was a standard VMX [VoiceMail Exchange] used for mid-market PBX [Private Branch Exchange] systems. It was pretty typical for voicemail systems of that era to use PC based architecture. Third-party voicemail systems, such as ActiveVoice Voicemail, often used off-the-shelf PCs. Even slightly later, on-card voicemail systems were still PC software running on specialized PC cards designed to fit inside of PBX cabinets. This was especially true after NEC acquired most of ActiveVoice and put their voicemail software on custom PC based cards (complete with DOS formatted flash drives) designed to be installed inside their NEAX and Electra Elite PBX lines. As you found, these outboard voicemail systems integrated with PBXes using Analog Voice Cards that were connected to pooled analog extension ports of a PBX. When you called the VMX pool extension, your call was routed to a free extension number in the VMX pool that connected to one of the ports on those Analog Voice cards. Dialogic Corp made the cards. Almost all PC architecture based voicemail systems of that era used Dialogic Cards. Most of these voicemail systems used out of band signaling to and from the PBX over a Serial port. This means when you called the VMX pool number from another extension, the PBX would send information about the call simultaneously to the VMX over RS232 using the Serial port indicating the originating extension. The RS232 connection was also used for MWI [Message Waiting Indicator] functionality. The reason you were having so many issues getting that motherboard to act normal is that they were hacked up to run without a keyboard or pointer device. They where intended to bypass the keyboard check, and boot to a drive running a basic OS (usually DOS) and the proprietary Voicemail software. Technicians would program the voicemail using only the serial interface. Technicians would never connect a keyboard or monitor to the system, and most telecom techs of that decade might not know how to service a PC. If there were an issue with the software, a technician would order a whole new drive or voicemail system with the software preinstalled.
FYI, voicemail software can be very sensitive to card in motherboard. i use to build one that was 486dx-100 board which had only one pci slot but 8 isa slots> when the video card got changed to a different model the software crashed. Only the one brand card could be used. The used i believe rocketport voicemail cards
That DALLAS chip not only has the realtime controller but also some bytes of memory and an include non-rechargeable battery - which drains slowly when the computer is powered off (and over time). On SGI Indy workstations that chip also contained the Ethernet MAC and the serial number. If the battery was toast you needed to buy another one of those chips and use a hex editor to reprogram it byte by byte. I did that twice back in time. Nowadays there is no replacement anymore for those.
I've looked over the random Brooktrout stuff of the time. The closest i can find to the telephony boards you have are the TR112 cards. It's possible for such a high volume customer, they made something custom. I've had the opportunity to develop for a lot of similar telephony boards - lots of fun, but the Brooktrout stuff has always kept their SDK a deep dark secret. If you take a look at the files on the hard drive, you might find a clue as to what these are; usually how these things work is there'll be an executable included along with the drivers that pukes a boot image into the processor onto the ISA board - sorta like a baby bird xD . The configuration file for that will more than likely contain some information as to the model. For whatever it's worth, there are still some manufacturers making systems just like these (with more recent cards/motherboards, of course) for some of the more modern, less "shove an ethernet port into an ATX machine and it's a PBX" phone systems.
Heh, okay, it's actually a TR202 card, as per the sticker. There's a development manual on one of those shady document hosting sites: manualzz.com/doc/33471336/brooktrout-fax-and-voice-application-programming-interface At the time I tried it, the document wouldn't load, so here's a Google Cache of it - webcache.googleusercontent.com/search?q=cache:0gIJ_ZnaF0QJ:manualzz.com/doc/33471336/brooktrout-fax-and-voice-application-programming-interface+&cd=5&hl=en&ct=clnk&gl=us
@Adrian's Digital Basement - I always use F2 & DEL quickly pressed over & over, because these 2 are the most used keys on PC from 2000s (when someone gives me a computer to look in the BIOS, because of problems) ... ... But nice to know that you can read this stuff out.
At a guess the cap and transistor in the other crystal socket is a watchdog timer with a several second timeout. It probably pulls on NMI and resets the machine if it isn't retriggered before it times out. Possibly the two rework wires on the back of the board are part of this (presumed) watchdog circuitry.
I'm seeing a bunch of good ideas posted already. The only trick I would add, is putting it back the way you found it plus the video card and see if it's giving you additional clues as it's booting up the way to manufacture intended it
My brother worked at Dollar General for a long time. They used DOS on their point-of-sale machines, until eventually management decided to upgrade the OS, but not the hardware. They opted for Linux, which of course ran like a slug doing anything on those old machines in comparison to DOS. The registers took five minutes to boot up from then on, so any time there were power issues, from a storm, or whatever the case may be, it became a nightmare scenario of long lines and frustrated customers while rebooting. The logical approach would have been an intermediary machine to connect the old registers into any newer back-end hardware, but they were never much for common sense, and certainly weren't going to spend money if they could avoid it. I think K-Mart took a similar approach, especially when I recall their systems taking a long time to start up when they went through a phase of electrical issues. Though we can all see how K-Mart ended up.
It's likely to present a servicing UI of some sort, that may be more interesting to see. I've dealt, in the past, with a sizeable fire detection system for multi-building facilities that was a 386 chip with custom mobo and software. It was a reasonable choice for the era, as it was for a voicemail system. I'd really wish some more delving into the software side of the thing, i.e. filesystem, software and servicing UI, rather than just trying to have a somewhat-working "yet another" 386 motherboard...
@@djangowatson217 I think, as many have pointed out, that this machine was configured to run over a serial console. Also, I think Adrian forgot to mention it but it seems like the hard drive controller was failing or failed to boot the disk. Either way, it seems the goal was to poke around a bit but ultimately using it as a standard 386 board.
Adrian - You were on the right track. The only thing missing was a test of the Battery. If the Dallas chip is dead, you cant "Save" a config. It will default to the wrong drive time after time. These motherboards do not support "Soft Boot" any reset command is a real reset and Bios is started over. (Packard Bells that had 486 were this way as well, they only take power from the battery and not the power rail. Man I wished I was in your basement, this would have been a fun win! o and btw... Not likely he will use this card.... OOOOO NOOOOOOOOOO I will pay for them as will many others! that love BBS systems
Brooktrout was a company from Needham, MA, specialising in telecom cards. The brand still exists, but I think only for Fax boards - have a look for Dialogic. The boards you have probably cost a good load of cash when new, but with an ISA bus will be of no use to anyone these days.
Those were a pretty penny back in the day. Brings back memories of those smaller Merlins, midsize Partners, and bigger Definity systems that I cut my teeth on as a young engineer.
Damn FINE, many thanks from the "ex radio, then ex fibre engineer", in Australia's dominant Telco. Retired at age 53 in Thailand, SO MUCH more enjoyable. Keep up your TERRIFIC work, great viewing! Mark
Late to the party. I met someone in the 90s who worked on systems like this. With ISDN trunk lines you would have industrial 286/386 motherboards hosting a whole bunch of ISA cards. The cards had freaking 486s on them to handle isdn and voice logic. Blew my mind.
The cards do have an order the first card can be a stand alone mm02 the second card can be ports 2&4 or ports 5&6. It will only work with a merlin system and normaly connects to the last of the station ports
Hi Adrian, I just found your channel great stuff, subed. I used to work on Merlin systems back in the early 2000's, when I was a tech at a local telecom business. It's been years, but I think I have a ton of service manuals for these systems in storage somewhere. Next time I'm in there, I'll look for them. I didn't work on these long,. VOIP was on the horizon, and we jumped on the Mitel bandwagon and ditched legacy AT&T products. We had one tech that still supported these for years though, for clients that were hesitant to mirgrate over to VOIP.
Im pretty sure there was a debug program in dos that you could write some value to a memory location which would cause the CMOS validation test to fail. You would then be prompted to access the BIOS by pressing del or F1/F2.
