Fun fact. A friend of mine used to sell Cirrus Logic chips in the UK. The offices backed onto Leavesden Studios where they filmed the James Bond films (pre Harry Potter days). She used to regularly have to explain the noise of the explosions going off in the background (from the studios) to callers. That's the sort of RETRO TECH STORY you came here for, you know you did.
26:47 8514 was hi res standard introduced with ibm ps/2. 8514/a adapter (mca bus) had only one mode 1024x768 256colors 43,5hz interlaced, no text modes or lower res. those were handled by ps/2's onboard mcga or vga chips. 8514 monitor supports 640x350, 640x400 , 720x350 and 720x400 at 70/31.5 , 640x480 at 60/31.5 and 1024x760 at 43.5/35.5. it is not "multisync" monitor and there is no support for 800x600 res.
I had the same card! It used to be in a PC I called PITA (Pain in the "butt"). I also noticed the TTL output was active when VGA was enabled, but never had a monitor to sync up with to see exactly what it was doing. This video truly made my day, after 26 years of wondering.
Here is what I think could be the problem with the CGA: * Check to see that you are not mapping high memory into B8000-BBFFF (CGA Memory). EGA/VGA use the A0000-AFFFF region and not all cards use the B0000-BFFFF region, although some do. Note: Some memory managers skip the last hex digit, and use the left four hex digits as a four digit address. It is possible another card is using that region for something, or a memory manager is mapping high ram into that region for device drivers.
@@SianaGearz it is possible that the trident card shares the ega/vga memory for this. I am thinking the dual output card with so many modes may be using different memory areas that mimic the old video modes. I have read that some EGA/VGA cards do not use the b8000-bffff region and others do. It probably reflects how the cards simulate the old video modes.
1:07:49 The original, unmodded Doom is capped at 35fps. Although, you were pressing your nose to a wall in a small space, so that's an artificially high framerate. But most of us played Doom closer to 10-15fps back in the day. Not a lot of people had the hardware to run it at full speed when it first came out, so it has a nostalgic look to it when it chugs.
You can easily add memory modules to that 286 board with SIPP modules without having to source or make SIPP modules. But you have to have some spare 30-pin SIMM sockets. Those can plug directly into the holes of the SIPP connectors and have regular 30-pin SIMMs installed in them. It works, I have one of those motherboards with SIPP sockets and this is how I use it.
@@adriansdigitalbasement2 have you heard of DooM8088? that might let you run [at least] Episode 1 of DooM on that 286 board, especially if you've got plenty of EMS/XMS available
We need Adrians digital dungeon. That's where we can put the copiers and printers and all the other retro hardware that the sight of makes us want to kick it and pagers. Plus there is no limit of chinese crap that cost under a tenner that can go in.
1:11:51 On the 286 board, there’s an oscillator OSC1 hidden between the two 16-bit ISA slots closest to the CPU. I can’t tread the frequency marked on it, so I don’t know if it’s used for the CPU clock or the ISA bus clock…
Heh yeah, I didn't see that. I have been working on a video for the main channel relating to this motherboard and I finally found that crystal when looking at things a bit closer:-)
I think that was all the fun of all the old hardware; making it work after hours or days of frustration. I do miss it as having things that just work leave you somewhat disconnected from what to do when things don't go exactly as planned. I enjoyed this video Adrian.
Get a modern ASUS gaming motherboard for AMD and you’ll get to relive that experience - except now with hundreds of settings and no documentation! Many hours of fun await
I confused myself. Earlier today I watched the video updating the 286 board. Then I watched this before dinner, paused when you reached going over the MBs. When you started going over the 286 I was thinking "didn't he go over that earlier in this video?" Then I realized it was in the next video I already watched.
I think VBIOS corruption on that Oti video card is worth exploring. After all, all the custom crazy hacked modes just work, so by all reason, the video hardware is alright! Perhaps there's something wrong with initialisation as performed during mode switching by the VBIOS. I mean the way you switch video mode on PC, you populate registers for function (AH=0 for video mode change) and parameter (AL=mode number), you pull on a 10h interrupt line, the BIOS intercepts that and sets up the video hardware. You expect then a jump table in the VBIOS that executes different code depending on the parameters, and there may be bitrot affecting initialisation code for mode 04 in particular. You should also toggle reserved dipswitches. I suspect the switches are just read and evaluated by the VBIOS, they probably don't do anything on the video chip itself, just connected to it as GPIO. Otherwise there would be no point populating "reserved" dip switches, but if they're fully software evaluated, they can change VBIOS and the printed manual last minute and add or remove functionality, while the boards are already being soldered up. You also don't have the manual for your exact card, that's the problem.
I don't think so ... it would be highly unusual that all the various modes work perfectly except just CGA graphics mode though, based on some switches that seem to do nothing else.
Note that DOOM runs in protected mode using its own dos extender (DOS/4GW) so you don't need emm386 to run it (edit: actually FastDOOM seems to use DOS/32A rather than DOS/4GW but the same thing applies!).
SIPPs weren't good if your computer was subjected to any vibration. I remember having to replug them from time to time, to keep them recognized and working with the system.
That 386 DX 40mhz was my first real PC. I was stuck on a Commodore 64 for over a decade due to finances. EDIT: Glad you are feeling better! Hope you get all the way back up to totally awesome!
I suspect the "Multisync" monitor profile outputs nearer to TTL logic levels instead of .7vpp levels that VGA uses. The analog output probably doesn't reach TTL levels, but the "analog on TTL output" is probably something like 1.2vpp, hence the blown out video.
You can insert another card, an additional MDA card, a monochrome display adapter. That may enhance the overall performance, and you can remove the mda card. MDA card has a higher resolution for the text, and a jumper is needed.
11:29 The GD5401 is equivalent to the AVGA1 and is "simply" an SVGA chip, no acceleration. It has clock and RAMDAC included, so it IS highly integrated though.
