Problem with TI! UGH!

[cbmeeks]

So, I installed the F18a in my TI-994/a about a week ago. I've had the computer disassembled on my desk and have tested it out over and over. All was great.

 

Today, I installed the motherboards back in the case and permanently mounted the VGA connector. Everything was working great.

 

So I thought I would let it "burn in" for a while. I left it running for about 1-2 hours and came back and the screen was black.

 

Now, when I turn on the computer all I get is the power LED and the splash screen to the F18a.

 

The audio is a loud continuous tone and nothing else works. I disconnected the speech synth too.

 

The area above the PSU was pretty warm but not too hot to touch. I'm going to let it cool down for a while and try again.

 

Have I lost my precious TI?

 

Oh, one other thing. I didn't put the RF shielding back on because I don't have a way to cut the metal. But I've used it assembled for a few hours but only a few minutes at a time.

 

Any help is appreciated.

 

Thanks.

[Stuart]

Try carefully removing and reseating the socketed chips - probably 3 GROMs and possibly the clock chip. Did the metal shielding have a heatsink block for the clock chip?

[cbmeeks]

I reseated the socketed chips. Still doesn't work.

 

Yes, the metal shielding had a block screwed on. But I'm not using the metal shielding at the moment.

 

Which chip is the clock chip? I have an o'scope but I don't know what to check.

 

Thanks

[Stuart]

It's the larger one to the right of the centre of the processor on here:

 

http://www.mainbyte.com/ti99/computers/motherboard_ti994a.html

 

Might be labelled TIM9904ANL, TIM9904 or 74LS362, depending on when the console was built. You should be able to Google for the datasheet. Has four clock outputs at TTL levels, and the same four clocks at +12V voltage levels. Plus a reset input and output. Scope away ...

 

(If the shielding only had one heatsink block, that would have been for the VDP that you've now replaced with the F18A.)

Edited June 25, 2016 by Stuart

[PeBo]

Is anything connected to the expansion port (speech, PEB, nanoPEB?)

[cbmeeks]

Nope. Nothing connected.

 

I've downloaded the datasheet to the TIM9904ANL that I have. Well, the datasheet I found was for the TIM9904A but I guess that should be good enough?

 

There was only the one heat sink block installed.

 

1) Do I need the keyboard attached to the motherboard to get the normal screen? I'd rather not take it back out if I don't have to. I currently have the main board and the PSU on my table.

 

** UPDATE **

 

I got the following frequencies from the clock chip.

 

O1 (12), O2 (11), O3 ( 8 ), O4 (9) = 3 MHz. However, it was an "inverted" 3 MHz meaning the duty cycle looked more like 20% high, 80% low. I forgot to record the exact duty cycle but I can.

 

XTAL2 (19) = 12 MHz but closer to a sine wave than a square wave. But I think this is normal because my scope is only 100 MHz.

 

XTAL1 (18) = Unreadable. Just a wavy mess with no signal to lock on.

 

OSCIN (17) = Flat 5v. Couldn't detect a frequency.

 

OSCOUT (16) = 12 MHz. (same as above)

 

O2 TTL (15) = 3 MHz duty cycle appears to be more high than low. Pretty much reversed from the O1-O4.

 

O1 TTL (14) = 3 MHz same as O2 TTL.

 

VDD (13) = 12v

 

VCC (20) = 5v

 

Not sure what the other pins are for.

 

Also, I should note the chip feels warm. Not hot but certainly warmer than the others. But I suspect that's because of the many clocks and 12v?

 

Thanks again

Edited June 25, 2016 by cbmeeks

[Stuart]

Clocks sound good (the TTL level clocks are inverted). It's normal for the chip to get a bit warm.

 

On the *processor*, check pin 6 (/RESET) - should be momentarily low when you switch on then high. Also check pin 7 (IAQ) - should be pulsing high at a variable rate as it reads instructions.

 

I'd also be inclined to remove the F18A and put the VDP back in, see if that makes a difference.

 

If that doesn't reveal anything, you might want to consider getting another cheap unit off eBay or maybe another member, unless you want to fault-find the board ...

[cbmeeks]

Update:

 

I put the TMS9918 back in. I connected it via RF and get a black screen with a very loud, continuous tone from the speakers.

 

I have other TI's but I really would hate to gut one of them if I didn't have to. One of my TI's could be my "parts TI" but I'd like to find the real fault if I could and see if I can get a replacement chip.

 

Any ideas now?

 

Thanks.

[cbmeeks]

Update 2:

 

I swapped in another TMS9918ANL and got the same thing. I can't prove the other 9918 is good or not and I bought it from eBay a long time ago.

 

Black screen, continuous tone.

 

:-(

[Omega-TI]

 

Any ideas now?

 

Thanks.

 

At least you know there is nothing wrong with the F18A. Personally I'd just swap motherboards and insert the F18A on a good one, so you can at least be up and operating while you figure out what's wrong with the motherboard in question.