I found it Solution 1: for AMI and Award BIOSs: C:\DEBUG (at a clear DOS prompt, don't do this in DOS-box in Windows) -O 70 17 -O 71 17 Q Solution 2: for Phoenix BIOSs: C:\DEBUG (at a clear DOS prompt, don't do this in DOS-box in Windows) -O 70 FF -O 71 17 Q
Didn't work. I just swapped the Dallas RTC with a with another one with a dead battery to promote that error but no luck. Nothing changes when CMOS is corrupt and it all just goes back to the default settings I show, no matter what.
@@adriansdigitalbasement Likely using the wrong setup.com. There were about eight different setup files to use, depended on the BIOS, and between PC, XT, 286, 386. Was wonderful when the setups started going onto the CMOS. Too bad, I just pitched a few manuals and PC books from that era, too. I'll see if there's stuff left in the attic (judging by the lack of room left, yes, I'm sure there is....) You might try holding down a key at boot (just after the Trident pops) and force a keyboard error. Sometimes that displayed the key combo for CMOS setup. If not, then I'm pretty sure it's a boot disk and the right setup that's needed.
Just a guess. Tons of those type of systems would be hard wired or programmed not to ask for keyboard on startup. So possible why you keep getting a keyboard beeping sound on startup. They also did the same thing sometimes for missing vga cards. To counter act the error for missing video card. I had a partner mail vs system that was not working one day. I pulled the 2.5" ide hard drive out and found I think some kind of unix release on it. Anyway I was able to clone it using ghost over to another drive without a issue. Going around the 1 bad sector and problem fixed. As for finding newer date on the machine being built due to cards inside. I talked to a few companies who due have special embedded stuff back in the day. They would order bulk to get a better price. Since they mostly write the software to work best with that hardware or in case they need the special modifications. They might have 1,000 motherboards in stock modified the way they want it and thus built the same system for 10 years. But I'm sure you already know this. As for the comment below asking about field service of the device. During that time you would not trouble shoot the bios or plug in stuff. Once you found a dead board. You would order another one from the mfg. Nice vendor lock in.
For me it's pretty obvious what to do. I think it is beeping every 108 minutes and then you have to enter a special sequence: 4 8 15 16 23 42 Love your channel btw :) Love this hardware hacking episodes. They are amazing!
Back in the 90's I converted an AT&T terminal computer to a dos/windows system. With the model I had, you needed an ATT boot disk to access the bios program. might be the same for this system. I was lucky enough to find one on an ftp back in the day.
I think a better description of the 386sx is that it's a 386 with a reduced interface design, similar to a Motorola 68k -- it has a 24bit/16bit external bus, but executes 32bit instructions.
The system was probably built in 1996 or so. and when the hard disk failed in 2000 they sent it back to Lucent and replaced the hard disk but kept all the other hardware the way it was. The Dallas chip has 9506 on it so it's probably original. Have you tried connecting something to the COM1 or COM2 ports? Maybe the modified BIOS does something with them. After all, there must have been some way to connect to this thing and do some configuration that couldn't be done with just a phone.
A few jobs ago, we had an NEC PBX that had an old Dell Optiplex tower running OS/2 handling both voicemail duty and as a programming terminal for the PBX itself.
Watching your attempts to get to the BISO setup. I noted; One is you waited to do the key presses too long. With these old PC systems it was always better to start trying the keypresses before the post messages had come up. As for the beeping. That was saying the keyboard buffer was full. Tip, try one or two key combo presses and if you don't get in before the end of post or you get keyboard buffer full beep as you did then restart and press another again. The keys to try are "ESC", F1, F2 or Del. Having to use the control key sometimes was needed but not as often as turning on and hitting the above key.
The bios memory chip battery back would still be working as being always on so did not much drain only on power cuts. I have used ones from systems like this and they still keep going. Having worked on these system we mainly used the RS232 to fault find and program them. So boot and connect up to RS232. I do remember something about the keyboard and the 286-386 boards so will have a dog in my loft for old manual. Similar problems with epos boards to enter bios you had to hold down non standard keys. I want to say it was something to do with memory over 640 on 286 but it was a long time ago.
that 'home made' oscillator is sick! Wonder if it's to under or overclock the thing or just because they didn't have the actual part when they refurbished it
Well, to be honest, I started watching thinking this doesn't really interest me. Boy, was I wrong? I was rooting for you the whole time. Very crafty using that alt PhoenixBIOS on a 720kb floppy -- nice one. I'm looking forward to seeing how this one might pan out in the end. Great work!
It has kind of been a dead-end this silly thing. Was hoping someone with better skills than I have could unlock the stupid BIOS to allow this to not be a total POS. :-)
I've worked on several of those Bell Labs > Lucent > Alcatel-Lucent > Nokia systems. The BIOS has been modified to treat all text modes as a serial console like seen on non-intel UNIX systems for years. This is similar to the management ports on newer rackmount Intel server systems where no local keyboard/display is expected. Pretty common stuff today, in the mid 90s it was totally non-standard. I'd imagine you could trace from the keyboard controller over to ports on the ISA bus for part of that redirect. That machine would have run a version of AT&T UNIX when it was in service. You might have success taking an image of the hard disk with dd and trying to mount it under linux as a loopback filesystem. As far as how old the hardware was for the installed date, Lucent buys a huge amount of hardware for their supported platforms so they could sell and maintain the hardware for decades, as people don't replace their voicemail system or PBX system very often. Incidentally, Lucent would charge several thousand for those machines more than the underlying hardware would warrent. Also, its pretty common for them to do component level troubleshooting to this day. Those bodge wires indicate the box has likely been repaired and redeployed at least once. The line cards are more likely to be DS0 boards than analog/POTS lines. The software had logic to transfer someone to another extension or forward to an external number that can only be accomplished with the DS0 signaling (POTS lines only really know on-hook and off-hook states).
Adrian - I built many desktop/transportable systems back then (including this board) and also managed this very email system. That board is what we called a 286/386 motherboard. It could come with DRAM sockets and/or SIMM sockets, and a choice of 286 or 386SX CPU. The board would barely handle 33MHz, and failed with a higher FREQ. the 386SX-40 is likely because of availability and later, pricing (40MHz was cheaper than the 33MHz part). The kludge board (early models) near the CPU was for use with a 286 CPU, and is bypassed when using a 386SX by that circuit. If it was a 286CPU, the other socket was empty or similarly bypassed. The PAL/keyboard and BIOS issue is due to differences between how the keyboards connect on 286 and 386. Many failures were "fixed" by replacing that PAL, and not the keyboard BIOS. And yes, changing from Phoenix to AMI would often (in those days) require swapping both, and all three on this 286/386 contortion.
Ah so that means there is little hope for this board as it sounds somewhat bespoke and customized for AT&T/Lucent. Clearly I won't find a better PAL and without a working BIOS there isn't much to be done.
@@adriansdigitalbasement AFAIK it is sometimes possible to read OTP PAL chips, and it may be possible to work around with some effort... but I wouldn't keep my hopes up, probably not worth the effort.
Though as someone else mentioned, might worth trying again and saving with ESC instead of F10...
Can’t you just bypass the PAL and wire the KB micro directly to the port? (With the necessary buffer circuit if need be.)
I had a look at the BIOS and it is custom, where the hard drive information is normally stored the BIOS is looking for what appears to be a date and time. The Keyboard protocol seems standard nothing out of the ordinary.
One of my previous employers had a system like this. According to IT is was a 386 running a custom version of Linux.
