Atari 2600 TIA chip source needed for a repair

[WyverntekGameRepairs]

Hi, I have an Atari 2600 that I believe has a faulty TIA (Television Interface Adaptor) chip. The part number is CO10444. It is quite common on these old Atari systems that the TIA chips go bad. Garbled graphics (and sometimes no graphics at all) are present when I try to run the console with my M*A*S*H cartridge. Sound is nill to none except for a clicking when the console is turned on and sometimes a buzzing when the console crashes. I have also traced a problem back to the Fast-Switching signal diodes (1n914's), and those could also be a contributing factor. I am ordering up some replacement diodes from Circuit Specialists, but I can't find any decent priced seller who sells the TIA chips. I found a few on eBay and on a website for Atari parts, but they are quite expensive (around $25 a piece, excluding tax). Is there any reliable, fairly-priced source for these specialist, obsolete parts? Or am I better off buying it off of eBay?

What I did to troubleshoot, for those who are interested:
-I performed a basic continuity check on the TIA chip's pins to confirm that it was not due to internal damage, and the test passed (I referred to the Atari 2600 Service Manual for the complete diagram of the chip's pins). I am aware this test does not apply to all situations, as sometimes an IC can be blown but have pins that are not shorted internally.
-I checked my solder work for any solder bridging or cracks in solder joints, both visually inspecting and using the continuity tester my meter. I referred to the Atari's service manual to make sure that any continuity I found was correct. There were no bad solder points or solder bridges.
-I swapped out the diodes with generic 1n4148 diodes and got different results, thus pointing to the diodes possibly being the true culprit. (However, I still need to prepare for the worst in case the TIA chip is also faulty, because again -- TIA chips are one of the most common failure points in these systems). The console does not work with generic 1n4148 diodes, so the only way I can truly know if it is the diodes is to order some 1n914's up - which I am doing.

Thanks!

[shakalnokturn]

If you're going to fix 2600's on a regular basis and the TIA is a common fault you could consider having a series of replacements made.
The whole 2600 has been emulated on FPGA, I don't know what the minimum FPGA requirement would be for TIA only.

[MagicSmoker]

...
-I swapped out the diodes with generic 1n4148 diodes and got different results, thus pointing to the diodes possibly being the true culprit. (However, I still need to prepare for the worst in case the TIA chip is also faulty, because again -- TIA chips are one of the most common failure points in these systems). The console does not work with generic 1n4148 diodes, so the only way I can truly know if it is the diodes is to order some 1n914's up - which I am doing.

The '914 and '4148 are the same diode, really. Both are gold-doped for high speed switching with 4pF of typical junction capacitance and rated for at least 75V PIV and 100mA If. I can't think of any situation where one couldn't substitute for the other, so if your generic '4148 aren't working then I would suspect you got a bad string of them, especially if you bought them off ebay, Aliexpress, Banggood, etc., rather than Mouser or Digikey.

[WyverntekGameRepairs]

...
-I swapped out the diodes with generic 1n4148 diodes and got different results, thus pointing to the diodes possibly being the true culprit. (However, I still need to prepare for the worst in case the TIA chip is also faulty, because again -- TIA chips are one of the most common failure points in these systems). The console does not work with generic 1n4148 diodes, so the only way I can truly know if it is the diodes is to order some 1n914's up - which I am doing.

The '914 and '4148 are the same diode, really. Both are gold-doped for high speed switching with 4pF of typical junction capacitance and rated for at least 75V PIV and 100mA If. I can't think of any situation where one couldn't substitute for the other, so if your generic '4148 aren't working then I would suspect you got a bad string of them, especially if you bought them off ebay, Aliexpress, Banggood, etc., rather than Mouser or Digikey.

Really? I read up on a forum (can't remember the name of it, I looked it up at school) that said the 1n914 does have slight differences that allow the Atari to function properly, and that replacing the diodes with 1n4148 generics is not recommended. Maybe the person who posted that was a dunce or something who didn't know what they were talking about? Anyway, I'll try a better pair of 1n4148's and get back to you when I get the results. If it still doesn't work, I'll get some 1n914s from Circuit Specialists, as well as a TIA chip.

If you're going to fix 2600's on a regular basis and the TIA is a common fault you could consider having a series of replacements made.
The whole 2600 has been emulated on FPGA, I don't know what the minimum FPGA requirement would be for TIA only.