SIPP = "Single Inline Pin Package" was how it was described to me. Essentially its a 30 pin SIMM (Single Inline Memory Module) that has a pin strip soldered on. The combination of pin header and pins was cheaper than the 30 pin socket to manufacture, it was very fragile and fell out of favor quickly. Ram was so expensive then that I desoldered 8 or 12 of the things to reuse the module in other boards.
You brought some great memories back! I had a card with the same or similar chipset (256 kB) I added to my PS/2 30 because I was so disappointed with its MCGA mode not emulating EGA. I remember CGA mode was working pretty well in VGA mode though, and I loved the extended text modes.
Great work. Really enjoyable. Would appreciate a video on setting up a NIC on an old bord (286, 386, 486) and getting it talking via DOS to a home server (NAS, or windows shared network, linux, etc...). Then it would be easy to grab any file you want from the server and pull it down onto any test bed with that NIC setup. So, if you have the time or you are interested in messing around with that, let it trip. I'm all in.
Often times, there is a jumper to disable VCC to the processor so you can use the socket. If not, add one. I've done this on boards with soldered 486sx chips.
I would double check the pins of the main OTI 067 chip to ensure no pins are touching or coming loose. Also check the resistor networks, I have had multiple cards have broken resistor networks which usually results from physical trauma and is quite obvious if you try to bend the component with the issue.
I recycled so many old motherboards, vid/nic/audio cards etc.. Wish I still had em. I started out on a tandy trs 80 gen 1.. lol so many years ago.. Another great video, ty!
I'm glad that you are feeling better. Once you replace the battery on the motherboard. You can use the chart that I emailed you to set the for systems that don't above the year 1999.
Idea for a project: You should be able to wire a 72pin SIMM module to this 286 board. Solder some wires to a 72pin socket (spare or from a scrap board) and push them into the SIPP sockets. Connect D0-D7 to the first socket, D8-D15 to the second, D16-D23 to the third and D24-D31 to the 4th socket. A0-A11 you can get from one of the sockets, the four CAS lines you can get from each of the four sockets, the two RAS lines from socket one and three (or two and four). I mean, there are adapters to turn four 30pin modules into one 72pin one, why not do it the opposite way?
Bits und Bolts published a 30p SIMM board design which takes EDO chips from 72p modules and optionally reconfigures them to run in FPM mode. With a sliding switch so you can select between EDO and FPM modes. As a rule, EDO compatible boards will run FPM if they have to, but the other way around doesn't work without some hackery like he implemented there.
I have a 386 board with SIPP sockets, and I've thought about doing the same thing. It should be about like the boards people make to adapt SIMMs to Amiga 3000 ZIP RAM sockets.
I've never seen a SIPP module, but it seems nice and simple. With how often SIMM slot retaining tabs get broken, I like the simpler idea of SIPP which just uses commodity pins and sockets and should be more serviceable. But I don't know how sturdy the pins are. When I was working on a Sega Genesis I noticed it uses ZIP memory - which has the pins almost inline, but in a staggered formation to squeeze the pins as close together as possible. It was an attempt to shrink the size of through-hole memory ICs - which tend to be numerous on computers so they can eat a lot of board space. Apparently it was trendy for only a short time before surface mount packages took over.
I had/have a 386DX40 that looks just like that and I remember the metal pieces breaking off the ram holders just like that! It happened when trying to push the ram to seat it! I kind of want to go look for it and see if it's the same brand! Maybe they had some bad ram holders! Great video! Thanks!
I've got a really awesome bootable 6.22 2GB (I think?) USB stick I created back in the day with the different HDD diagnostic tools and other things along with a very comprehensive/customized boot menu. I don't use it at all anymore, but, I keep it around because it's just really cool!
I had one of those oak VGA cards with no manual - but I did manage to get it to output VGA to my cga monitor just with wrong colors... I'd have to adjust the vsync but it meant I could play VGA only games, though in potato mode, so that's was nice
I've never actually been able to figure out what the heck Dave Jones is saying -- comeagotcha? I still have no idea. We need an aussie to translate for us!
According to the venerable Urban Dictionary: come-a-gutsa Verb. (Australian colloquialism) To be thrown off a moving object at high speed, arms and legs flailing, landing on a hard surface and sustaining multiple wounds and broken bones.
@@adriansdigitalbasement2 said like that it is lol, it can refer to something as simple as a child falling off a bike or a swing as long as there is a fall and a small injury (even pride) then you have come a gutsa.
As an Aussie, I always thought it just meant catastrophic failure, I never knew it had that specific gory definition but the guts in the name should’ve been a clue.
The problems with the picture quality on capture might be due to the pixel clock on the graphics card interacting (intermodulating) with the sample clock on the capture device. Dunno how to fix it, but one idea might be to try some sort of low pass filter on the video lines, which should be ever so slightly over the pixel clock frequency of the video card.
Since the table on the right side of the page at 27:37 tells you that VGA is the default, I assume that picture on the left is the VGA configuration, so you should be able to figure out which switches are up and down by cross referencing the picture of the DIP switches with the VGA defaults.
Would you also please dump the VBIOS from it, if at all possible, and document the card how it looks? Because i suspect the cause of issue may be VBIOS corruption. Video mode change on PC is by via VBIOS 10h routine which will have a jump table executing different code for different modes, and perhaps specifically the code setting up the card for mode 04 is corrupt by bitrot.
Here's the thing I don't get about DOS video output and digital scalers... What is causing the dark vertical lines in a solid background? Even if you were sampling 640 pixels instead of 720 (or vice versa), and even if you had the phase off, the signal for a solid background color should be pretty much DC for the entire width of that line. You could sample _at any point_ and get the same value, except at the edges. So what's causing the sampling artifacts? Are some graphics cards emitting black video _between_ pixels? This could be interesting to look at on a scope.