Networking gear is another great source of surprise PC re-usable equipment. An example was about 17 years ago while cleaning up old equipment I noticed an old router oddly had a VGA port, serial ports, and PS2 connector. I cracked it open and found a tiny motherboard sporting 6 network ports, an IDE connector + 2.5 HDD, one DDR slot, and an old Cyrix VIA chip. I connected up a dual connector IDE cable and a CD-ROM and the thing booted off of an XP CD. I had no problem installing XP on the little sucker. XP crawled but it ran. I put in the biggest DDR memory I could find - 512MB - and wound up transferring our old keycard system onto it from an old dying dust packed tower that was sitting under a desk in the basement for the last 5-6 years. I screwed that little system to the wall and it ran like that for another 10+ years before finally a complete security system upgrade.
I ran into a similar situation with a small form factor pc used as a controller for a remote access KVM device in a server room. It was an AMD K-266 with 128mb RAM. I acquired a scrap pass to bring it home and ran it as a firewall/ router running Coyote Linux booted off a floppy.
There used to be PC Magazines that did how to build routers out of old 286/386 units
I’d put it back to the stock config and connect a terminal or emulator to a serial port. At a previous job, We had an AT&T audix voice mail system that was similar, and it ran AT&T Unix and expected a serial console not a vga with keyboard. At my current job we’ve got some modern server running Linux that are configured the same way, the primary console, including post and bios setup, are directed to a com port.
Beat me to it. :) I'd imagine that when an official repairman worked on it, he had some kind of handheld serial monitor that he could plug in to talk to the hardware.
What he said... most of the ATT & Lucent are expecting Unix and not screens. Hook up a terminal via serial with the hard drive attached. As I remember it, it was 9600,N,1 but I probably have that wrong.
That's exactly how it worked. Those two cards plugged into the phone system controller and would have been treated as extensions in the system. From there, the users would access the system through their phones, meanwhile the operator would have a serial console at their desk where they could manage the system as a whole.
downloads.avaya.com/css/P8/documents/003677613
Yep telecoms stuff is often RS-232 and a further hint is the lack of a video card. A field tech wouldn't likely be dragging around a video card and monitor. Good luck.
The lack of video card , and odd looking bios gives it away as com1 console
Is it bad that I see things made in 2000 and think "oh, that's quite new" before realising it's 20 years ago?
I know what you mean
......shit
you're right, i feel old
Yeah, I had the same thought, "Oh that's gotta be really old" then he shows that it was made in 2k, "oh thats only a few years ago...oh wait"
Tom Barber yes we’re getting old.....😱
Same here. Every. Single. Time.
I had to administer a similar voicemail box back in the early 2000s. It was running aDOS based system and was not friendly to configure. I'm so glad those days are gone.
executone?
Honestly I have no idea. That was at least 15 years ago.
Just after the Win95 systems came in easier to admin and use
dude, that thing is worth a fortune! us telephone collectors that still use Merlin Classic systems pay good money for these since they are pretty rare! You should offer it up on eBay or something so it can go back into service!
These systems were introduced in the mid 90's. The build sticker is likely the date it was rebuilt by Avaya, which im suprised they didn't rebrand it since Lucent was gone by 2000 or so unless this was done right before they spun off into Avaya from lucent. These things were rock solid, and likely the reason it was refurbished is because the hard drive failed. These cost about $10k when they were new. This would wedge into the merlin system using extensions. The system would be programmed to forward a call to a VM port (one of those 2 cards) where the VM system would take over. The end user would call an extension or dial a code (depending on Classic or Merlin Plus system) and the system would prompt for the VM box and passcode. It was a highly advanced system for its day. My Avaya IPOffice 500 from 2008 or so (which is still sold today) does everything this huge box does on a 16gb CF card!
The last analog phone system I supported was a System 75. It was bought in the early 90s and lasted until 2005. I miss the Charlie's Angles speaker phone.
I would love to find an mid to late 90s PBX to route calls to a multi node retro BBS.
Hell yeah, letting a telephone collector get use of it instead of simply cannibalising the motherboard sounds nice.
I would hate to see that salvaged for parts; I like the idea of it going back into service.
its worth close to 0, nowadays you can setup software PBX on a $100 worth of hardware, and it will include voip endpoints and all the modern features.
We had one, called Smooth Operator, when I was working in the early 90s; they had specifically configured it so we couldn't use it as any kind of standard PC. I seem to remember it coming up in the conversation with the sales people at the time.
So it is implemented as security by obscurity specifically to stop people trying to use it as a normal PC.
A lot of people are mentioning the fact that the "serial port/console" is your way in! I agree. Probably the reason for this weird keyboard setup + the keyboard connector was not accessible from the outside. I wouldn't be surprised that they have created some weird serial keyboard (or special console device) for their technicians
The systems i work on got programmed from the phone system itself by command down the phone line interface
A program on a Toshiba Laptop should do it. (some laptops did not carry the correct pin out or line voltage to work the programming card on phone systems, same could be on this).
@@highpath4776 could be also a special serial port, I work on several devices whose serial port required a special cable, but I would say knowing most probably programed though phone system
Adrian - in the setup.com program you were supposed to hit Escape, F10 was just to exit the program. Escape was reboot, which I assume would apply the settings.
Oh that would make sense. Hope he revisits this.
@@NLDXBOX me as well. As another commenter mentioned, I would also like to see what happens when it boots off of the hard drive that was in there. If it even boots at all from it.
Check out the help menu as well for other commands. Looks like it never saved or it's wired to reset it self. Try to change the controller board placement, could be an IRQ conflict thing. Or change the controller and other boards. Try to unpatch the cable under it.
That's what I was thinking, ie, F10 exit without save.
Yeah, and if that fails, I remember that there was also a utility for these PhoenixBIOS chips that could lock the configuration. It would have been a separate utility that would 'toggle' the lock at each run. Back in the 80386 era, I had a computer with the exact same BIOS as this one, only it was accessible through F2, as he showed. You could also use the supplied disk to access setup. The disk also contained this other utility to protect access to setup. You could even set a password for it, but that wasn't required. If protection was enabled, the 'Press F2 for setup' message disappeared at boot. Therefore, it wouldn't save the config if you tried to edit it using setup.com.
We had one of these at the company I work for until about 2012, when we went to IP phones and got rid of our Avaya PBX. Which BTW, was so reliable we never had a single problem with it to my knowledge in the 10 years it was in operation. The "Audix" voicemail PC was identical except it did have a front panel with a floppy drive and an optical minidisk for backup. Once I had to take it apart and reseat everything and as far as I can remember, the interior was identical. We managed the system with either an amber screen terminal or a piece of software that replicated all the features and connected to the PBX's IP. *Edit: Reading the other comments on its value, I really wish I'd rescued it from the recycling once it'd been retired.
Hmm. The original BIOS that replaced the IBM PC BIOS code was Phoenix, it was a "cleanroom" rewrite and not a copy. Back in those days, things were a bit different. The configuration program being built in to the EPROM was not there in the early days. If it has it you can hold down a key during startup and you will get a keyboard error and the option to press F2 to enter setup. The parameters were maintained by battery backed CMOS, not EEPROM or Flash so no battery and you lose configuration. I would guess that little circuit was to maintain the CMOS memory? Those ISA interface cards would have been a small fortune back when that was new.
Terrible how I have forgotten most of what I knew. And it took be ages to learn. And the manufacturers changed the systems and next sales point in so had to learn the new tricks.
Brook Trout was the company that made chips that say "Bt" on them. They were used a lot in TV Tuners and I think DAC chips for video cards. They were purchased or merged with Conexent which was originally part of Rockwell, who also made Modem and Telecom chipsets.
I used to be a CallWare Reseller/Installer. It was a Voicemail system that ran on top of Novell Netware and we used a lot of the Brooktrout boards. Those were very common in-band signaling board (that means the phone system sent a hook flash then the DTMF code for what type of call (ie. Voice Mail Retrieve, No Answer Message, Caller Busy Message, etc) and then it sent the extension number that it was coming from and then another hook flash, and transferred the call. There were all kinds of systems and each brand had is own codes and methods. Some were well documented others we had to use a box called a digigrapper to monitor the line and see what codes it was sending and do a bunch of test calls. There were also systems that used out of band signaling as well like SMDI ( a seperate cable) or TSAPI (A network protocol that signaled across IPX/SPX). You could run the non DSP cards for up to 4 ports but (each one of those RJ11's is wired for two lines), after that you had to get the DSP version to off load the voice processing from the CPU. As we started going to faster processors that became a non-issue. Then VOIP started to take over the world and it all changed.
