Working on a Super Rare TI-99/4
Вставка
- Опубліковано 3 чер 2024
- The TI-99/4A is a really cool machine with a pretty terrible design, but you know what would make it even better? If it was way worse! The TI-99/4 is what predated the 99/4A and it’s an incredibly rare machine, with only around 20,000 units built. The 99/4A get’s the “A” in its name from the TMS9918A Video Display Processor, or VDP. But, the 99/4 uses a TMS9918 (no A), which is why it doesn’t have an A in it’s name. The 9918 is much worse though, lacking a bitmap mode. Throw in the weird chiclet keyboard, and the 99/4 is worse in everyway, except for rareness and coolness. And I love rare and cool. So, let’s see if we can get this machine going!
(Spoiler alert, we don’t…)
If you want to support the channel please hop over to Patreon:
/ usagielectric
Also, we now have some epic shirts for sale!
my-store-11554688.creator-spr...
Be sure to check out RetroBits right here on UA-cam:
/ @retrobitstv
Come join us on Discord and Twitter!
Discord: / discord
Twitter: / usagielectric
Intro Music adapted from:
Artist: The Runaway Five
Title: The Shinra Shuffle
ocremix.org/remix/OCR01847
Thanks for watching!
Chapters
0:00 It’s good to be back
1:17 Introduction
4:11 TI-99/4 vs. TI-99/4A exterior
6:25 Disassembly
7:53 TI-99/4 vs. TI-99/4A interior
10:10 Cleaning
11:35 We’re gonna need a monitor
12:54 Monitor disassembly and cleaning
13:53 First test
16:06 Troubleshooting begins
16:45 A bad video circuit
19:40 A bad 74LS138?
21:35 A bad 4116 RAM chip
24:45 We’re getting there, just not there yet
26:10 Bunny! - Наука та технологія
TI99s have to be the hardest machines to troubleshoot, well done on progress so far
That is most definitely the truth there! This thing is a convoluted mess of obtuse engineering decisions, haha.
@@UsagiElectric A lot of engineering came from El Salvador, so that checks out
Back in the day, when testing for faults on 8-bit micro's, I used to "dead bug" the RAM chips. That is, sit a replacement chip on top of the suspect one with all the pins touching. Saved me a lot of time desoldering stuff...
I actually did that same test, pressing a new RAM chip atop the soldered in ones, but it didn't change anything, no matter what that line was always pulled high. Which feels like one of the 4116s failed with an internal short to VCC on that pin that was strong enough to overcome a known good chip sitting on top.
isn't it a "live bug" if the pins are pointed down, and a "dead bug" if the pins are pointed up?
When desoldering a bunch of chips, desolder one pin per chip, that is pin one of chip one, then pin one of chip two, then pin one of chip three and so on, then pin two of chip one and so on until you're done. This way the chips have more time to cool down in between the desoldering and have a greater chance of surviving the process.
May not best way to get thermal cycling old components. I think beter and faster to do preheat the board and do the continius desoldering with moderate heat then do it with cold board and components.
Good work. I used to use a FLIR camera to hunt for bad chips. Depending on the failure mode they don't always heat up but sometimes they do and it makes for quick troubleshooting.
Thank you!
Unfortunately don't have a FLIR camera at my disposal, but I did the touch test, and nothing really felt any hotter than anything else except the VDP, but they run hot already.
@@UsagiElectric If you leave your finger print on the chip its hot. :)
@@ahbushnell1 If you leave your chip print on your finger than it's hotter! ;)
I had transistor markings on a finger for about a week a long time ago.
I saw a video from Adrian's Digital Basement where he sprayed on some alcohol and watched to see how it evaporated. It disappeared off the hot chip super fast.
@@nurmr I've seen others use copious amounts of flux. It bubbles up where hot.
The "lugs" on the shield for the power supply wiring are feedthrough capacitors, for EMI filtering
Oooh, I didn't think about that, that at least makes some sense. It's still a nightmare, haha.
Yes, I call them pass-through capacitors and I mostly used them on RF to pass power or very low frequency signals. They usually were 1nF.
Shielding easiest fix is to spray with a coat of zinc based primer, or even aluminium silver spray paint, which will keep them looking nice. Otherwise you will need to chemically derust them, degrease and do a copper plate, then a nickel plate to get the surface bright with metal.
After he's wire brushed it all he needs to do is rub it down with wd-40
Or spray just it with clearcoat. The WD-40 will smell awful as the computer warms up.
@@HTMLEXP does matte clearcoat actually protect metal and keep its surface conductive?