Quite a bit of flicker and focus hunting on the face cam, what camera are you using for that? because it seems like it's not aging well. @53:33 You can notice it most in the upper left corner of the video, that grey square.
55:16 The CGA emulation problem is probably a bug in the BIOS code. I can't see it being a hardware issue, given that the card works in all the other modes.
At @1:09:30 I fondly remember 80286 motherboards with SIPP memory and C&T Chipset. Those were right at the end of the 80286 era. 😎 At @1:14:14 Yes, that is an official C&T BIOS, not AMI nor Award. 😎
I played DOOM beginning to end on a 486 SLC-33 back in the day, in a little window on low-detail mode. It's the best I could scrounge up on a teenager's income. I really wish FastDOOM was a thing when I was 14.
how dare you diss the canadian ati all in wonder XD i had one that could display more colours than VGA and do stereo vs the sound blaster mono UNFORTUNATELY no games supported those modes XD so it kind of was a PoS, great video Adrian!!
I once had a VGA card with TTL out, it seemed like the TTL output was determined by the palette index of the VGA mode, not the actual colour of the VGA pixel. Had interesting results as when the VGA palette was set to all black like while a game was loading things and the game draws things to the screen while everything was black in the VGA palette the TTL output would reveal what was actually being drawn into the video ram even though it was all black on the VGA output.Have long been curious about what they actually do, The strangeness of the EGA output with vertical lines running in 256 colour VGA mode in Doom has me think that it might be trying to the spread the 8 bits of the VGA palette index across every pair of pixels of the scan doubled TTL output instead of just outputting the same for each pair. It's certainly not trying the colour match the TTL output to the VGA output it has to just be using some bit trickery on the palette index. In 16 colour modes you'll get identical output on both outputs unless the 16 colour VGA palette is reprogrammed to be different from the standard 16 colour cga palette.
56:28 I’m sure I saw a video on YT recently by an Aussie guy with a board very similar and he was able to get it running with the onboard cpu and external one fitted, I think it was as simple as fitting a jumper wire on the pcb. It’s late I’m on my phone so I’m not going to go searching but ping me if you want info I will check
Ah yes, I can see that. The 386 and 486 actually have a pin that completely disables the CPU. This is used on Intel 486SX motherboards with a soldered CPU so you can replace it with the overdrive or upgrade 486DX. So it would make sense the 386 boards with a solder CPU also have this pin exposed as a jumper.
That OTI card may require something like a VESA adapter on the option expansion. Some of the Voodoo cards and I think some of the S3 Verge cards did something similar.
I would have loved to see you put a WEITEK Abacus FPU in that 386DX 40mhz. I had one with an AMD Processor and it was pretty speedy for a programming college student back in the day! I got it at a computer fair and I could barely afford MB of RAM for it at the time. I ran it in the box it came in and I hacked apart the PSU from my AT&T 6300 (RIP)
It's possible that the RAM from that board was not removed by hand, but by having something dropped on it (presumably when in that box o' junk?) that dislodged the RAM from the sockets, bending the clips in the process, though being metal clips it could be possible to bend them back, though how easy that would be without damaging them even more, I wouldn't know... :)
The CGA problems could be a high memory conflict, I'm not sure what you're loading up there but CGA used a different set of memory addresses for direct access than does EGA or VGA and you may be loading something into that space.
I played Doom on an A1200 with a 25MHz 030 back in the day, I'm pretty sure there's folks who've used worse. I've learned to never underestimate what someone would use to play Doom back then.
The first "home computer" we had was a 386SX-25. Somehow or another we not only played DOOM on it, but even made our own silly mods. Of course, when we got that first Pentium and saw DOOM as it was meant to run, it was incredible.
@@CaptainSouthbird Home computers weren't a thing in England between Commodore going bankrupt and around 1999 unless you had some serious cash or worked from home. A PC good enough to play games would've cost around 3-4 times the price of an A1200, which itself was about twice the price a game console. Was a real oddity, went from 80% of households having a computer in the '80s, and by around '93 or so it had practically reversed.
My Magewell capture card has a similar issue with autodetecting graphics vs text mode resolutions and as you might expect, I get those same lines in games sometimes, though they can be difficult to detect depending on the game.
I know I remember using an Oak card with the feature connector, but I also remember my 386SX/25 having an Oak card too. Doom was barely playable in fullscreen but it was much faster with the screen reduced to fill about 1/4 of the screen.
I do support this experiment, on principle, but why would there be a switch to break the cga graphics mode and lock up the computer, that seems like the most useless switch in existence.
@@SianaGearzIt might be for multi-monitor support - I don't know if this would even be possible with a VGA card, but I think that was why you could disable CGA emulation on the PGC Edit: That probably isn't what's going on, because then the switches would probably disable the CGA text modes as well
@@SianaGearz It might farm out the CGA to a daughter board connected via the feature connector for maybe composite support? Or it replaces the CGA modes, which by this time were mostly useless, with a different set of features. It could be a factory test switch or something like that and the loss of CGA modes is a side effect. *shrug* I seen weirder things
@@OffstagePfaffa Problem with hypothesis 1 is that for the feature connector to work and be usable with a daughterboard, the computer has to not lock up in vbios routine int10 0x0004 first. Problem with hypothesis 2 is that you have an endless unexplored space of possible mode switch instructions, there's no need to override 0x0004 in particular. A switch that kills CGA for no particular good reason is a better working theory, if it indeed turns out that the switches matter :D
I always got a kick out of how turbo mode really was the opposite. When you close the switch you got a slower machine. It really should have been called De-Turbo.
@@Synthematix It ate my comment -.- short version - the performance botleneck is because writing to a card takes time, while it blocks the shared bus and the CPU. So writing half the image each to two cards takes up as much time as writing one whole image to one card. Nothing to be gained. The second issue is that if you put two VGA cards on ISA, you get same image out of two cards, if you were to solve their VBIOS fighting issue. Multi card things only really became possible on PCI.