This thing was running DOS 4.0 and had a single executable and autoexec on it. I remember seeing one of these in use a long time ago and there were nothing but RJ11s between it and the PBX, so I'm assuming the DTMF signalling you mention was used. Another comment said this was available as a 286 as well which was very slow and struggled. Hmm
@@adriansdigitalbasement I am cleaning out my storage unit that I have been backing full since 1994. I am pulling a lot of the boards ram, keeping a few cases and stuff. If there is anything in particular you're looking for let me know. The vast majority of it is Pentium 1 and up. I scrapped the XT/AT stuff years ago, which I am kicking myself over now.
14:36 9506 is the date code on the Dallas RTC, so it's original. Also, the case says "refurbished". It was an old machine from 1996, they replaced the hard drive and the power supply in 2000, then sold it as refurbished. It would have been wise to replace that RTC chip as well...
Yes. 25 Years is a normal lifetime for these RTCs. I have one from 94, it is still working.
wise yes
needed no
25 years later and its still going strong
@@frogz I'm not sure if it's still running. The BIOS utility said that the RTC date is 1989. It might be caused by the utility was unable to read the SRAM/RTC data.
But if you ask few hundred dollars for a refurbish job, yes, replacing a 5 year old RTC battery is a minimum. Okay, the expected running time is around 10 years according to the datasheet (they don't say it's guaranteed!), but still. A working RTC in such equipment is quite important.
My guess is that it was built in the mid 90's and the HDD and battery were replaced in 2000 with a refurbishment as well as new software.
That is what I was thinking, a factory refurb replacement was swapped instead of repair to keep phone systems up.
I was building something like that in late 90s into early 2000, used a 486 motherboard, ran dos 6.22
Considering it had a sticker mentioning refurbishment I'd say that'd be the obvious answer that its baffling the guy didn't figure out on his own.
Those little isolation transformers are very useful to make your own passive ‘DI Box’, useful when your computer’s sound card is picking up any ground loop humming. Your video has very excellent audio, but you never know when a ground isn’t appropriately connected to the local wiring and you start picking up ground loop noise. Reminds me of my first PC, a Pentium 3, 300 MHz. It’s a worthwhile educational toy if you stick with it and poke around and learn to be resourceful. The 8086 book became a favorite of mine. THANKS FOR SHARING!!
It's very likely that the motherboard is setup to run with a serial console... I'd try plugging in a terminal or serial cable into another machine with a terminal emulator... also if I remember correctly the older at&t voicemail systems from the early 90's were unix based and used a serial terminal vs. a vga monitor and keyboard. Looking at the data on the hard drive might give you some clues as well.
You could use the old Unix voicemail or buy a much cheaper DOS based system. Phone equipment is/was freaking expensive.
This. I came here to say this. Try connecting via serial
@@mstandish because you don't know a damn thing about telephony.
@@Steveos312 why would you think that?
@@mstandish because you're using dos to run your systems for cheaper haha /\
thats the dif between a 10 user system and a 1000 user system
Hi Adrian !! This Dallas clock chip have a battery embedded !! Just open this IC, remove the battery and replace it with 2 alcalines AA socket/support !!
I did it very very times in the past !!
Brooktrout is one of the few companies that make Analog voice boards for VoiceMail and Fax. Each of the two boards has two Analog ports, as you pointed out. These do not have dialtone but will answer a ringing line, and are just an audio circuit (usually) and can send and receive DTMF, etc. The VM communicated with the phone system over one or more serial ports (RS-232), those handled the signalling of what port to answer,a nd with what, light Message Waiting lights, etc. Common serial settings were 9600 8N1, try hooking a serial console to them! It looks like the original build was 1994-1996, and was factory refurbished in 2000 as indicated by the sticker on the box. I never worked on the Merlin Mail, but AT&T commonly used System V Release 3, 4 or 5 of AT&T Unix, or MS or other DOS as the OS. All of your bios control is probably done through Serial, or hard coded at the factory to ignore the keyboard. You can drop the HDD, and Brooktrouts in any 386 or 486 motherboard and get it running probably, did that on quite a few VMs in an emergency.
Hi, fantastic video so far. I just paused to reminisce. Anybody who was buying PCs in the late '90s will remember the fact that they all came with modems installed, as telephony was being exploited to the max in those days, particularly in the consumer space. And the fax a voicemail systems in those home PCs was starting to rival those of expensive office setups like this, with individual mailboxes for each member of the family, etc. OK, old man wistful thinking ended lol
14:35 the bottom number is the datecode, the dallas chip is from '95 as well, 9506 sixth week of 1995
If these chips can switch between internal battery and getting power from the PC, being in a machine that was turned on 24/7 for decades, there's a good chance there's a few years worth of juice still in there!
was just looking at the comments for this!!
either that or its a REALLY new voicemail system post julian calander
made in the 95th week of 06
I've found several reasonably nice PCs out on the curb waiting for trash pickup. Usually, all I have to do is press the key for the restore utility while it's booting. People mess up the OS and the software and throw it away, but there's nothing wrong with the hardware and they must not realize that there is a built in factory-condition restore. Every now and then, I can tell they were poking around inside especially if the BIOS reset jumper is in the wrong position or some cables are unplugged. Hardly anybody bothers to remove or erase the hard disks.
Which restore utili, one on mine wont work, its going to linux
You were right when you said the cards had analogue trunk lines, that's exactly what those are. They're also known as FXO, where the O stands for office, as in central office, because they act like a telco's central office (AKA telephone exchange), and act like an analogue phone line. The DSPs are analogue to digital and digital to analogue converters (ADCs and DACs), digitising the voice of people leaving messages and playing back recorded messages.
There would also be a CODEC chip for encoding/decoding the digitised audio to/from a more compressed format to save space on the hard drive. The chips at the rear of the cards look like SRAM, which would be used to buffer the digitised samples for playback. Basically the CODEC chip takes the audio file and decodes it to the raw digital audio samples and stores them into SRAM, which the DSP would read and convert to an analogue signal.
Modern systems don't need hardware CODEC chips, it's all done in software these days, but old systems like this wouldn't have been powerful enough for the CPU to encode and decode the audio in real-time. This only became possible in the Pentium MMX era when Intel introduced the SIMD instructions which allowed CPUs to encode and decode audio faster than real-time. That's why MP3s didn't take off until the mid-late 90's, because before then PCs weren't powerful enough to decode the audio for real-time playback (without hardware assistance anyway).
Brooktrout! Man what a blast from the past. They used to make cards for faxes and voicemail systems. Way back in the 90s I had to maintain a bunch of OS/2 based voicemail systems, and they all used Brooktrout cards. Most of those systems were tied into Mitel PBXs, and had an external interface box to do the trunk line signalling. But the brooktrouts did the actual voice line processing. Pretty cool cards, the ones I dealt with did on board DTMF recognition, and could play back a decent set of PCM sample rates. We would use the Voicemail systems to do all our IVR work as well. I wonder if Asterisk or some other soft PBX still supports them? Also... plug the serial board back in and see if the BIOS does serial redirection... most Telco stuff had service done thru the serial port.
I have a college that wrote a fax system using those cards a few years ago, still in use by us, Brooktrout has a Linux SDK we used.
I once trashpicked an analog pbx system from an e-waste bin, ended up selling it minus the case and the power supply for $900 on ebay - best trashpick of all time!
As others have said, these things are usually controlled by the vendor over serial. A standard modem is usually connected to a COM port. They may have done something to allow the keyboard to be controlled over the modem for more control.