@@SonicBoone56 the RF shielding probably doesn't need to be conductive across its hold surface, so the areas where full electrical conductivity is required could be masked off. Dielectric grease in the uncoated area could protect metal from corrosion. Clearcoat will be fine in an indoor environment. Clearcoat is used by those who want to preserve the 'rat-look' on their vehicles. It isn't as good a primer, base coat and then clearcoat of course.
I wonder if there's a company somewhere you can just send stuff to and they'll galvanize it and send it back.
I did development on the TI 990 using the DX10 operating system and COBOL. It was an amazing and fun machine.
That's awesome! I would love to get hands on with a TI990 someday, I feel like with the right programmer they can be an incredibly capable system!
I worked in a TV repair shop in high school around 1970. Color TV's have to be converged to fix your color problem. Look in the back of the monitor and see if there are adjustments for color convergence. If there is you may be able to fix your problem. Also there were magnets on the neck of the tube. I remember my boss would spend hours converging a set while I held a mirror so he could see the screen. Good video. Good luck.
That one might be a bit beyond my comfort zone! Maybe someday I'll carry it with me and go visit Adrian and get him to help me out!
@@UsagiElectric If the convergence controls are there mark the settings with a felt tip pin and try adjusting. Or maybe there is a manual on line. Keep up the great work!!
@@UsagiElectric The color fringing might not be a convergence issue; it could be a signal problem. You'd definitely want to check the monitor with a known-good test signal before you tweak anything.
I remember that fringing being a very common problem for computers of the 80's & early 90's. Maybe due to stray capacitance or to limits on the technology at the time. (either TVs or monitors on a budget)
I'm so glad I sent you my TI-99/4a I got when I was a kid. You're the person I always wish I could find for the things I never can do much with myself but want to find a good home for.
Someday I hope to come visit and use your TI-99/4A setup for a few hours because I've wanted to get a running system started from that one I found. I wanted to so much that it sat at the foot of my bed until I sent it to you, but I just never found anything else for it so it was always by itself. Getting to play with your TI systems would fulfill that goal I had, especially since the only part I ever found got to add to that collection!
I grew up with the 99/4A in my family. Nice to see a little bit of the heritage behind the machine. Easy to forget, at least for me, that it was a 16 bit machine that early back. I sometimes wonder where things would have gone, if they had been able to keep on developing the line of computers.
I was fortunate to live in Colorado Springs in the mid 80s. A TI engineer at the local office brought in a 99/8 prototype to our users group meeting. That was quite a treat. I also was gifted a 300 baud TI demonstrator modem (plexiglass case) by him when they closed up shop. Fun days.
Having data on the data lines usually means that the CPU is working. I would start focusing on the ROMs.
I think you're definitely on point here!
Several bits in the ROMs typically go bad. Best path forward is to unsoldered both ROMs (note which goes where!), Read them out, and compare to the known good code. Then program new UV EPROMs (with minor wiring changes). I spent WAY too long on other chips before discovering bad ROMs. I have a simple Arduino board and program to read, compare, and program them...for the 99/4A.
There is a technique for hunting for chips that hold pins inappropriately high or low if the chips are soldered, it may work to save you some superfluous desoldering. I have used this technique in the past to find the one bad chip on a RAM expansion board which had over 100 memory chips, all soldered, of which one was bad. Essentially you take a known good chip, and you stick it on top of each of the chips you suspect are bad, one at a time, in a piggyback configuration. You have to make sure that all the pins of the chip on top contact the corresponding ones of the chip below. If you're lucky and the defect in the chip isn't pulling a pin too hard high or low, the good chip can fight it and bring back the level in spec. So when you find a position where piggybacking corrects the fault, it's likely that the chip at the bottom is the bad one.
I actually did that same test, pressing a new RAM chip atop the soldered in ones, but it didn't change anything, no matter what that line was always pulled high. Which feels like one of the 4116s failed with an internal short to VCC on that pin that was strong enough to overcome a known good chip sitting on top.
I love hearing that classic 15khz whine. Music to my ears!
Oh, could you hear it in the video? I tried to roll it out as much as possible, but I literally can't hear it even if I hug the CRT!
@@UsagiElectric Yeah I can, but I'm also a professional live audio engineer with excellent hearing so... I actually don't like it being rolled out much; I'm a fan of it because it's part of the experience of running a CRT! That's how you know it's on!
Nice to see the 99/4A again. Back when it was being sold, I briefly worked for a store that sold them and Xerox 820, and I worked with the local TI 99 users group to teach computer ltieratcy using LOGO. Although it's version of LOGO was a little bit funky, it did have access to the hardware sprites, and we did some fun stuff with the sprites. Fun memories.
Wow! There are bodge wires on both the 99/4 and the 99/4A! You'd think they'd get the PC board right on the 4A.