Is there a clear explanation of what VRAM was used for on such cards? Obviously theres a frame buffer, but if you can have cards with such an ostensibly ... relaxed ... chip such as that OKI, what did the other 265Kb do? Presuming buffering and stuff but as Adrian says, it's a no frills chip that doesn't seem to provide any specific acceleration? Thanks!
There is no datasheet for oti 067 that i can find but there is for 077. I don't know whether 067 supported hi colour mode, i have severe doubt. It can do 1024x768 at 16 colours and 800x600 at 256, that's both half-megabyte graphics modes. Even 640x480 at 256 colour needs a little more than 256k. It doesn't look like it has bit blit feature. You can support scrolling and page flipping (double buffering) on any VGA adapter, if enough memory is present. So you can have two pages of 640x400 256-colour to flip between. Hey univbe setup should be good to discover what it can do.
I have a stack (like 6 or more) of these half-height OTI067 512k cards. I kinda wanna hack the TTL output into these now. They are pretty slow tho.I don't remember them not being able to display CGA graphics. That was in the SVGA jumper setting though. I do run prefer Windows to use the higher resolution graphics modes. Enabling Fast Gate A20 makes himem.sys crash on almost all of my machines and I have a lot.
Someone made a version of the Graphics Gremlin with HDMI out. But it's EGA, not VGA compatible, i think. Also it's only 8 bit. HDMI can be connected to most DVI devices and most capture cards are HDMI.
How do you glue a radiator onto the 386? Is it a permanent solution or it can be removed? I am thinking of doing something like this to my am386dx because it seems too hot at 40mhz.
For one you can buy thermal double sided adhesive tape. One made by Akasa is likely to be available from your nearest PC store, otherwise thermal pad brands generally also make this 0.1mm thin tape as well. Note there is a difference between pads and tape, pads aren't adhesive, and come at larger thicknesses 0.5mm and more. This is removable, but it holds on strong, so you can damage the chip, so perhaps it can be removed a little easier if you heat up the heatsink but it can stand up to pretty high temperatures, i think it changes little with reasonable heat. I have not tried removing it using alcohol or acetone, maybe it helps. The way many glued on heatsinks were done in the past is using thermal conductive cement, you can buy nondecript white tubes for pennies. I see no reason to go this route. It is not really removable, damage likelihood is very high. Third option is using electronics silicone, say Kafuter, its thermal conductivity is eerily good, 1.6W/mK claimed, it doesn't actually reach that for sure but no matter, it's much better than one would expect. It may have trouble curing in such a thin gap, but so far i have been lucky. It is removable, but i don't think heat can help remove it. Perhaps there exists a solvent which can degrade silicone, but i have no experience.
In this case, the manual for the card explicitly said it'd work on an XT, so there shouldn't be any worries with 186/286-only instructions in the BIOS (like there is with some later cards).
![Adrian's Digital Basement ][](http://yt3.ggpht.com/ytc/AIdro_k7CrQRe71qN6c0Ua7A1P_I8KkxjqLa70u0zYeCvdvnKw=s48-c-k-c0x00ffffff-no-rj)
Fun fact. A friend of mine used to sell Cirrus Logic chips in the UK. The offices backed onto Leavesden Studios where they filmed the James Bond films (pre Harry Potter days). She used to regularly have to explain the noise of the explosions going off in the background (from the studios) to callers. That's the sort of RETRO TECH STORY you came here for, you know you did.
It was probably easier to explain those explosions than the ones from RIFA caps.
26:47 8514 was hi res standard introduced with ibm ps/2. 8514/a adapter (mca bus) had only one mode 1024x768 256colors 43,5hz interlaced, no text modes or lower res. those were handled by ps/2's onboard mcga or vga chips. 8514 monitor supports 640x350, 640x400 , 720x350 and 720x400 at 70/31.5 , 640x480 at 60/31.5 and 1024x760 at 43.5/35.5. it is not "multisync" monitor and there is no support for 800x600 res.
I had the same card! It used to be in a PC I called PITA (Pain in the "butt"). I also noticed the TTL output was active when VGA was enabled, but never had a monitor to sync up with to see exactly what it was doing.
This video truly made my day, after 26 years of wondering.
Glad you're feeling better Adrian! Thanks for the vids!
Here is what I think could be the problem with the CGA:
* Check to see that you are not mapping high memory into B8000-BBFFF (CGA Memory). EGA/VGA use the A0000-AFFFF region and not all cards use the B0000-BFFFF region, although some do. Note: Some memory managers skip the last hex digit, and use the left four hex digits as a four digit address. It is possible another card is using that region for something, or a memory manager is mapping high ram into that region for device drivers.
Very smart, very good.
WAIT WAIT why would it be an issue if the Trident and whatnot just works in CGA mode?
@@SianaGearz it is possible that the trident card shares the ega/vga memory for this. I am thinking the dual output card with so many modes may be using different memory areas that mimic the old video modes. I have read that some EGA/VGA cards do not use the b8000-bffff region and others do. It probably reflects how the cards simulate the old video modes.
I always love the longer videos. The Oak and Cirrus cards are a blast from my past. Love the testing videos.
1:07:49 The original, unmodded Doom is capped at 35fps. Although, you were pressing your nose to a wall in a small space, so that's an artificially high framerate. But most of us played Doom closer to 10-15fps back in the day. Not a lot of people had the hardware to run it at full speed when it first came out, so it has a nostalgic look to it when it chugs.
You can easily add memory modules to that 286 board with SIPP modules without having to source or make SIPP modules. But you have to have some spare 30-pin SIMM sockets. Those can plug directly into the holes of the SIPP connectors and have regular 30-pin SIMMs installed in them. It works, I have one of those motherboards with SIPP sockets and this is how I use it.
Funny you mention this -- I have been working on a video on the main channel this week around this awesome 286 board!