Have you tried using a serial console? Since there wasn't natively a video card, maybe the BIOS was configured with a serial console.
I'd give it a try as it had a multi-io board with a serial port. I bet there is a console there.
That would be my guess. A old dumb terminal like a VT-100. Would also explain the 2 rs-232 ports. I had a older AT&T system that only had 1 rs-232 for a terminal or you could dial into it. If you called the correct port directly and entered the right DTMF code you would get the modem. That could be why one of the cards has a DSP chip on it.
I have a Merlin 820 KSU and 8 Merlin voice terminals (that’s what they called the desk phones) installed in my house and I love the system. It even integrates with VoIP and my mobile phone. Amazing for a system that was designed in 1984!
That's pretty cool. I guess they knew IP phones were coming sooner or later, what with the ISDN buzz, and wanted to be sure they didn't have to make a whole new system just to accommodate it later. What kind of mobile phone integrations does it have?
@@kaitlyn__L This era, some of them would optionally notify your pager you had a desk call when you were away. That was a pricey ($3,000-5,000 IIRC) option. And that was hot back then. I kid you not.
Kaitlyn L The nice thing about the Merlin system is that just about anything can be used as a trunk line. For my mobile phone integration I just have a small unit that connects to my phone via Bluetooth, but also can be wired in to the KSU with just regular twisted pair. I can make and receive calls via my mobile from any Merlin voice terminal. It’s called the “Dock ‘n Talk.”
@@kaitlyn__L ISDN lines cost a lot of money to plain copper.
@@nbntelevision1 There are/were options to bung a SIM card in an onboard socket to translate mobile to IP /ADSL or PBX / POTS Lines
I mean obviously there are security features to prevent snooping of recorded calls! Also the 386SX is not a fancy 286. It has a virtual 8086 mode like all 386s and much more sophisticated memory management. It has a 16 bit external interface but is 32 bits internally. You should boot the thing with an external serial console (e.g. Putty) connected to COM1 at 9800,8,N,1. That will almost surely bring up an admin interface to the Merlin system. Don't forget the null modem.
If you still remember SEAGATE disk manager, there is an option to write a small boot overlay into boot sector, and override the Type 48, in order to use a large hard drive.
Brooktrout was famous of making voice and fax (back in the old days we also have the GammaFax multiline fax card), may be running DOS 5.0 and may have TSR to do background voice handling (from the GammaFax experience I got). The daughter board should be a replacement of oscillator, this board is just a bit newer to use DIMM instead of DIP (pin).
We had a similar Vocemail system at work up until around 2 years ago, and was even older (around 95-95) and ran NT4. This was being used for around 2000 users and we only replaced it when the hard drive decided to start failing (Luckily when it decided to fully fall over it did it on the morning of replacement).
NT4? Was it Meridian Mail?
Adrian - I think these systems output to the serial port. You should see if anything is there.
"Fixed Disk Controller Failure" means that probably won't help.
That is possibly a Quadtel BIOS. Phoenix bought Quadtel in 1992, so if this ATT/Lucent product existed before 1992 it would have been a Quadtel BIOS (later rebranded as "Phoenix" after 1992). Try "Ctrl-Alt S" for Setup.
FXS (Foreign Exchange Subscriber) is the port that actually delivers the analog line to the subscriber. In other words, it is the “plug in the wall” that delivers a dial tone, battery current and ring voltage. ... FXO (Foreign Exchange Office) is the port that receives the analog line. you can use them on a POTS lie.....omg i'm dying here
I presume you can also feed a POTS or similar phone into it, for listening / if fax, faxing in /out and also programming mailboxes for voice and fax 2 email or similar
Wow, had to click on this one because my father used to work for At&t and then Lucent. Thought it would just be for lulz, but then I remembered one day going to work with him about 20 years ago (I had the day off school or something...), and I "helped" him upgrade very similar looking machines from their operator pool. I had no idea what we were doing at the time, but we were swapping out EPROM chips in motherboards like this. I realize now they were probably the proprietary bios chips they were using either in machines like these, or their operator terminals. I remember him telling me later they used some proprietary *nix variant. Don't tell Ma Bell, but I kept a handful of the chips we replaced just to play with. Unfortunately, if they didn't get tossed, they are on the opposite side of the country in my parents basement so I cant just check them out... but thanks for the memories!
You CAN get custom BIOSes with custom initial parameters for when the CMOS battery goes on the fritz, but the system won't be crippled.
25 years ago, I worked for a company that churned out big LED matrix displays for advertising and they used a PC with such a BIOS so they could work without a keyboard and a display and even if the CMOS died. (It booted DOS on a floppy that contained the advertisements to be displayed; the floppies were mailed each month to the branches with new advertisements).
I used to work at the Lucent factory/Bell labs in Westminster CO where they assembled these merlin boxes and Definity switching cabinets, I was there 1998-2002 .i worked for IBM supporting Lucent desktop machines and servers. I believe this was a Lowtech security feature where you had to connect a modified keyboard or special keyboard that would allow you to interact with the BIOS. Basically to prevent someone ,other than a technician , just plugging a regular old keyboard and messing with the box. Also, what other said about the rs232, I think you can get console access with a tty terminal and see the exported video, like old school serial over Ethernet, so they could remotely dial into the box and diagnose it. I think all this was to make it like an appliance as much as possible, headless just plug it in and connect to your phone system.
Yep, 386 motherboards in custom-purpose hardware. Kind of like the CD duplicator we had at Future Vision. It was in a very sparse-looking beige metal housing, with nothing more than a row of membrane buttons, and a 2-line LED display. It could run 4 SCSI 4x CD writers at once (each on their own card/cable).
Some time later we had to open the system, and found that 386 (or 386SX) motherbard, 5 SCSI-1 cards (one each for the external writers, one for the SCSI HDD you would dump your master copy down to, then read off of to write to the drives). You'd program your choices of what you wanted to copy, how many copies, etc through a labyrinthine menu on that 2-line display.
12:19 That 'thing' and those bodge wires are almost certainly the issue. I've never seen anything like that on standard PC's of the era. I doubt that the BIOS, or keyboard controllers have been modified beyond something about that bodge wire and assembly intervening to probably prevent CMOS changes, as you are able to use the keyboard in DOS. Also, one of those bodge wires runs to a group of 5 pins, which might be related to keyboard or unpopulated PS/2 interfaces that a lot of these motherboards had provisions for, but weren't used. You might want to start with those bodge wires and see what happens if you disconnect them. Or, try to find a schematic for the motherboard, which looks super generic and is probably an unlicensed copy of somebody elses board, which was also super common for boards of this era.
thankfully someone else commented who used to work with these, the bodge wires are for converting the 286 into a 386SX, before later board revisions incorporated the needed changes natively.
I think the last line on the Dallas device is the date code + some chip revision, i.e. it was manufactured in week 06 of 1995.
I thought the 12887 was a fairly recent replacement part for the 1287..?
Exactly this definitely was not made in 00 but perhaps serviced then. It had to be earlier.
I’ve seen other comments here saying what I’m gonna say...drop a 9-pin serial/console cable into COM 1. Spin up a terminal and I’m pretty sure you’ll get output. Try various rates but likely as not 9600 will be the one.
Assuming it boots, then you'll need whatever key sequence or default or custom admin account names and passwords to get further.
BTW, the date code on that is year then week, then a 2 digit die revision code, so that RTC chip was made in 1995.
A few observations:
- Both Brooktrout cards use a DSP: On the later revision it's the small TMS320BC52PZ on the left side ("U2"). Always suspect a DSP when you see "TMS320" ;-)
Some boards are prepared to use different versions (package sizes) of the same chip and this seems to be one of them.