They definitely decreased the number of long jumpers considerably on the 4A, but sometimes, you've just got no space to run that extra line on the PCB itself.
To de-rust easily, get some evapo-rust, just de-grease and then drop the part in and wait an hour or so, its fantastic stuff and re-usable. Once done, just wash off with water and use a brush or something to remove the black film that is left. You will be left with metal that is completely rust free.
Never used it, but can vouch for evaporust and anything based on the same idea. No harsh chemicals, completely safe, clean to the environment. My only question is what's something that can be used on electronics or anything grounded that prevents rust and oxidation from coming back? Obviously there's tons of methods in normal applications, but I've never figured out what's an electronics safe sealant and rust preventer? Would love to restore my 90s Dell Dimension that has rust all on the back.
@@SonicBoone56 conformal coating is the ideal, but a thin application of nail lacquer works just as well :) most people choose clear or green inside systems, because of typical PCB aesthetics, but any colour should work!
Just avoid the ones with glitter (big or small) in! You can’t always tell whether they’re plastic or metal (the latter can cause problems). That shouldn’t be hard if you’re buying it specially, as the cheapest ones don’t usually have extras like glitter. It’s more of a caution against asking for any old bottle from your sister/mother/spouse without checking first :P
Would probably recommend a clear matte one for external metal, as that doesn’t show up except on close inspection. Of course plastic paint of any kind works, and depending on the size a spray applicator may be easier. Nail lacquer is always water resistant in a way other paints sometimes aren’t though.
@@kaitlyn__L thanks for the tips! Didn't know conformal coatings actually protected metal too
Very much recommend Evapo-Rust. Excellent product, does a great job, and best of all non toxic. You can get many uses out of a gallon. I like that it gets out ALL of the rust, even down in pin holes and such. As mentioned, clean with water and slightly abrasive pad afterwards.
Very nice A-7 picture on the wall.
Thanks, my father used to fly A-7s off the Independence with the "Waldos"!
An American with a kettle and a tea cup? It's a Christmas miracle😮
Not to mention having kettle temp shown in good ol ' celcius.
Fran drinks tea, not sure if she uses a kettle tho.
@@TheErador don't watch her channel. If we are talking about the same lady, she is smart, but there is something in het behavior and video making i find very annoying.
@@bzuidgeest franlab. Probably the same, her neurodivergent tics probably put some off
@@TheErador yes that's the one. Neurodivergent... Is that the way we say weird, these days😃. Everybody is different and doesn't need a label. That they annoy me is not her problem, nor mine.
Looks almost as rusty and crusty as the VCF Mac :)
Good to know of this rare transitional model, and nice to see your progress. It's definitely worth saving.
As for paint, I'd go either with transparent polyurethane spray paint, or something automotive-specific.
Not quite, the VCF Mac was rough!
We'll get there on it, and I'm pretty sure I have an idea of what needs troubleshooting next!
I actually watched you ane FPGA troubleshoot the video circuit in voice chat, youe vacuum tube way of understanding it was funny for sure.
I was muted because I was still a bit shy to talk haha.
Great to see that you got it partially working!
my first computer was a TI 99-4A. I had the speech synthesizer, I never have been too much into games,. I would write BASIC programs on it. I loved that computer! my parents had a window Air-con unit in the living room window for the main part of the house, but I had none in my bedroom, I remember in the summer, being so into coding, that I would work on programs until the sweat dripping off my face, would get the keyboard wet.
Ya I was that enthralled in it ! 😀
Great effort so far, I hope you can get this rare system up and running! I've never experienced the 99-4 but the 4/A has a special place in my heart. Little yellow screwdriver... KYB AGX? ❤ the short film!
Thank you so much for the coaster!
The 4/A is a considerably better system, but it is fun to experience the obtuseness of the OG.
And I have no idea where the little yellow screwdriver came from, I've typically been a Bilstein guy, though I do have a set of KYBs on my Bellett!
Nice!
Don't worry about the 15k noise... Those of us old enough to remember the Ti-99/4 and /4a can't hear that range anymore anyway...lol!
The last time I checked my Ti-99/4a to see if it was still working was around 2015 I think. It was working perfectly back then, but now I am afraid to try it because the capacitors are probably all dry now. I have the Speech Synthesis module, the Peripheral Expansion with 32K memory and a Floppy drive in it.
You may be able to use a thermal imaging camera to look for bad chips (not always possible).
You could also socket all of the suspect chips then remove each one in tern. This is highly dependent on the architecture might not apply in this case but your stuck line might change when the bad one is removed...
If you don't like blanket replacements, why not get a retro chip tester or the like? Considering how much TTL hardware you have there, that seems like a valid tool purchase. And you can save as many chips as possible. And save some frustration.