@@adriansdigitalbasement2 have you heard of DooM8088? that might let you run [at least] Episode 1 of DooM on that 286 board, especially if you've got plenty of EMS/XMS available
now up to almost 90 mins, it still keeps getting better and better =)
I missed the longer form videos. Yay!
"Welcome to Adrian's Digital Dungeon with special guest: The 8-Bit Slot."
We need Adrians digital dungeon. That's where we can put the copiers and printers and all the other retro hardware that the sight of makes us want to kick it and pagers. Plus there is no limit of chinese crap that cost under a tenner that can go in.
When the PC/AT was designed, IRQ2 on the first PIC was assigned as the cascade interrupt to spill down to the second PIC that was added.
came to say basically the same thing, but then I scrolled and saw that someone had already said it. Fancy meeting you here... small world
Not only write what works/not works on cards but also date of testing so you can go back to the footage if/when necessary.
gotta love the Plexus at the background watching over what's happening at the bench
He’s like: “you guys be careful - he sticks a screwdriver in me … !”
Soon the Plexus will talk like HAL from 2001. "I'm sorry Adrian, I'm afraid I can't do that."
@@retropuffer2986he could have an Android named Data.;)
The Plexus is giving off the same vibes as the Crushinator robot from Futurama. 🤖
Wow... that's some old hardware! Glad to see most of it performed as expected.
Also glad to see you feeling better.
1:11:51 On the 286 board, there’s an oscillator OSC1 hidden between the two 16-bit ISA slots closest to the CPU. I can’t tread the frequency marked on it, so I don’t know if it’s used for the CPU clock or the ISA bus clock…
It is "something".0000 MHz, so that also fits the CPU/bus clock.
Heh yeah, I didn't see that. I have been working on a video for the main channel relating to this motherboard and I finally found that crystal when looking at things a bit closer:-)
I think that was all the fun of all the old hardware; making it work after hours or days of frustration. I do miss it as having things that just work leave you somewhat disconnected from what to do when things don't go exactly as planned. I enjoyed this video Adrian.
Get a modern ASUS gaming motherboard for AMD and you’ll get to relive that experience - except now with hundreds of settings and no documentation! Many hours of fun await
honestly, i like these test and check videos just fine. i'd rather see some hardware tested than just watch you open packages or something
I love these circuit board vids. Love seeing soldering action too.
I confused myself. Earlier today I watched the video updating the 286 board.
Then I watched this before dinner, paused when you reached going over the MBs. When you started going over the 286 I was thinking "didn't he go over that earlier in this video?"
Then I realized it was in the next video I already watched.
I love your channel so much... no politics, no agenda just unadulterated nerdiness.
I used to have one of those Oak VGA/TTL cards with the feature connector. 8514/a is an IBM proprietary extension to the VGA standard.
I think VBIOS corruption on that Oti video card is worth exploring. After all, all the custom crazy hacked modes just work, so by all reason, the video hardware is alright! Perhaps there's something wrong with initialisation as performed during mode switching by the VBIOS. I mean the way you switch video mode on PC, you populate registers for function (AH=0 for video mode change) and parameter (AL=mode number), you pull on a 10h interrupt line, the BIOS intercepts that and sets up the video hardware. You expect then a jump table in the VBIOS that executes different code depending on the parameters, and there may be bitrot affecting initialisation code for mode 04 in particular.
You should also toggle reserved dipswitches. I suspect the switches are just read and evaluated by the VBIOS, they probably don't do anything on the video chip itself, just connected to it as GPIO. Otherwise there would be no point populating "reserved" dip switches, but if they're fully software evaluated, they can change VBIOS and the printed manual last minute and add or remove functionality, while the boards are already being soldered up. You also don't have the manual for your exact card, that's the problem.
Did you try CGA mode with the DIP switch 5 and 6 set to the position the card came with?
I don't think so ... it would be highly unusual that all the various modes work perfectly except just CGA graphics mode though, based on some switches that seem to do nothing else.
@@adriansdigitalbasement2You should also try with the bus speed set to slow. edit: ahh, the very next thing you tried. Nevermind then.
Your 2nd oscillator is between the ISA slots
Note that DOOM runs in protected mode using its own dos extender (DOS/4GW) so you don't need emm386 to run it (edit: actually FastDOOM seems to use DOS/32A rather than DOS/4GW but the same thing applies!).
SIPPs weren't good if your computer was subjected to any vibration. I remember having to replug them from time to time, to keep them recognized and working with the system.
That 386 DX 40mhz was my first real PC. I was stuck on a Commodore 64 for over a decade due to finances.
EDIT:
Glad you are feeling better! Hope you get all the way back up to totally awesome!
I suspect the "Multisync" monitor profile outputs nearer to TTL logic levels instead of .7vpp levels that VGA uses. The analog output probably doesn't reach TTL levels, but the "analog on TTL output" is probably something like 1.2vpp, hence the blown out video.
You can insert another card, an additional MDA card, a monochrome display adapter. That may enhance the overall performance, and you can remove the mda card. MDA card has a higher resolution for the text, and a jumper is needed.
11:29 The GD5401 is equivalent to the AVGA1 and is "simply" an SVGA chip, no acceleration. It has clock and RAMDAC included, so it IS highly integrated though.
SIPP = "Single Inline Pin Package" was how it was described to me. Essentially its a 30 pin SIMM (Single Inline Memory Module) that has a pin strip soldered on.
The combination of pin header and pins was cheaper than the 30 pin socket to manufacture, it was very fragile and fell out of favor quickly. Ram was so expensive then that I desoldered 8 or 12 of the things to reuse the module in other boards.
You brought some great memories back! I had a card with the same or similar chipset (256 kB) I added to my PS/2 30 because I was so disappointed with its MCGA mode not emulating EGA. I remember CGA mode was working pretty well in VGA mode though, and I loved the extended text modes.
@1:04:00 Van Tassle is an Anglicization of Van Texel, meaning from Texel, and island in the north west of The Netherlands.