- You should check out these EEPROM-like chips on both cards if they are actually EPROMs. Often cheaper, non-reusable EPROMs are used (you can't erase them as they don't have a window like the large EPROM DIP packages). Better remove the sticker and check the IC number. If it's begins with "27C" you have an EPROM (I sometimes had luck with programming them ONCE, though). EEPROMs often have "28C" which work with your MINIPRO (a 29C would indicate a FLASH memory "PEROM" - also good).
- You point out the date on the I/O controller chip on the I/O card. Right next to it, on the card itself is also a datecode: 3895 ;-)
- You also missed the datecode on the DALLAS clock module: 9506.
- Also the 386SX is not "essentially a 286". It's to the 386DX what the 8088 is to the 8086: It can run the same software, but externally supports older hardware (and less memory) - and it was much cheaper. The 286 can actually be faster when clocked the same but it was obsolete the moment 32-bit software became the standard. The 386SX can run these, though (I believe Win 3.11 and later).
- As for the "oscillator module": You should look up the number on the TO-92 package. It wonder if it's really a transistor. Never seen such a module. Maybe desolder it and draw a schematic - as tiny as it is?
- Also: The long botch wire on the back runs from this module to resistor R16. This seems to go to unused turbo switch connector. You may want to check that out.
The "oscillator module" at 10:45 might be a primitive watchdog. Of course, the CPU needs some GPIO pin to periodically reset the timer (after it is enabled). Since since this box only has 1MB of RAM, I'd repurpose the A20 gate for that (which is famously controlled by the keyboard controller). A PAL (PLA, GAL, CPLD, ULP, ULA, ...) with feedback logic could be used to enable the watchdog when the A20 gate gets asserted for the first time after power up, and from there on, it needs to be periodically reset.
Would make sense to an unattended voicemail system to have a watchdog which automatically resets the system if the system hangs/crashes/behaves irregularly.
Sorry for mentioning the A20 gate. A truly ugly hack...and Intel CPUs supported it until Q2 2013.
Well, officially, Intel just mentions that you should not expect the A20 to work as expected on 64 bit Haswell (and later) architecture CPUs. That makes me wonder...just how does an Intel Core i9-10900X access more than 1MB of memory under DOS 6.22?
Oh, man... It's always so nostalgic seeing the old Lucent logo. My dad used to work for Bell Labs/Lucent.
Headset Guy yeah none of this avaya rubbish!
Lucent name brings back bad memories of fighting with Winmodems and "added value" software that came with them in the mid to late 90's. Before ditching dialup completely I resorted to using an external hardware modem.
Here's a suggestion on how to get this working: Make a daughter card that plugs into the PAL socket and implements the stock IBM logic (7400 and 7407) to interface to the keyboard, then replace the keyboard controller and the main BIOS with standard parts. This should bypass the PAL tomfoolery and get it to behave like a stock AT board.
Funny you're doing something on this. I run a 30 year old multi conference/chat line that uses old dialogic d/40b cards. Its a intel P2 400 mhz running DOS 5. The cards are from 1988 or there abouts. It has a dialogic bridge board D/80. The program is written MS C+ with some asembler code.
Really! Would you mind dropping the number?
I second dropping the number. Would love to dial in.
Nice find. 386SX-40 are very rare today.
I have an Alcatel-Lucent OmniPCX and it is one of the most custom and locked down things I have ever seen. I was just barely able to get a console working but it does follow the same principle as this device: A slightly custom computer (CPU) board/module and then multiple add-ons for whatever your needs were.
Thanks Adrian, your videos help us get through tough times.
Yes the vids are interesting but I wouldn't go that far...
I love seeing weird, custom PC stuff like this. Hope to see more tinkering with it!
The assembly at 10:56 is an oscillator. It has "OSC2" beside it in the silkscreen. At 14:23 you were wondering about the battery in the clock chip. The box said it is a Refurbished unit. The 07/00 date may be when it was repaired.
I feel like I discovered this video a bit late, but from my memories of obtaining a Merlin Mail system many years ago, and 'going down the same rabbit hole', the boards were configured for a specific set of hardware in certain slots.
I managed to use Linux to do a 'dd' image of the hard drive when I had the system, and I want to say it, or at least my system, ran some version of AT&T Unix or BSD Unix.
Having maintained the 'big brother' of this system, 'AT&T AUDIX', it wouldn't at all surprise me if the board was, in some way, custom for AT&T/Lucent...
I do remember that, on my system, the port labeled COM 1 had the same messages as your VGA port. It now makes me think that the bios, along with the custom PAL chip are modified by the manufacturer to enable use of a serial console for management and maintenance....
Good old Brooktrout cards. I've seen these used in fax servers, though I don't know if the ones you have are the same or not. That BIOS stuff can be a bit of a fiddle. BIOS manufacturers do include a number of "hidden" settings that can be set up using a special utility or something. Unless you happen to get lucky with finding the right one somewhere you probably aren't going to be able to figure out what needs to be changed to re-enable what's been disabled.
Brookcom were heavily into telcom cards in the 90's. My former company was using them until we switched to Voip that was in 2015. The Power supply likely has both a -5 and -12 one of them was removed so modern power systems don't have both. We had a hard time finding power susplies because of that.
I ran an AT&T system 75 for a few years (the Merlin is a small version IIRC). They NEVER crashed and we had DOS based 386 voicemail system for it until 2005. We eventually upgraded when ebay was getting expensive for parts.
Matt I know where you can get some spare parts for it.....
@@JesusisJesus That was a long time ago. I am looking for an old Merlin or System 75 though. I am building a retro BBS cluster and thought it would be neat to have a system like back in the day.
Adrian. Brooktrout cards were used a lot for fax services. Particularly a program called Zetafax. You can still get a virtual brooktrout device on relatively recent versions of the software for doing FOIP.
yes those are DRAM spots. 386 motherboards often had a choice where you could have both 30 pin Simms DRAMs ( usually socketed ) because early on there was a huge price difference between the two memory types. ( 8x128kbit +1parity... or maybe x256)
Those are set up to do a serial console. It's 1200 baud, 7 bit. (It was designed to have a modem hooked up, so AT&T could dial in to make changes)
The password to get in is probably: p1zza
They run Concurrent CP/M!
You’re right, it is a rude computer. At 22:05 the on screen text even says Foff.
Good eye! They probably designed it not to be changed or tampered with and put that at the end of the model number LOL
What's on the HDD??? Think of the 'gold mine' of information...
So disappointed the video doesn't go into any detail about booting the actual system on the drive or looking at its filesystem!
Years ago, we had a voice mail system at work that ran on a DOS box similar to this, complete with the DSP cards and the cheap IDE hard drive (a Western Digital in our case). The system it went with was a Panasonic; we ditched it all and put the Panasonic up for sale after switching to IP phones and Asterisk.
The board with the resistor diode and capacitor is a Bios lockout reset, Its design to lockout the bios, reset it to default pre programmed configuration after reboot, even if you manage to get in, and here's the kicker if you remove it doesn't boot at all.
Looking at that small board, it looks like it would heve been the spot for the 14MHz oscillator, and they instead removed it, to put in another clock signal, probably a divided down 66MHz clock, so likely either 16.5MHz or 8.25MHz, which is probably used by one of the add on voice cards as a clock synchronous with the processor, as pretty much no other cards would use the 14MHz clock anyway, unless it was an early CGA card generating NTSC video.
At 12:20 the windows logo was because there had been some glitch with Windows and some non Intel CPUs so I guess Bill seized the opportunity to make some logos and sell them! This logo thing went on for a while and was evetually dropped or maybe not because it is kind of the thing you could still see in some laptops today that the sticker is displayed to the owner instead of being buried in the CPU where nobody could see it anyway because of the fan and stuff, starting with 486 DX2-66 CPUs started needing fans officially recommended by Intel. Great video thank you!
i´d put my money on the often mentioned serial console. from a service perspective that makes a lot of sense. technican comes with his laptops hooks it up to the serial port and does his thing. no need to add a vga card or keyboard.