Edit cheap options for testing: dramduino. You likely already have the parts. Can be made on perf board
I actually have a 4116 specific tester, I was more lamenting the fact that I was desoldering them in the first place.
But really, I had to desolder them anyways, even if I could locate the single bad chip, there was no way my OCD would allow me to have just one of the eight RAM chips socketed, haha.
@@UsagiElectric You could live dangerously and just solder the chip back without a socket. 🙂
I've written quite a bit of 9900 code back in the day. A really weird and unusual software architecture. Silicon real estate was expensive back then, so it didn't have any on-chip general registers. Instead, there was a pointer to a block of 16 locations in main memory which functioned as registers. Long before caches became a thing! Didn't even have any real support for stacks: no push or pop, just a 'branch and link' instruction which switched to a new 16 locations;. You had to keep track of the old location and restore it on return from a function call. Interesting times!
the TI-99/4A was the very first computer i encountered in my life. a family friend had one and he was kind enough to let me play with it. It was GREAT! i was so young so i didnt do much but the smell, the sounds, the games.. that was so awesome!
I ran TI-99/4As well into the 1990s. I even surfed an early Internet using a TI-99/4A as a dumb terminal to a dial-up service; I sent my first e-mail on one. In many years of being a TI-99/4A User Group member, I have seen exactly ONE TI-99/4, and it was a classmate's home in 1980 or so. I remember things like the keyboard overlays. Take good care of that beautiful and special machine.
The TI-99/4A V2.2 is pretty rare as well; they say "(C) 1983 TEXAS INSTRUMENTS V2.2" in the title screen and have a lockout to prevent 3rd party cartridges (ahem... "Solid State Software Modules") from running. Another example of TI shooting themselves in the foot. The V2.2 machines are all beige QI machines (not all QI machines are V2.2, though!), but the motherboards are interchangeable across all TI-99/4A variants AFAIK.
Ahhh, if TI had just released full schematics and internal data, extended the entire 16-bit-bus to the expansion port, and maybe made use of the TMS9918A's genlock feature for the burgeoning camcorder and VCR market of the era, the TI-99/4 and TI-99/4A might not have died an undignified death on the shelves of K-Mart.
What could a stock TI-99/4A do with nothing more than the 32kx8 RAM expansion? (Despite the speech, there is no speech synthesizer used!) To avoid composite video quality issues, this is captured from an emulator but it looks and runs the same on real 1981 hardware. I don't mean to spam your channel with an outside video, but check this out:
ua-cam.com/video/ZhSUhE03XFw/v-deo.html
I think replating is best. you could either do a separate side video showing how to plate or if your other projects are more pressing, you could take the shields to a plating company. The big thing would be to polish it to the desired finish first. Any scratches or tooling marks won't get fixed by plating.
That "curious transistor configuration" is the Common-Base configuration, used to amplify voltages in high-frequency applications.
0:34, love that geek-in-the-sunlight squint :)
A suggestion for RAM troubleshooting: if one address line is shorted, but you don't know which RAM chip is the problem, desolder just that one pin on each chip, one by one until the short disappears. Sometimes after desoldering a pin there can be a little hidden bridge of solder left that keeps the pin connected, but it's worth a try - can save a lot of time!
When I work with CRTs, I use a wire that has alligator clips on each end. I clip one end on a flat screwdriver and the other to ground. I then slide the blade of the screwdriver under the rubber "umbilical" wire to the side of the CRT. Sometimes, I hear electrical zaps, sometimes I don't. I make sure I tap the sires under the rubber several times before I call it safe. Just to be safe, I touch the screwdriver to several other parts inside the display's electrical connections... just to make sure everything is grounded out.
Q200 is configured as a common-base amplifier, which has a low input impedance, a current gain of slightly less than 1, and a high voltage gain. The output is hooked to an emitter follower (common collector) to achieve low output impedance (high current gain), since the common base configuration generally has a high output impedance when configured for reasonably high voltage gain (both are determined by the collector resistor). The reason for this arrangement is to avoid the miller effect, which would otherwise severely limit the bandwidth of the amplifier and produce a very mushy video signal.
You tend to see a lot of weird things in video amplifiers, and the reason for them is almost always "needs more bandwidth", since video quality relies on being able to output square-edged signals (or as close as possible to it), which means a flat amplification of everything from DC to daylight.
Ugh.... if it comes down to the 9900 be extra careful desoldering and resoldering... the pinout is very unforgiving and a solder bridge across the wrong 2 adjacent pins will not work out well...
I still have my original 4a from 1983, and it still works!
I bought it a few friends over the years in case i ever need spare parts.... it's not the TI99/4a of Theseus yet, i did re-cap the power supply board and the motherboard as a couple caps were bulging.