Great work. Really enjoyable. Would appreciate a video on setting up a NIC on an old bord (286, 386, 486) and getting it talking via DOS to a home server (NAS, or windows shared network, linux, etc...). Then it would be easy to grab any file you want from the server and pull it down onto any test bed with that NIC setup. So, if you have the time or you are interested in messing around with that, let it trip. I'm all in.
Often times, there is a jumper to disable VCC to the processor so you can use the socket. If not, add one.
I've done this on boards with soldered 486sx chips.
I would double check the pins of the main OTI 067 chip to ensure no pins are touching or coming loose.
Also check the resistor networks, I have had multiple cards have broken resistor networks which usually results from physical trauma and is quite obvious if you try to bend the component with the issue.
I recycled so many old motherboards, vid/nic/audio cards etc.. Wish I still had em. I started out on a tandy trs 80 gen 1.. lol so many years ago.. Another great video, ty!
I'm glad that you are feeling better. Once you replace the battery on the motherboard. You can use the chart that I emailed you to set the for systems that don't above the year 1999.
Idea for a project: You should be able to wire a 72pin SIMM module to this 286 board.
Solder some wires to a 72pin socket (spare or from a scrap board) and push them into the SIPP sockets. Connect D0-D7 to the first socket, D8-D15 to the second, D16-D23 to the third and D24-D31 to the 4th socket. A0-A11 you can get from one of the sockets, the four CAS lines you can get from each of the four sockets, the two RAS lines from socket one and three (or two and four).
I mean, there are adapters to turn four 30pin modules into one 72pin one, why not do it the opposite way?
Bits und Bolts published a 30p SIMM board design which takes EDO chips from 72p modules and optionally reconfigures them to run in FPM mode. With a sliding switch so you can select between EDO and FPM modes. As a rule, EDO compatible boards will run FPM if they have to, but the other way around doesn't work without some hackery like he implemented there.
I have a 386 board with SIPP sockets, and I've thought about doing the same thing. It should be about like the boards people make to adapt SIMMs to Amiga 3000 ZIP RAM sockets.
I've never seen a SIPP module, but it seems nice and simple. With how often SIMM slot retaining tabs get broken, I like the simpler idea of SIPP which just uses commodity pins and sockets and should be more serviceable. But I don't know how sturdy the pins are.
When I was working on a Sega Genesis I noticed it uses ZIP memory - which has the pins almost inline, but in a staggered formation to squeeze the pins as close together as possible. It was an attempt to shrink the size of through-hole memory ICs - which tend to be numerous on computers so they can eat a lot of board space. Apparently it was trendy for only a short time before surface mount packages took over.
I had/have a 386DX40 that looks just like that and I remember the metal pieces breaking off the ram holders just like that! It happened when trying to push the ram to seat it! I kind of want to go look for it and see if it's the same brand! Maybe they had some bad ram holders!
Great video! Thanks!
At 1:10:05 There's an oscillator ("OSC1") between the last two 16-bit ISA slots just up and left of the big chipset chip. Next to R16 and the PAL.
I've got a really awesome bootable 6.22 2GB (I think?) USB stick I created back in the day with the different HDD diagnostic tools and other things along with a very comprehensive/customized boot menu. I don't use it at all anymore, but, I keep it around because it's just really cool!
2 GB flash in the DOS days must have cost a fortune! I had a 32 MB JumpDrive in 1999 and I thought it was so cool lol.
I had one of those oak VGA cards with no manual - but I did manage to get it to output VGA to my cga monitor just with wrong colors... I'd have to adjust the vsync but it meant I could play VGA only games, though in potato mode, so that's was nice
I've never actually been able to figure out what the heck Dave Jones is saying -- comeagotcha? I still have no idea. We need an aussie to translate for us!
According to the venerable Urban Dictionary:
come-a-gutsa
Verb. (Australian colloquialism)
To be thrown off a moving object at high speed, arms and legs flailing, landing on a hard surface and sustaining multiple wounds and broken bones.
@@kpanic23 ROTFL, I never looked that one up. Quite a gory term!
@@adriansdigitalbasement2 said like that it is lol, it can refer to something as simple as a child falling off a bike or a swing as long as there is a fall and a small injury (even pride) then you have come a gutsa.
@@adriansdigitalbasement2 you should look at the switches. CGA mode is only usable in the proper configuration.
As an Aussie, I always thought it just meant catastrophic failure, I never knew it had that specific gory definition but the guts in the name should’ve been a clue.
Did you switch back the 5/6 switches before testing the CGA mode problems?
Adrian, when you are in the BIOS, if you do CTRL+F1 it can sometime expose extra options.
The problems with the picture quality on capture might be due to the pixel clock on the graphics card interacting (intermodulating) with the sample clock on the capture device.
Dunno how to fix it, but one idea might be to try some sort of low pass filter on the video lines, which should be ever so slightly over the pixel clock frequency of the video card.
Since the table on the right side of the page at 27:37 tells you that VGA is the default, I assume that picture on the left is the VGA configuration, so you should be able to figure out which switches are up and down by cross referencing the picture of the DIP switches with the VGA defaults.
just wanted to say that I love all your videos, keep it up!
I have a similar Oak card and will give it a go in CGA mode when time permits.
Would you also please dump the VBIOS from it, if at all possible, and document the card how it looks? Because i suspect the cause of issue may be VBIOS corruption. Video mode change on PC is by via VBIOS 10h routine which will have a jump table executing different code for different modes, and perhaps specifically the code setting up the card for mode 04 is corrupt by bitrot.
Always worth splitting the composite video signal and sending one output to a vhs video and the other to your capture device
SIP (in this context) stands for Single Inline Package, as opposed to Dual Inline Package (DIP).
Here's the thing I don't get about DOS video output and digital scalers... What is causing the dark vertical lines in a solid background?