Super Nifty! Those old boards (when working!) are great as "floppy interface" machines for working with older 8-bit machines for floppy disk transfers, especially on computers that don't have flash systems, like old Kaypros, etc. I have a very similar board installed in my "386 Frankenstein" machine - sitting on the bench right now for work with a Commodore 64!
Yeah, I have a IBM PS/1 that i swapped out the 1.2 drive for a 360k and is primarily used to make floppies for my trs-80/kaypro/osborne. Technically it's a nice dos gaming machine if i had any interest in moving past 8 bit machines :)
@@brianv2871 Are there any computer boards for sales and trade other than ebay I would prefer to seek out my stock of 386s to be collected from me I have SIMMs as well, no means of testing them though
Avaya, previously Lucent and formerly AT&T used hacked up PC architecture for their Merlin Mail system. Merlin Mail was a standard VMX [VoiceMail Exchange] used for mid-market PBX [Private Branch Exchange] systems. It was pretty typical for voicemail systems of that era to use PC based architecture. Third-party voicemail systems, such as ActiveVoice Voicemail, often used off-the-shelf PCs. Even slightly later, on-card voicemail systems were still PC software running on specialized PC cards designed to fit inside of PBX cabinets. This was especially true after NEC acquired most of ActiveVoice and put their voicemail software on custom PC based cards (complete with DOS formatted flash drives) designed to be installed inside their NEAX and Electra Elite PBX lines.
As you found, these outboard voicemail systems integrated with PBXes using Analog Voice Cards that were connected to pooled analog extension ports of a PBX. When you called the VMX pool extension, your call was routed to a free extension number in the VMX pool that connected to one of the ports on those Analog Voice cards. Dialogic Corp made the cards. Almost all PC architecture based voicemail systems of that era used Dialogic Cards. Most of these voicemail systems used out of band signaling to and from the PBX over a Serial port. This means when you called the VMX pool number from another extension, the PBX would send information about the call simultaneously to the VMX over RS232 using the Serial port indicating the originating extension. The RS232 connection was also used for MWI [Message Waiting Indicator] functionality. The reason you were having so many issues getting that motherboard to act normal is that they were hacked up to run without a keyboard or pointer device. They where intended to bypass the keyboard check, and boot to a drive running a basic OS (usually DOS) and the proprietary Voicemail software. Technicians would program the voicemail using only the serial interface. Technicians would never connect a keyboard or monitor to the system, and most telecom techs of that decade might
not know how to service a PC. If there were an issue with the software, a technician would order a whole new drive or voicemail system with the software preinstalled.
FYI, voicemail software can be very sensitive to card in motherboard. i use to build one that was 486dx-100 board which had only one pci slot but 8 isa slots> when the video card got changed to a different model the software crashed. Only the one brand card could be used. The used i believe rocketport voicemail cards
I think they can work on the interrupt cycles and sensitive to voltage changes
That DALLAS chip not only has the realtime controller but also some bytes of memory and an include non-rechargeable battery - which drains slowly when the computer is powered off (and over time).
On SGI Indy workstations that chip also contained the Ethernet MAC and the serial number. If the battery was toast you needed to buy another one of those chips and use a hex editor to reprogram it byte by byte. I did that twice back in time.
Nowadays there is no replacement anymore for those.
I've looked over the random Brooktrout stuff of the time. The closest i can find to the telephony boards you have are the TR112 cards. It's possible for such a high volume customer, they made something custom. I've had the opportunity to develop for a lot of similar telephony boards - lots of fun, but the Brooktrout stuff has always kept their SDK a deep dark secret. If you take a look at the files on the hard drive, you might find a clue as to what these are; usually how these things work is there'll be an executable included along with the drivers that pukes a boot image into the processor onto the ISA board - sorta like a baby bird xD . The configuration file for that will more than likely contain some information as to the model.
For whatever it's worth, there are still some manufacturers making systems just like these (with more recent cards/motherboards, of course) for some of the more modern, less "shove an ethernet port into an ATX machine and it's a PBX" phone systems.
Heh, okay, it's actually a TR202 card, as per the sticker. There's a development manual on one of those shady document hosting sites: manualzz.com/doc/33471336/brooktrout-fax-and-voice-application-programming-interface
At the time I tried it, the document wouldn't load, so here's a Google Cache of it - webcache.googleusercontent.com/search?q=cache:0gIJ_ZnaF0QJ:manualzz.com/doc/33471336/brooktrout-fax-and-voice-application-programming-interface+&cd=5&hl=en&ct=clnk&gl=us
@Adrian's Digital Basement -
I always use F2 & DEL quickly pressed over & over, because these 2 are the most used keys on PC from 2000s (when someone gives me a computer to look in the BIOS, because of problems) ... ... But nice to know that you can read this stuff out.
Why didn't you check the serial console? Remember it didn't have a video card originally. Maybe it's completely controlled through the serial port?
I've tried to clean the splotches on my monitor for too long before I realized they were on yours.
At a guess the cap and transistor in the other crystal socket is a watchdog timer with a several second timeout. It probably pulls on NMI and resets the machine if it isn't retriggered before it times out. Possibly the two rework wires on the back of the board are part of this (presumed) watchdog circuitry.
I'm seeing a bunch of good ideas posted already. The only trick I would add, is putting it back the way you found it plus the video card and see if it's giving you additional clues as it's booting up the way to manufacture intended it
My brother worked at Dollar General for a long time. They used DOS on their point-of-sale machines, until eventually management decided to upgrade the OS, but not the hardware. They opted for Linux, which of course ran like a slug doing anything on those old machines in comparison to DOS. The registers took five minutes to boot up from then on, so any time there were power issues, from a storm, or whatever the case may be, it became a nightmare scenario of long lines and frustrated customers while rebooting. The logical approach would have been an intermediary machine to connect the old registers into any newer back-end hardware, but they were never much for common sense, and certainly weren't going to spend money if they could avoid it. I think K-Mart took a similar approach, especially when I recall their systems taking a long time to start up when they went through a phase of electrical issues. Though we can all see how K-Mart ended up.
I was hoping to see what happened when you boot it from the HDD.
It's likely to present a servicing UI of some sort, that may be more interesting to see. I've dealt, in the past, with a sizeable fire detection system for multi-building facilities that was a 386 chip with custom mobo and software. It was a reasonable choice for the era, as it was for a voicemail system. I'd really wish some more delving into the software side of the thing, i.e. filesystem, software and servicing UI, rather than just trying to have a somewhat-working "yet another" 386 motherboard...
@@djangowatson217 I think, as many have pointed out, that this machine was configured to run over a serial console. Also, I think Adrian forgot to mention it but it seems like the hard drive controller was failing or failed to boot the disk. Either way, it seems the goal was to poke around a bit but ultimately using it as a standard 386 board.
you may just see a system running screen, most of configuration came from the phone system sending code to it, that is for the one I work on
Adrian -
You were on the right track. The only thing missing was a test of the Battery. If the Dallas chip is dead, you cant "Save" a config. It will default to the wrong drive time after time. These motherboards do not support "Soft Boot" any reset command is a real reset and Bios is started over. (Packard Bells that had 486 were this way as well, they only take power from the battery and not the power rail. Man I wished I was in your basement, this would have been a fun win! o and btw... Not likely he will use this card.... OOOOO NOOOOOOOOOO I will pay for them as will many others! that love BBS systems
Brooktrout was a company from Needham, MA, specialising in telecom cards. The brand still exists, but I think only for Fax boards - have a look for Dialogic. The boards you have probably cost a good load of cash when new, but with an ISA bus will be of no use to anyone these days.
Those were a pretty penny back in the day. Brings back memories of those smaller Merlins, midsize Partners, and bigger Definity systems that I cut my teeth on as a young engineer.