Damn it! That's why the 4A always looked weird to me! I had the 4 when I was a kid and didn't realize it, the darn thing was stolen when I was a teenager too. 😭
You can spot the TI-99 enthusiast by the extension he's had to build onto his house so he can fit all his sidecars in.
The intro has hints of Bad Obsession Motorsport 😂
Oh my god. I started writing code for Apple II, but never had a computer myself until years later when I finally had one of my own. And it was a TI -99/4A ! This brings back memories.
Good to see another video on a historically interesting computer. You will figure this one out, I'm sure. I look forward to the next episode!
As for the shielding? I'd just hit it with some lubricant (like 3-in-1 Silicone or CRC 2-26 if you prefer silicone-free) or maybe some wax-based conformal coating. It's not permanent or expensive and is easily removable/renewable.
I'm impressed that there's more than one American that gets how an electric kettle works!
That little kettle is awesome!
At University in 81, the undergraduate computer club found itself in possession of a prototype - wired-wrapped I think - with 9900 processor and some sort of display. May have been a UART/serial terminal. A handwritten monitor rom with a single error message QUE? The 16-bit instruction set looked gorgeous on paper, especially compared to Z80 or 6502. Had a go at writing a simple cross-assembler on my BBC MIcro, but it didn't go anywhere. I discovered theatre, I think. Alas, the 9900 never gained sufficient traction in the market no doubt because of the memory performance issues you mention at the head of the video. I certainly is a fascinating path not taken.
That's a wonky bistable circuit normally used for flashing LEDS!
wow I actually remembered something!
That's the first time I've seen anyone else using those solder-sucker bulbs. I don't have the attachment, mine just have nylon tips but I find them far superior to those one-shot-wonder spring-loaded things. I used to get mine from Tandy/Radio Shack or RS Components for about £5.
Only drawback, it can be a PITA to empty big chunks of solder out. I use a home-made pin ring made from a paperclip to keep the nozzle clear.
Okay I love the way you filtered the CRT whine. I can still hear it but you managed to tone it down to a degree where it's just "comforting". Like "Ah yes it's on and it works like it should." Great! :3
LOL it looked like you were greeting your co-workers after coming back from a vacation! But we all know who your favourite co-workers are. Of course it's the buns!
The composite amplification circuit is a common-base amplifier that feeds a common-collector amplifier. The common-base is intended to have an impedance to match the output of the TMS9918 composite output. The base voltage is (4.7/(4.7+6.3)) X 12 V = 5.1 V which means there's approx. (5.1V-0.7V)/2.2k=2 mA of emitter current, or the input impedance is VT/2 ma = 26 mV/2 mA = 13 ohms. Because this is very small, the input impedance is dominated by the 470 ohms+2.5 k pot and so the voltage gain is 2200 ohms/(470 ohms+2.5k pot). This is in turn goes into the high input impedance common collector stage which will have a gain of 220/(220+47)=0.82. The output impedance of the common collector is approximately (Vt/Ic + 2.2k/beta)+ 68 ohms so probably something like the 75 ohms that the composite video expects.
you could try cold tinning to cover the shields with thin layer of tin metal. you use special liquid for that that does electrolysis without electricyty , just rub it in those liquids work on principle of base metal properties and you could copper coat steel, tin coat steel nickel or copper and nickel copper or steel but not the other way around with those
it could be found in electronic stores or musical stores that specyfi in brass instruments like tubas and trumpets, necause they are often tinned in this way
A tip on the RF shielding thing (I’ve yet to watch the future videos but who knows, could still be useful info for someone!)
If an anti-rust coating of some sort isn’t Feasible for whatever reason, find another material that will corrode in your conditions Faster & more readily than the material you wanna save, and attach the two together - the sacrificial metal will corrode, but the other metal won’t
Practical Engineering has a fascinating video series on the subject of corrosion in infrastructure
The easiest way for rustproofing the rf shield would be some Tremclad! I'd use it on my metal cases. You can even get a clear coat if you want to keep the original color.
Thanks for the video. I had a 4A as a kid, didn't know about the 4.
RetroBytes coaster? Nice!
And that ceramic package is indeed a beauty!
So what I do when I'm working on arcade PCBs with common RAM address lines like you experienced, I clip that address leg on all of them with side cutters. Then I confirm that the address line is working properly. If the RAMs were a problem, it should start toggling. Then I solder a blob of solder on each clipped leg, one at a time, until it goes back to stuck high. That way I don't have to desolder a bunch of chips just to find one bad one.
Yay! I've never owned a TI-99 but I still may yet. Thanks!