Even if you were sampling 640 pixels instead of 720 (or vice versa), and even if you had the phase off, the signal for a solid background color should be pretty much DC for the entire width of that line. You could sample _at any point_ and get the same value, except at the edges.
So what's causing the sampling artifacts? Are some graphics cards emitting black video _between_ pixels?
This could be interesting to look at on a scope.
Quite a bit of flicker and focus hunting on the face cam, what camera are you using for that? because it seems like it's not aging well. @53:33 You can notice it most in the upper left corner of the video, that grey square.
56:16 I think ive seen this same motherboard model on the CPU Galaxy channel.
55:16 The CGA emulation problem is probably a bug in the BIOS code. I can't see it being a hardware issue, given that the card works in all the other modes.
They couldn't have just shipped a card with non working mode 04, that would be brutal.
But it could have been struck by bitrot.
7:00 Can you show / explain / share a link how you did setup the Extron with the Stream Deck ?
At @1:09:30 I fondly remember 80286 motherboards with SIPP memory and C&T Chipset. Those were right at the end of the 80286 era. 😎 At @1:14:14 Yes, that is an official C&T BIOS, not AMI nor Award. 😎
I vote that white is the switch setting and black is the gap
I played DOOM beginning to end on a 486 SLC-33 back in the day, in a little window on low-detail mode. It's the best I could scrounge up on a teenager's income. I really wish FastDOOM was a thing when I was 14.
I have a dual dac Analog/Digital TSeng ET4000ax made by Wyse. Was great for cad back in the day.
Not only were there no Fs given when removing the RAM, but many possible FFs were taken.
how dare you diss the canadian ati all in wonder XD i had one that could display more colours than VGA and do stereo vs the sound blaster mono UNFORTUNATELY no games supported those modes XD so it kind of was a PoS, great video Adrian!!
That vertically striped screen really reminds me of when you're looking at multicolor graphics on the Commodore 64 and change it to hires mode.
I once had a VGA card with TTL out, it seemed like the TTL output was determined by the palette index of the VGA mode, not the actual colour of the VGA pixel. Had interesting results as when the VGA palette was set to all black like while a game was loading things and the game draws things to the screen while everything was black in the VGA palette the TTL output would reveal what was actually being drawn into the video ram even though it was all black on the VGA output.Have long been curious about what they actually do, The strangeness of the EGA output with vertical lines running in 256 colour VGA mode in Doom has me think that it might be trying to the spread the 8 bits of the VGA palette index across every pair of pixels of the scan doubled TTL output instead of just outputting the same for each pair. It's certainly not trying the colour match the TTL output to the VGA output it has to just be using some bit trickery on the palette index. In 16 colour modes you'll get identical output on both outputs unless the 16 colour VGA palette is reprogrammed to be different from the standard 16 colour cga palette.
4:26 isn’t it you are experiencing ground loops with your capture device? That that is causing the lines for different types of equipment?
56:28 I’m sure I saw a video on YT recently by an Aussie guy with a board very similar and he was able to get it running with the onboard cpu and external one fitted, I think it was as simple as fitting a jumper wire on the pcb. It’s late I’m on my phone so I’m not going to go searching but ping me if you want info I will check
Ah yes, I can see that. The 386 and 486 actually have a pin that completely disables the CPU. This is used on Intel 486SX motherboards with a soldered CPU so you can replace it with the overdrive or upgrade 486DX. So it would make sense the 386 boards with a solder CPU also have this pin exposed as a jumper.
That OTI card may require something like a VESA adapter on the option expansion. Some of the Voodoo cards and I think some of the S3 Verge cards did something similar.
I would have loved to see you put a WEITEK Abacus FPU in that 386DX 40mhz. I had one with an AMD Processor and it was pretty speedy for a programming college student back in the day! I got it at a computer fair and I could barely afford MB of RAM for it at the time. I ran it in the box it came in and I hacked apart the PSU from my AT&T 6300 (RIP)
It's possible that the RAM from that board was not removed by hand, but by having something dropped on it (presumably when in that box o' junk?) that dislodged the RAM from the sockets, bending the clips in the process, though being metal clips it could be possible to bend them back, though how easy that would be without damaging them even more, I wouldn't know... :)
The drafting lab in high school had a 286 running 16mb of RAM, only time I've seen it.
The CGA problems could be a high memory conflict, I'm not sure what you're loading up there but CGA used a different set of memory addresses for direct access than does EGA or VGA and you may be loading something into that space.
I played Doom on an A1200 with a 25MHz 030 back in the day, I'm pretty sure there's folks who've used worse. I've learned to never underestimate what someone would use to play Doom back then.
The first "home computer" we had was a 386SX-25. Somehow or another we not only played DOOM on it, but even made our own silly mods. Of course, when we got that first Pentium and saw DOOM as it was meant to run, it was incredible.
@@CaptainSouthbird Home computers weren't a thing in England between Commodore going bankrupt and around 1999 unless you had some serious cash or worked from home. A PC good enough to play games would've cost around 3-4 times the price of an A1200, which itself was about twice the price a game console. Was a real oddity, went from 80% of households having a computer in the '80s, and by around '93 or so it had practically reversed.
My Magewell capture card has a similar issue with autodetecting graphics vs text mode resolutions and as you might expect, I get those same lines in games sometimes, though they can be difficult to detect depending on the game.
i love those old boards!
The 286's isa slot was broken that you tried first. Pins are very uneven.
I am always inspired watchin you work on electronics. What capture card are you using or use with your PCs commodores and Macs?
1:10:00 That motherboard almost looks like the one that I had in my old Zenith.
I can't remember the system it was in but I've used that Oak VGA card before.
I know I remember using an Oak card with the feature connector, but I also remember my 386SX/25 having an Oak card too. Doom was barely playable in fullscreen but it was much faster with the screen reduced to fill about 1/4 of the screen.
With the Oti card: try messing with switch 5 & 6, one of them might disable CGA.