Damn FINE, many thanks from the "ex radio, then ex fibre engineer", in Australia's dominant Telco.
Retired at age 53 in Thailand, SO MUCH more enjoyable.
Keep up your TERRIFIC work, great viewing!
Mark
Late to the party. I met someone in the 90s who worked on systems like this. With ISDN trunk lines you would have industrial 286/386 motherboards hosting a whole bunch of ISA cards. The cards had freaking 486s on them to handle isdn and voice logic. Blew my mind.
The cards do have an order the first card can be a stand alone mm02 the second card can be ports 2&4 or ports 5&6. It will only work with a merlin system and normaly connects to the last of the station ports
Hi Adrian, I just found your channel great stuff, subed. I used to work on Merlin systems back in the early 2000's, when I was a tech at a local telecom business. It's been years, but I think I have a ton of service manuals for these systems in storage somewhere. Next time I'm in there, I'll look for them. I didn't work on these long,. VOIP was on the horizon, and we jumped on the Mitel bandwagon and ditched legacy AT&T products. We had one tech that still supported these for years though, for clients that were hesitant to mirgrate over to VOIP.
Im pretty sure there was a debug program in dos that you could write some value to a memory location which would cause the CMOS validation test to fail. You would then be prompted to access the BIOS by pressing del or F1/F2.
I found it
Solution 1: for AMI and Award BIOSs:
C:\DEBUG (at a clear DOS prompt, don't do this in DOS-box in Windows)
-O 70 17
-O 71 17
Q
Solution 2: for Phoenix BIOSs:
C:\DEBUG (at a clear DOS prompt, don't do this in DOS-box in Windows)
-O 70 FF
-O 71 17
Q
Didn't work. I just swapped the Dallas RTC with a with another one with a dead battery to promote that error but no luck. Nothing changes when CMOS is corrupt and it all just goes back to the default settings I show, no matter what.
@@adriansdigitalbasement Likely using the wrong setup.com. There were about eight different setup files to use, depended on the BIOS, and between PC, XT, 286, 386. Was wonderful when the setups started going onto the CMOS. Too bad, I just pitched a few manuals and PC books from that era, too. I'll see if there's stuff left in the attic (judging by the lack of room left, yes, I'm sure there is....)
You might try holding down a key at boot (just after the Trident pops) and force a keyboard error. Sometimes that displayed the key combo for CMOS setup. If not, then I'm pretty sure it's a boot disk and the right setup that's needed.
Just a guess. Tons of those type of systems would be hard wired or programmed not to ask for keyboard on startup. So possible why you keep getting a keyboard beeping sound on startup. They also did the same thing sometimes for missing vga cards. To counter act the error for missing video card. I had a partner mail vs system that was not working one day. I pulled the 2.5" ide hard drive out and found I think some kind of unix release on it. Anyway I was able to clone it using ghost over to another drive without a issue. Going around the 1 bad sector and problem fixed. As for finding newer date on the machine being built due to cards inside. I talked to a few companies who due have special embedded stuff back in the day. They would order bulk to get a better price. Since they mostly write the software to work best with that hardware or in case they need the special modifications. They might have 1,000 motherboards in stock modified the way they want it and thus built the same system for 10 years. But I'm sure you already know this. As for the comment below asking about field service of the device. During that time you would not trouble shoot the bios or plug in stuff. Once you found a dead board. You would order another one from the mfg. Nice vendor lock in.
Looks like a perfect CNC controller box!
For me it's pretty obvious what to do. I think it is beeping every 108 minutes and then you have to enter a special sequence: 4 8 15 16 23 42
Love your channel btw :) Love this hardware hacking episodes. They are amazing!
Back in the 90's I converted an AT&T terminal computer to a dos/windows system. With the model I had, you needed an ATT boot disk to access the bios program. might be the same for this system. I was lucky enough to find one on an ftp back in the day.
I think a better description of the 386sx is that it's a 386 with a reduced interface design, similar to a Motorola 68k -- it has a 24bit/16bit external bus, but executes 32bit instructions.
Early 386SX boards straddled the DIP DRAM/SIMM card switchover and often having sockets for both.
The system was probably built in 1996 or so. and when the hard disk failed in 2000 they sent it back to Lucent and replaced the hard disk but kept all the other hardware the way it was.
The Dallas chip has 9506 on it so it's probably original.
Have you tried connecting something to the COM1 or COM2 ports? Maybe the modified BIOS does something with them. After all, there must have been some way to connect to this thing and do some configuration that couldn't be done with just a phone.
A few jobs ago, we had an NEC PBX that had an old Dell Optiplex tower running OS/2 handling both voicemail duty and as a programming terminal for the PBX itself.
Heh interesting. This had DOS on it and maybe a serial interface as there was that COM port I never checked when running the DOS based VM software
Watching your attempts to get to the BISO setup. I noted; One is you waited to do the key presses too long. With these old PC systems it was always better to start trying the keypresses before the post messages had come up. As for the beeping. That was saying the keyboard buffer was full. Tip, try one or two key combo presses and if you don't get in before the end of post or you get keyboard buffer full beep as you did then restart and press another again. The keys to try are "ESC", F1, F2 or Del. Having to use the control key sometimes was needed but not as often as turning on and hitting the above key.
p.S. My bad I started this comment before the end of the video but it might still give some ideas
These are really really cool. I used to admin one.
That one ran Linux. It was built just before the turn of the century. It had a serial terminal.
The bios memory chip battery back would still be working as being always on so did not much drain only on power cuts. I have used ones from systems like this and they still keep going. Having worked on these system we mainly used the RS232 to fault find and program them. So boot and connect up to RS232.
I do remember something about the keyboard and the 286-386 boards so will have a dog in my loft for old manual. Similar problems with epos boards to enter bios you had to hold down non standard keys. I want to say it was something to do with memory over 640 on 286 but it was a long time ago.
that 'home made' oscillator is sick! Wonder if it's to under or overclock the thing or just because they didn't have the actual part when they refurbished it
Well, to be honest, I started watching thinking this doesn't really interest me. Boy, was I wrong? I was rooting for you the whole time. Very crafty using that alt PhoenixBIOS on a 720kb floppy -- nice one. I'm looking forward to seeing how this one might pan out in the end. Great work!
It has kind of been a dead-end this silly thing. Was hoping someone with better skills than I have could unlock the stupid BIOS to allow this to not be a total POS. :-)
@@adriansdigitalbasement LOL. Has no one in the ADB fan crowd come forward to assist?
I've worked on several of those Bell Labs > Lucent > Alcatel-Lucent > Nokia systems.
The BIOS has been modified to treat all text modes as a serial console like seen on non-intel UNIX systems for years. This is similar to the management ports on newer rackmount Intel server systems where no local keyboard/display is expected.
Pretty common stuff today, in the mid 90s it was totally non-standard.
I'd imagine you could trace from the keyboard controller over to ports on the ISA bus for part of that redirect.
That machine would have run a version of AT&T UNIX when it was in service. You might have success taking an image of the hard disk with dd and trying to mount it under linux as a loopback filesystem.
As far as how old the hardware was for the installed date, Lucent buys a huge amount of hardware for their supported platforms so they could sell and maintain the hardware for decades, as people don't replace their voicemail system or PBX system very often. Incidentally, Lucent would charge several thousand for those machines more than the underlying hardware would warrent.
Also, its pretty common for them to do component level troubleshooting to this day. Those bodge wires indicate the box has likely been repaired and redeployed at least once.
The line cards are more likely to be DS0 boards than analog/POTS lines. The software had logic to transfer someone to another extension or forward to an external number that can only be accomplished with the DS0 signaling (POTS lines only really know on-hook and off-hook states).
The boot up beep when a key is pressed... Does remind me of my old 286 unit with a wrong AT/XT switch on the keyboard.