Retired Texas TI99/4A engineer needs help please.
 

Its nearly 20 years since I stopped fixing 99/4As and now I'm stumped with a simple power board repair.

The board presented with LED lit, computer not powering up.
Examination revealed R6 100 ohms burnt O/C. A replacement went the same way, fast.
Component checking revealed Q2 TIP31 O/C, replaced with a tested TIP 31, D10 diode O/C replaced with a 1N4007, perhaps a 1N4148 would be better, C15 Hi, leaky? replaced, D9 and Q1 checked OK.

Now I have + & - 5 v and +12v with no computer connected but with the computer I have a partial boot to a distorted screen and Q2 begins to heat very quickly.
The computer is not faulty.

I'm stuck, forgotten all I knew about this power board, they were always too reliable. Any help gratefully received.

Schematic attached.

Sam.

Re: Retired Texas TI99/4A engineer needs help please.
 

Hi Sam

A quick google shows the PG1992 to be an ultra fast switching diode.

I suspect you're right the IN4007 is not a great sub

Cheers

Mike T

Re: Retired Texas TI99/4A engineer needs help please.
 

Interesting circuit. Is that really a switching buck converter made with a 723 regulator? That had the reputation of being one of those things that was shown in the data sheet but wasn't recommended in real life. L4 is the giveaway. If I haven't got the wrong end of the stick, then D9 and D10 probably both need to be fast diodes. D9 is doing the heavy lifting so D10 could probably be a 1N4148 at least just for testing. A random schottky diode from a scrap power supply would also do.

Chris

Re: Retired Texas TI99/4A engineer needs help please.
 

Switching regulators are not my favourite things.

PG 1992 equivalent is listed as 1N4148 so I'll replace it with one of these.
Is 1N4148 as fast as the original? We are dealing with bits designed in 1980 here.

Maybe it is working OK but the TIP31 gets warm damn quick, it has a clip on heatsink but looks to be not very efficient.
I know that the area on the console where the power supply board is gets warm normally, but...........
How hot can a TIP31 run safely?

Everything stays cold until I connect to the logic board.

Re: Retired Texas TI99/4A engineer needs help please.
 

Yes, a 1N4148 will certainly be fast enough. This circuit is unlikely to be running at more than about 20kHz so 'fast' doesn't mean crazy fast, just not absurdly slow like the 1N4007.

TIP31A should be OK running hot but I wouldn't expect it to run really hot here unless you can see signs that it's run hot for a long time in the past. It will get quite warm because it will switch fairly slowly (by modern MOSFET standards) and so be quite wasteful of power.

See if getting the diode sorted out improves matters. It may do.

Chris

Re: Retired Texas TI99/4A engineer needs help please.
 

I have TIP31B in now. It should make no difference.
Pushed for time so it will have to wait till later, I'm on a deadline for going away for winter, lots of boat maintenance to do in the next month.
In view of the distorted picture when this psu is running the computer I may have a ripple on one or other of the supplies.
If I scope them just how clean can I expect them to be?
I'm trying to understand what instigated the failure, was it the leaky cap or the O/C diode?
I would have thought that even with the power transistor O/C the 100 ohm resistor would not have passed enough current to burn out, they look to be 1/4 watt.
Sam.

Re: Retired Texas TI99/4A engineer needs help please.
 

I think there is a clue in the fact that the PSU is not yet able to run a known good TI99. Something is still wrong with the PSU.

For the time being forget about the computer as a load and try running the PSU up with combinations of high wattage resistors or lamp loads on the 5V and 12V outputs and see if it can deliver ripple free output with the maximum expected load on it.

Testing / faultfinding the PSU with the computer as a load poses a certain amount of risk to the computer if the 5V rail decides not to limit itself to 5V for any reason.

Re: Retired Texas TI99/4A engineer needs help please.
 

I intend to do the testing with resistor loads, I don't have access to a TI99 now, it went back with the guy who owns the psu.
But as to how much load is correct, i have no idea.
Obviously to fry the resistor, it has passed 50mA, 1/4w plus.

Re: Retired Texas TI99/4A engineer needs help please.
 

On a home 'console style' computer of that era I would think it would draw somewhere between 0.5A and 1A from the 5V rail alone. It's a shame you don't still have the computer, you could have tried running it up on a twin bench PSU (if you have one available) and that would have shown you that (a) the computer really was working and (b) how much current the machine really takes from each rail. The computer itself may also have its power requirements from each rail marked on a label, maybe ask the owner.

Ordinarily the majority of the current drawn should pass through the transistor, so when the transistor failed the load tried to draw what it wanted through the resistor alone.

Re: Retired Texas TI99/4A engineer needs help please.
 

I wouldn't worry about the other rails for now. Just concentrate on the +5V one since that's the one with the problem, and its circuit is radically different from the other ones (they're much simpler). Sticking my finger in the air, I'd say the +5V rail needs to be able to deliver about 2A to power the computer happily, judging by my experience of other machines of the era. It might be somewhat less, though.

Chris

Re: Retired Texas TI99/4A engineer needs help please.
 