I played with one of these things when they were first released. The main thing I noticed about them was they are slow, really bloody slow. Thanks for the vid, it brought back a lot of memories. 🙂
Coat the rf shield with clear coat paint. It won't rust after and will still look like a regular rf shield.
Don't be surprised if you get a call from The 8-Bit Guy, he may want to documentary on the TI 99/4.
Once it's up and working, I'll shoot him a message and see if he's interested in borrowing it for a video!
For the ghosting, check the video board on the monitor. check filters and resistors. Thats what we always did at our shops if we didnt have the schem or microfilm.
I'm glad I'm not the only person who walks into a room and points at various machines while drinking something. Usually it's me trying to think of what to do lol.
At this point, they're all family members and I was just saying "Hi" after being gone for two weeks!
@@UsagiElectric the Centurion has basically become a good friend who's your roommate lol
Always loved the bus bars on Ti-99 PCBs! 😎
The artifacting on the monitor at 19:26 looks very much like a Lorenz attractor. Found that to be pretty interesting.
Cool video thanks for sharing
On my old Taxan monitor a tech fixed the ghosting by adjusting the colour guns on the CRT tube plus adjusting horizontal hold.
Never set a CRT on the screen. There can be trash inside that will then settle on the front and with the grid will get caught. I have seen it a few times and mostly with a few short sharp whacks will dislodge it (also don't hit it too hard as that mesh can move), but I have seen a couple of screens where it was ruined by dust/trash over the screen.
Learned this at the very beginning of the 90's when I was stripping down screens for parts (and sometimes repair) My dept did over a hundred screens a day so you got to see a lot.
Nice kettle, I'm jealous
Nice work on the collab with Veritasium.
Thank you!
Sweet! My uncle worked for TI in the 70s, so we got one of these right when they were released in '79. I didn't realize the plain ol' 99/4 was as rare as that. I have "fond" memories of being half way done with a program and hitting SHIFT-Q instead of SHIFT-2 (double-quote). And if you've used one of these beasts, you know SHIFT-Q resets the machine. Hmm... now I'm wondering if we got a UK version somehow, yours has the @ symbol at SHIFT-2. Hmm. Anyway, I was VERY happy when my uncle came over with a disk controller and a pair of drives so saving early and saving often did not include a cassette.
FWIW... there were a couple different revs of the 99/4. I know there were different revs of the PCB, earlier models had several different lengths of rework wire installed at the factory. (Pretty sure mine had some green wires on the BACK of the mainboard which yours doesn't seem to have.)
I frequently visited the Dallas Users' Group at University of Dallas where several TI marketing people were members. I think they got early factory prototypes to take home since there were a couple different versions: Beige 99/4's with black "Solid State Software" badges, 99/4's without headphone jacks, 99/4A's WITH headphone jacks, etc. I remember seeing a picture of a 99/8 and 99/2 (along with the marketing presentation on both machines.) I was very much looking forward to having a 99/8.
I pleaded w/ my uncle to get an early-release copy of the Editor Assembler so I could write accelerated routines to handle drawing bitmap graphics into character tiles. By the time I figured out how to do that, the 99/4A had come out with it's full-screen bitmap mode.
If you wanted to be SUPER authentic, you could try to find one of the Zenith TV's TI re-badged as their 13" color monitor. ( as seen in the old-computers.net page on the 99/4: oldcomputers.net/ti994.html ) Alas, mine gave up the ghost in the early 90s.
Good luck with your next debugging session!
Very nice video :) I really appreciate that you try to turn on the computer "live" and not with a voice over. It makes me so much more immersed in the video.
Always wanted a /4 (and a /8, but that's a different story)
Once thing is certain, I wish I would have rekindled my TI-99 hobby with your skill set.
Not having it means that I don't fix anything until it breaks for fear of making it worse, severely restricting the fun I could be having. But at least I have your videos.
TI stuff is out of my comfort zone and alas no time to learn the architecture behind its thought processes. I'll stick with the larger business solid state computers and storage, but best wishes to your attempts and the enthusiast community on this one
Keep the 99/4(a)s ALIVE! thank you for preserving these little guys. parsec forever "harder than it looks, isn't it" 😊
You didn't give us too much to go off but given how rusted everything was my first thought was spraying everything that makes contacts with contact cleaner, especially all socketed chips, reseating them in process.
Some corrosion may also have damaged traces on the motherboard - good luck with that - you will need it.
Some of your RAM chips may still be bad - have you tested them?
I vaguely remember a video of someone working on TI99 - Adrian maybe? - and using some diagnostic ROMs - if they do exist your best bet is probably to try them.
You have a working computer handy - this is awesome - try powering them both up and comparing what is different.