[cbmeeks]

I might do that. I'm sort of a hoarder with my vintage collection so I always cringe when I have to "steal from Peter to pay Paul". But I may not have a choice on this one.

 

Shoot. I might just buy another couple TI's. LOL

 

But, the engineer in me wants to solve this!

 

 

 

UPDATE:

 

Now this is weird. Remember the continuous tone I mentioned? Well, I turn on the console and get a high pitch tone. Turn it off and back on and then I get a low pitch tone. Back and on again and it's high pitch.

 

I did it like 10 times and each time it toggled between high and low tone. I could see a random tone. But a continuous toggle of high/low?

Edited June 25, 2016 by cbmeeks

[Stuart]

The tone indicates that the processor isn't getting far enough in the boot code to silence the sound chip. So it may be a problem with the processor, memory, or a host of other ICs. Did you check the /RESET and IAQ pins as per the earlier message? Also check for -5V on the processor pin 1.

[mizapf]

The sound chip must be initialized on startup. When power is turned on, the sound starts up, and is muted very quickly, then you should hear the beep. When the machine crashes, this muting is not reached, so you should hear a sound from the undefined setting in the sound chip.

 

I do not exactly know the uninitialized behaviour; the sound generator is just a simple square wave generator. It is not impossible that after powerup there is some internal state that is toggled on the next power up. Maybe you could do a test series with growing pauses?

[cbmeeks]

That sounds like a reasonable idea on the sound. Previously, when I turned it on I would get a really loud beep before the normal "welcome to TI" beep. That makes sense. The sound IC comes up in a random state and the CPU reads from ROM the silence commands.

 

I checked pin 1 and it was -5v. I'm not sure how to test the reset. My scope shows it at low when the power is off and when I turn it on, it seems to stay low for a second and then go high. I don't know if it is actually holding low or just immediately jumping to high on power up.

 

I checked the IAQ and it looked like a square wave of about 300KHz (IIRC) but it appeared to jump around a bit. I assume it was executing commands.

 

Any other ideas on testing the CPU better?

 

I actually have a spare CPU I bought on eBay. But I have to admit...I'm not sure I want to de-solder the old one. I'm pretty good at soldering but I always have a dickens of a time de-soldering. Not sure why.

[Stuart]

All the power supplies are good then and the processor is ... processing. You might want to try swapping the bank of 4 socketed chips (3 of which are GROM memory) with one of your spare consoles - might be a GROM that has gone bad and it's easy to check. Other than that, there's a copy of the console schematic at [http://www.mainbyte.com/ti99/man/ti99_tech.pdf] and you'd have to start tracing through the processor/memory circuit. Anything suspect you'd have to desolder to replace.

[iKarith]

Check the voltages coming out of the coffee warmer. I had problems like you describe when my nanoPEB was not connected to power. If your coffee warmer is undervolting, it'd probably cause the problem you're seeing.

[cbmeeks]

UPDATE:

 

I took the motherboard from my donor TI and plugged it into the power supply of the dead TI. And it worked. So I believe the PSU from the original TI is good.

 

I swapped out one of the socketed chips from the good motherboard into the bad one and it didn't work. I didn't do the other three because it was proving to be difficult to pry them out and I didn't want to take a chance on damaging the donor TI *working* motherboard.

 

I started thinking that whatever has caused the original TI to die could possibly cause it to happen again if I start swapping parts out. So I think the original TI motherboard is going to become my donor TI motherboard and I'm just going to swap the boards out.

 

HOWEVER. There is something different about this board. The clock chip is socketed (the original is not) and the clock chip on this one had a smaller heat-sink block for it. The chips look to be the same model number.

I won't be able to use the RF shielding on my F18a mod board because I don't have a cutting tool that can cut the metal shielding.

 

 

** EDIT **

 

I've been running this for 5 minutes and the clock chip is really hot. I don't think I want to run it without cooling.

 

Any recommendations on how to attach a heat-sink when I can't use the RF shielding?

Edited June 26, 2016 by cbmeeks

[Stuart]

Any recommendations on how to attach a heat-sink when I can't use the RF shielding?

 

You can get slide-on heatsinks, such as http://www.digikey.com/product-detail/en/aavid-thermalloy/580200B00000G/HS179-ND/228497.

 

Or ones like this http://www.digikey.com/product-detail/en/aavid-thermalloy/501200B00000G/HS227-ND/265412 that you have to fix on with adhesive thermal tape.

[dphirschler]

Easy to swap out the sound chip. I had to do that once. The TI had similar symptoms.

 

Darryl

[matthew180]

Random thoughts as I read through these posts:

 

The clock chip runs hot. I have seen motherboards that have a heat sink on it just like the original VDP heat sink (only smaller). Not sure if the clock chip heat sink was on earlier or later motherboards. Some have it, some don't.

 

As previously mentioned, the sound chip powers up making noise and must be silenced. If your TI is screaming, then the CPU is not executing very much, if any, start-up code.