Actually I'm going to revise my guess (0.5A-1A) because if that was the current requirement the 5V regulator would probably have been a three terminal 7xxx series, just as the others are.

Therefore, the maximum output current required from the 5V rail is likely to be in excess of anything that a 7805 could comfortably supply. The standard versions do 1A maximum. There are tougher 'evolved' versions of the 7805 which can handle more output current but, being linear regulators, they would need a very big heatsink to do so.

Re: Retired Texas TI99/4A engineer needs help please.
 

There's a gentleman here:-

http://www.stuartconner.me.uk/ti/ti.htm

-who writes:-

Quote:

The console motherboard, fitted with the TI Extended BASIC cartridge, draws the following current from the power supply board: +5V: 950mA +12V: 330mA -5V: ??? Didn't measure it, but should be pretty low - probably just a few milliamps On the 'standard' power supply board, the +5V regulator is a switched mode design, and the +12V and -5V regulators are of the linear type. The NanoPEB draws from the +5V supply ~75mA 'at rest', rising to ~85mA when accessing the CF card .

Cheers

Re: Retired Texas TI99/4A engineer needs help please.
 

Ah, there you go, right near the limit of an original 7805, probably too much for one to handle continuously unless mounted on a huge heatsink.

Sam, if you can draw ~ 1A from the output of the 5V rail without it sinking by any significant amount or looking like the teeth of a saw, then you're doing well.

Any ripple that there is ought to be high frequency, up in the high audio / ultrasonic frequency range since this is a switching type regulator. If you see low frequency ripple on the output you may have a problem with the bridge D1-D4 or smoothing (C9) on the input to the 5V regulator, although that would be making the output from the 12V rail rough as well.

Also worth checking that neither L1 / L2 have gone high resistance, and the same goes for both sides / sections of the power switch, one half of which switches the AC input to the positive supplies, the other half of which switches the AC input to the -Ve supply.

Re: Retired Texas TI99/4A engineer needs help please.
 

Wow, lots of useful help there. I was thinking in excess of an amp from the 5v rail, this powers the optional speech synthesizer and the cartridge port as well.
Am I correct in thinking that as I am getting +5v from the output driven by the UA723 that this chip is likely to be OK?

Re: Retired Texas TI99/4A engineer needs help please.
 

Are you saying that with no load on any of the outputs you are getting spot-on 5V from the 5V output? If so then I would assume for now that the uA723 is OK, and the next thing you need to do is apply an ever increasing load on the 5V output, ideally monitoring it with a scope as well as a meter.

If the output collapses or develops bad ripple when you load it up, then you can go on from there.

Re: Retired Texas TI99/4A engineer needs help please.
 

Within a few tens of millivolts, yes I am getting 5v, off load, but I did when the TIP31 was O/C as well. The resistor only burnt out when the rail was loaded.

The logic powers up but there is massive data corruption, like a line hold fail, the screen is unreadable.

I need to take stock, follow up on all the suggestions and scope the output.
I'll be back. Thanks for giving me the pointers.
Sam.

Re: Retired Texas TI99/4A engineer needs help please.
 

Had some time on this board yesterday.

Coupled a 20v transformer and sorted out the PG1992 diode, stole a good one from a scrap box.

Sadly I found that though the 12v rail is OK at 12.2v and the diode bridge is all good, the 5v rail is at 25v !
There is no regulation at all. All I can now presume is that either or both chips are duff as all the discrete components test out OK.
Hence with an O/C TIP 31 it explains why the 100 ohm resistor would fry when any load was put on.
My previous measurement of the 5v rail was obviously incorrect, perhaps I was on the -5v rail.

Or am I missing something?

I know nothing of the UA723C or the TL331 chips other than they are hard to source in dil package and that they seem so expensive that I may well be wasting my time trying to go any further with the repair, which is sad as TI99/4As are now a bit thin on the ground.

Unless I can find some chips at a reasonable price I may have to admit defeat.

Sam.

Re: Retired Texas TI99/4A engineer needs help please.
 

The original 5V regulator based around the 723 is terribly noisy and inefficient by modern standards. The quickest, easiest fix is almost certainly to replace it with one of the ubiquitous LM2596-based buck regulators available for a pound or two on eBay (lots of them seem to have displays attached, but there are some without). Connect the input to the TI's unregulated supply, set the output to 5V and wire it to the 5V pin. Remove Q1 and Q2 and that's the job done.

Chris

Re: Retired Texas TI99/4A engineer needs help please.
 

Hi Sam,

I think I may still have some half broken pieces of TI99 in my workshop which I'm clearing out. I'll try and get a look today or tomorrow. If I find any bits you're welcome to them. IIRC the last time I tried it it worked (about 10 years ago) but the modulator was hacked to death whilst I was a teenager, the PSU was good.

D

Re: Retired Texas TI99/4A engineer needs help please.
 

That thought had occurred Chris, I'd have to ask the owner.

If you find any bits at all Dominic, I'm sure that the guy who owns this board would be very grateful, he has been playing with TI stuff since 1983.

Sam.