I guess the repair manual already walked you through that but just in case:
1) check the power rails with the scope during startup
2) check the RESET signal
3) check all clock signals
4) check ALL address and data lines being 1 or 0, not floating at random voltages
5) slow-scope some line like D0 and A0 on working and on dead machine from reset and compare. Does it attempt to startup?
6) try running the computer without ROM. Compare what address lines are doing on alive and dead computers.
7) do a visual inspection of the board. Do you see any possible corrosion spots?
8) does any chip get much warmer than in should?
We are seeing new motherboards for other vintage computer but nothing for the TI-99/4A. If any computer deserves a redesign to fix the issues that crippled it out of the gate the TI-99/4A is one. It's my understanding the design halved the possible speed.
The TI's performance problems run deep. The sequential 8 bit access to VRAM and GROMs and that essentially all software is not even machine code but GPL means you'll never fix it and still have something that is a TI-99/4A. Yes changing expansion RAM to 16 bit access would help but it'd still be like trying to sprint with one leg tied behind your back.
A redesign would likely introduce software incompatibilities.
And the system was not a big hit. It's not that interesting. There for not a lot of designed for it. Though it's one of the few that has separate video memory.
TI was working on an 8-bit processor for their home computer, but that never came to be, so they kludged in a 16-bit processor they had on hand.
It would be interesting to see someone use the TMS9900 in its full potential.
Congrats on being featured in Veritasium's latest video! Hopefully that'll help the channel grow!
The power "lugs" are feedthrough capacitors to shield the power noise
loved that machine
13:53 Your technical skills are amazing, and clearly far beyond mine. I still have not figured out how to make things work if you put the cover back on _before_ testing it; for me invariably screwing the case back together before testing means that the device won't work. :-)
As far as debugging computers goes, yeah, after doing basic power and clock checks my next step is almost invariably start with the display output and work backwards from there. Because CVBS (full video or just sync) is so easy to recognise and debug even on a cheap handheld 'scope, I've found this to most quickly get me to whatever the source of the "I'm not seeing a display" issue is. On the Tomy Pyuta I happened to be debugging the other day, I had good sync but no luminance at the output, which was the same all the way back to the TMS9918A, so the next step was to check the DRAM interface on that. Turned out a couple of DRAM chips were showing oddly different data line signals than the rest, and also getting hotter than the rest, so it's probably bad DRAM there, meaning the next step is to desolder all the DRAM, put in sockets, and find some new DRAM chips and ideally a friend with a DRAM tester.
That's where I would have started here, too, since actually the '138 you replaced was "deeper" into the system than the the DRAMS, which are directly connected to the TMS9918. If those DRAMs are bad, it doesn't really matter what you do in the rest of the system since bad VRAM will probably mean you never get a display.
Lol I have one of these and never gave it a second thought. Never powered it up either. my grandad used it to control his furnaces to grow gallium arsenide
The TI-99/4 you have has the exact same ceramic TMS9900 package as my TI DS990 Model 1 computer. I think they have a very similar DOM too!
Problem is also the same, mostly. If I turn on my computer, it will (sometimes) attempt to read from the first external floppy drive, then fail and show LOAD ERROR 22. If I power cycle the computer, I either get a very similar error, or nothing at all. If that happens, I cannot get an error message again until I've left the machine for hours. I am thinking it is a reset circuit problem, but only guessing. All 4116 RAM checks out...
Also, the red/orange keys on the keyboard are the same colour as on the DS990 Model 1!
Looking forward to the next video!
Just got a cool surprise watching the latest episode of Veritassium!
Are the aluminum shells swapable?
If the short causes enough current to flow, you can find it with the hp547a current tracer. CuriousMarc has an episode on using it for exactly this kind of fault.
Unfortunately those are very rare and expensive. Without one, an easy and quick solution is to cut the legs on the ICs one by one until the fault goes away. Then just resolver the legs on the good ICs.
@@adriansdigitalbasement That's actually really smart, I'm kind of embarrassed I didn't think of that, haha.
* scratches head * I'm puzzled. I keep on seeing videos about the "Texas Instruments TI99" and it's always a full computer... but...
I distinctly remember using a borrowed calculator which was designated the TI-99 and was made by Texas Instruments.
This TI-99 was a calculator, not a computer. It was pretty much the Rolls Royce of calculators and it had a huge (for the day) memory which could be divided between program memory and data storage (number) memory.
It also had the ability to write out its memory to custom-made magnetic stripe cards. You needed 4 of those to store the whole memory space. Once each "block" had been written, the information could be read back into the same block by reading the stripe.
I remember programming it to play a "blackjack" style game, using the built-in random number generator.
U-505 is actually a Submarine in Chicago.
thermal camera would spot a slightly hotter ram chip which can indicate one of two things. Either it's the only one that is working and thus is warm, or it's a bad chip and heating up. It all depends on how the chip operates when good whether warm == good/bad.