 

If you plug in the F18A and see the "green screen", you know several things from this. First, the system is *not* being held in reset (otherwise the F18A output would be blank). You also know that the +5V is most likely ok, or again you would not have any F18A output. Also, seeing the green screen means the CPU did not execute enough start-up code to begin VDP initialization.

 

When you put the shield back on, you need to make sure it is not shorting anything out. There are only 3 screws that hold the shield in place, however the shield is also "supposed" to be soldered in several places around the motherboard. I don't know anyone who resolders the shield, but you need to make sure it is properly aligned (or don't put is back on at all).

 

Those white strips around the motherboard that you might think are just extra shielding (that's what I thought), they are actually power and ground buses! They bend and break very easily, and I can tell you from experience that they are critical to a working system. I used to just casually bend them out of the way until I learned (the hard way) what they they really were. If a leg on one breaks they are very hard to resolder since they are very short, and they tie into power and ground planes which suck the heat out of your soldering iron. As far as I can tell they are also specific to their location, i.e. you can't swap one for another.

 

TI power supplies are known to run hot and fail often. Check your voltages under load.

[cbmeeks]

I like the idea of those slip-on heat sinks. I may go that route. One thing I noticed is that the original motherboard that did NOT have a heat-sink on the clock chip is newer than the motherboard that did have a heat-sink on the clock chip. Same goes for the chip itself. It seems newer too.

 

I wonder if the newer revision runs a little cooler so TI thought a heat-sink wasn't needed?

 

I went out and bought some snips that can cut metal. I was cutting through that RF shield like a hot knife through butter. LOL

 

So I might just go ahead and put it back together and use the RF shield until I can order some of those heat-sinks. I actually don't want to use that shield. Seems way too easy to accidentally short something out. I'm sure the engineers at TI knew what they were doing but the whole TI seems like an odd design sometimes.

 

And yes, I'm familiar with those weird, white voltage rails. I've never seen that before in any other computer. Not like that anyway.

[cbmeeks]

OK, I looked at the slide-on heat-sinks mentioned by @Stuart.

 

Those look great but I think the issue is that they require a clamp that goes under the IC so that it snaps on. I'm not sure that would work on the socketed clock chip.

 

Unless there is another way it can clip on?

 

Is there a thermal glue or epoxy that could transmit the heat and not melt?

 

Thanks.

[Stuart]

>> I wonder if the newer revision runs a little cooler so TI thought a heat-sink wasn't needed? <<

 

Out of interest, there should be a grey crystal next to the clock chip. Does the old one say 48 MHz, and the new one 12 MHz? If so, that indicates different versions of the clock chip. It may be that the old one was less reliable, so it was socketed for easy replacement and had a heatsink, and the new one was more reliable so was soldered direct to the board, which would conduct and dissipate more heat to the board anyway.

 

>> I actually don't want to use that shield. Seems way too easy to accidentally short something out. <<

 

I cut mine, and as well as fixing back with the screws, I soldered (just a blob of solder) the cut ends to the GND trace round the edge of the board just to make sure it couldn't move. A cheap 30W iron is enough to solder the shielding to the board, and easy enough to unsolder if you need to remove the shielding again.

 

>> And yes, I'm familiar with those weird, white voltage rails. <<

 

Be aware that there's actually two metal strips in each, distributing GND plus another voltage. If you're tempted to cut one to isolate something, you're liable to short them together. ;-(

 

<< Those look great but I think the issue is that they require a clamp that goes under the IC so that it snaps on. I'm not sure that would work on the socketed clock chip. >>

 

Should work OK. If you look at a socketed chip, you'll see that they often don't sit right down on the socket anyway. Normally enough room for that clamp to slide under.

 

<< Is there a thermal glue or epoxy that could transmit the heat and not melt? >>

 

Google "adhesive thermal tape" - look for something like Akasa AK-TT12-80. You can probably pick some up locally. Often used to fix small heatsinks to modern RAM chips and the like.

[cbmeeks]

UPDATE

 

The motherboard swap appears to be working. The computer has been running Parsec for 2.5 hours with no issues. Fully assembled.

 

I ended up using the RF shielding and putting a dab of thermal grease on the small heat-sink.

 

Overall, I'm happy that I have a F18a equipped TI. Just too bad that I now have a "parts" TI that doesn't work. Oh well...maybe I will buy a couple more to make up for it. ;-D

 

Thanks for all the help everyone.

[RickyDean]

UPDATE

 

The motherboard swap appears to be working. The computer has been running Parsec for 2.5 hours with no issues. Fully assembled.

 

I ended up using the RF shielding and putting a dab of thermal grease on the small heat-sink.

 

Overall, I'm happy that I have a F18a equipped TI. Just too bad that I now have a "parts" TI that doesn't work. Oh well...maybe I will buy a couple more to make up for it. ;-D

 

Thanks for all the help everyone.

If your handy at unsoldering/soldering you could start with the logic chips around the processor and or the crystal and start socketing/replacing chips. I used to do this to fix my MB's till I packed everything away for a bunch of years. Just starting to get back to fixing like items.