I do support this experiment, on principle, but why would there be a switch to break the cga graphics mode and lock up the computer, that seems like the most useless switch in existence.
@@SianaGearzIt might be for multi-monitor support - I don't know if this would even be possible with a VGA card, but I think that was why you could disable CGA emulation on the PGC
Edit: That probably isn't what's going on, because then the switches would probably disable the CGA text modes as well
@@SianaGearz It might farm out the CGA to a daughter board connected via the feature connector for maybe composite support? Or it replaces the CGA modes, which by this time were mostly useless, with a different set of features. It could be a factory test switch or something like that and the loss of CGA modes is a side effect. *shrug* I seen weirder things
@@OffstagePfaffa Problem with hypothesis 1 is that for the feature connector to work and be usable with a daughterboard, the computer has to not lock up in vbios routine int10 0x0004 first.
Problem with hypothesis 2 is that you have an endless unexplored space of possible mode switch instructions, there's no need to override 0x0004 in particular.
A switch that kills CGA for no particular good reason is a better working theory, if it indeed turns out that the switches matter :D
Does Rammy have any SIPPs to see if that board can take the full complement?
The "CPU speed" setting only really toggles what the turbo jumpers do IIRC (open = slow or open = fast)
I always got a kick out of how turbo mode really was the opposite. When you close the switch you got a slower machine. It really should have been called De-Turbo.
Wonder if anyone ever made a driver to run two of these cirrus logic cards in "SLI"
Pardon? What good would it do?
@@SianaGearz Double the performance
@@Synthematix the cards need to be made to communicate with each other.
You would not get a performance increase.
@@Synthematix It ate my comment -.- short version - the performance botleneck is because writing to a card takes time, while it blocks the shared bus and the CPU. So writing half the image each to two cards takes up as much time as writing one whole image to one card. Nothing to be gained.
The second issue is that if you put two VGA cards on ISA, you get same image out of two cards, if you were to solve their VBIOS fighting issue. Multi card things only really became possible on PCI.
How many motherboard do you have test before put them back together that is crazy but educational at same time-DLH
Is there a clear explanation of what VRAM was used for on such cards? Obviously theres a frame buffer, but if you can have cards with such an ostensibly ... relaxed ... chip such as that OKI, what did the other 265Kb do? Presuming buffering and stuff but as Adrian says, it's a no frills chip that doesn't seem to provide any specific acceleration?
Thanks!
Just realised the obvious case being higher bit-depth!
There is no datasheet for oti 067 that i can find but there is for 077.
I don't know whether 067 supported hi colour mode, i have severe doubt. It can do 1024x768 at 16 colours and 800x600 at 256, that's both half-megabyte graphics modes. Even 640x480 at 256 colour needs a little more than 256k.
It doesn't look like it has bit blit feature. You can support scrolling and page flipping (double buffering) on any VGA adapter, if enough memory is present. So you can have two pages of 640x400 256-colour to flip between.
Hey univbe setup should be good to discover what it can do.
Almost makes me wanna dig out my old PCs languishing in my own basement. lolz
Did you try the 9 pin cga connector for cga ourput?
I have a stack (like 6 or more) of these half-height OTI067 512k cards. I kinda wanna hack the TTL output into these now. They are pretty slow tho.I don't remember them not being able to display CGA graphics. That was in the SVGA jumper setting though. I do run prefer Windows to use the higher resolution graphics modes.
Enabling Fast Gate A20 makes himem.sys crash on almost all of my machines and I have a lot.
Are there currently (DIY) projects creating FPGA based VGA ISA card swith DVI/DP output?
Someone made a version of the Graphics Gremlin with HDMI out. But it's EGA, not VGA compatible, i think. Also it's only 8 bit.
HDMI can be connected to most DVI devices and most capture cards are HDMI.
Regarding SIP RAM, I've always wondered if one could just solder some headers on a SIMM module
Yes, and the other way too. Soldered 4 sipps into an Amstrad that had 4 unpopulated simm sockets
@@rtechlab6254 Neat! :)
didnt packard bell use non standard RAM?
Adrian, let me know if you want any documents or firmware for your Extron gear. The company I work for is a dealer.
I have the 8-bit version of that video card.
How do you glue a radiator onto the 386? Is it a permanent solution or it can be removed? I am thinking of doing something like this to my am386dx because it seems too hot at 40mhz.
For one you can buy thermal double sided adhesive tape. One made by Akasa is likely to be available from your nearest PC store, otherwise thermal pad brands generally also make this 0.1mm thin tape as well. Note there is a difference between pads and tape, pads aren't adhesive, and come at larger thicknesses 0.5mm and more. This is removable, but it holds on strong, so you can damage the chip, so perhaps it can be removed a little easier if you heat up the heatsink but it can stand up to pretty high temperatures, i think it changes little with reasonable heat. I have not tried removing it using alcohol or acetone, maybe it helps.
The way many glued on heatsinks were done in the past is using thermal conductive cement, you can buy nondecript white tubes for pennies. I see no reason to go this route. It is not really removable, damage likelihood is very high.
Third option is using electronics silicone, say Kafuter, its thermal conductivity is eerily good, 1.6W/mK claimed, it doesn't actually reach that for sure but no matter, it's much better than one would expect. It may have trouble curing in such a thin gap, but so far i have been lucky. It is removable, but i don't think heat can help remove it. Perhaps there exists a solvent which can degrade silicone, but i have no experience.
SIPP = Single Inline Pin Package. At least that's what I think it means. 🙂
Bring the 286 board back and pimp it out!
I just finished up a video on the 286 the main channel. I totally love the thing!
Working in 8bit slot on AT doesn't mean it will work on XT. Some card can, but not all.
In this case, the manual for the card explicitly said it'd work on an XT, so there shouldn't be any worries with 186/286-only instructions in the BIOS (like there is with some later cards).
Great 😊