I wasn't aware that they were that rare. That's what our high school started with, moving to the 4A soon afterwards.
After the not so great experience with the chicklet keys on the 4, I was surprised when IBM decided to use basically the same for the PC Jr. And now... a lot of slim laptops. Shudder.
BTW, those are linear power supplies, so no surprise the voltage is fine with no load. They just make lots of heat - which was good for cold days in Monte Vista, Colo.
I need to dig my modified 4A out and see if it still works. Modified because it has static RAM in place of the 4116's, so it runs twice as fast and has 48kB on-board without an expansion card. Or... maybe I could just let you have a go at it.
would be cool to get the Peripheral Expansion system for the 4a with the P-Code system so you can have both type of expansion system for the TI-99 side car and Peripheral Expansion box.
Yep, 15,734 Hz. Anyway, I'd love to aquire an IBM Selectric with ASCII interface. Oh, and ghosting is likely due to your video input cable. Check impedance and and the length is the same as original with any ferrites that might have been present. It's a signal reflection or delay. Just suggesting possible sources for the ghost. I'm 73, so it's been a while since I did analog diagnostics.
14:13 Spectroid FTW. Great spectrum analyzer app.
What a nailbiter - hope you can save this thing
I have two of those but only one is functional. I think the DRAM is bad in the nonfunctional one as it comes up and the screen is stable but the text is unreadable. Could be the VDP. I'll eventually get around to troubleshooting it. I have quite a few of the 4A's also, but they are plentiful.
10 years ago i saw one of these in the box at a flea market. but i didn't buy it because i was only collecting game consoles. but i should have bought it since pre 90s computers are very rare at flea markets.
unbelievable😱😱😱
Amazing !
That video amplifier uses a common-base for the first stage followed by a common-collector (voltage follower). The common-base stage is a voltage amplifier with near unity current gain, while the second stage is near-unity voltage with low output impedance.
Isn't it operating as a transconductance amplifier?
That doesn't seem to be the case. The first stage (unity current gain, voltage gain=rc/re = 1) is just a stabilizer circuit. The second stage (unity voltage gain, current output is load-dependent) produces an output signal that matches the input signal, but with as much current as is required. The two amplifiers work together to make sure that the output signal is a perfect mirror of the input signal, regardless of load. However, a composite signal is supposed to be 1V from white to black plus 0.4V sync, and looking at the sync pulse on the scope, I bet the chip is producing more than that. The variable resistor R201 must be to attenuate the signal to the proper range.
That video circuit is a tough one... I tend to get very confused if the input to a transistor circuit doesn't go into the base.
Retro Chip Tester might be a good idea?????
Right! Having the input not into the base always does my head in.
I actually have a 4116 tester, but I would still have had to desolder the 4116s anyways. And the real reason I swapped entirely to 4164s was so that they all look the same, it would have driven me nuts having just one modified RAM chip and 7 unmodified, haha.
I wonder if anyone built any multiuser systems around the TMS9900? It's actually based on the architecture of the TI 990 minicomputer.
It's not just based on the architecture of the TI-990, it actually is the TI-990 CPU!
Later models of the TI-990 minicomputer used the TMS9900 in place of the TTL logic boards, so the CPU in the little TI-99/4 is incredibly powerful, it's just the rest of the system that hamstrings it.
@@UsagiElectric Litton used a more consolidated version called the TMS9995 in aircraft weather radar processors, taking advantage of the hardware multiply to speed up the task of processing the raw data.
Pity you didn't get it going. The 99/4 and 99/4A had the bones of some really good architecture (like how the device ROMs are switched in and out and how the console can find GROM routines in cartridges) likely borrowed from the 990 minicomputer, but it was also incredibly compromised. They tried to develop a version of the 9900 that drove an 8-bit data bus, but they ran out of time and it was faster to throw in that data bus multiplexor than redesign the thing to be properly 16bit. The 9900 itself was also not a fast processor and TI struggled to even grasp the idea of making a fast home computer. The 99/8 had some real promise, but it also lacked enough forward thinking (it still used a 9918A-derived video chip instead of at least using Yamaha's 9938A!) so it was too little too late. Oh what might have been...
Really nice👍
Woot :) there are a few variations of the 4.. some have a headphone jack, some have a volume control and a headphone jack and some have neither
Also there are different variations for the 4 power supply, the newer models are compatible with the 4a external supply the earlier models (with the volume control) are NOT
13:50 That looks like a little lamp on the tube base. I wonder what these are for? IIRC my Sony TV-990UB sets have a little neon lamp attached to their tube base. Maybe flash-over protection? I'll have to investigate next time I open one up.