If I can get a good price for a DIP FPGA (or even a custom wafer, but I imagine those are really expensive), I'll definitely consider it. I know that there are a limited amount of OEM TIA chips in the world, so it would be up to me to ultimately supply myself in the long run. Do you have any suggestions on who I can try for FPGAs? (Keeping in mind, the FPGA chip package needs to be DIP to replace the actual TIA chip)

[shakalnokturn]

I doubt any PLD/PGA of the DIP-only era is going to hold enough logic to fit the job, I could be wrong, I haven't looked in detail what the TIA is about.
I was thinking more of a minimum pin count TQFP/BGA FPGA on a DIP adapter PCB.

[MagicSmoker]

The '914 and '4148 are the same diode, really.
...

Really? I read up on a forum (can't remember the name of it, I looked it up at school) that said the 1n914 does have slight differences that allow the Atari to function properly, and that replacing the diodes with 1n4148 generics is not recommended.

Well, back in 1977 (or whenever it was that the 2600 came out) there might have been some minor differences between the '914 and '4148, but these days it is all but certain the exact same diode will be inside regardless of the part number printed on the package. Furthermore, a '914 or '4148 made today will almost certainly vary in some respects with ones made in 1977, too.

EDIT - I got curious about your emphatic belief that there is something about the '914 that made it critical in the 2600 and found several versions of the schematic so not sure which one applies, but there is one particular use - ironically with a pair of 1N4148, not '914 - that could be sensitive to the specific diode: when used as a varactor substitute in a variable frequency modulator (shown in the schematic excerpt below). Then again, the junction capacitance - or, more specifically, its variation with reverse bias voltage - is unlikely to be significantly different between even vintage '914 and '4148 diodes.

[WyverntekGameRepairs]

I doubt any PLD/PGA of the DIP-only era is going to hold enough logic to fit the job, I could be wrong, I haven't looked in detail what the TIA is about.
I was thinking more of a minimum pin count TQFP/BGA FPGA on a DIP adapter PCB.

If you need additional information on the TIA chip, Wikipedia actually has a pretty in-depth description: https://en.m.wikipedia.org/wiki/Television_Interface_Adaptor
While it is possible to get a BGA on a BGA-DIP adaptor, I imagine it would be expensive (especially because I would need the supplier to mount the BGA chip to the adaptor as I do not currently have the proper tools or confidence to mount them myself). I could use a TQFP package, as that is much easier to manage and I do have the proper tools and confidence to mount it to an adaptor myself. Maybe even a Chip-Carrier package could be used. With a PLCC-DIP adaptor, it could work.

[WyverntekGameRepairs]

Well, back in 1977 (or whenever it was that the 2600 came out) there might have been some minor differences between the '914 and '4148, but these days it is all but certain the exact same diode will be inside regardless of the part number printed on the package. Furthermore, a '914 or '4148 made today will almost certainly vary in some respects with ones made in 1977, too.

EDIT - I got curious about your emphatic belief that there is something about the '914 that made it critical in the 2600 and found several versions of the schematic so not sure which one applies, but there is one particular use - ironically with a pair of 1N4148, not '914 - that could be sensitive to the specific diode: when used as a varactor substitute in a variable frequency modulator (shown in the schematic excerpt below). Then again, the junction capacitance - or, more specifically, its variation with reverse bias voltage - is unlikely to be significantly different between even vintage '914 and '4148 diodes.

Don't worry, I wasn't being emphatic in that sense, I was more displaying a tone of confusion and curiosity. According to the parts list on the official Atari 2600 field service manual, the diodes are '194, so I try and stay as faithful to the original parts as possible to maintain consistency and reduce the need to retune parts of the circuit to compensate for slight differences that an alternative part number would posess. So you can probably imagine me trying to stay faithful to the official documents. But I am willing to take your word for it, you are more experienced in the details than I am.
I have a few NOS '4148s I am willing to try. They're from around 1990, so not too old. Maybe they'll work? Of course, I do have some new '4148s on the way from a better source (I need them for some other repairs, so I just bought an ammo-pack of about 70 or 80). I will be ordering some '914s, just in case. But you are most likely correct in that they are no longer really different, so you can tell me "I-told-you-so" if I come back saying the '914 and '4148 diodes are actually interchangeable.

Also, I found out that I accidentally used 1N4149 (high-conductance fast) diodes instead of 1N4148 (fast-switching signal) ones. Turns out it is easy to accidentally mix those when you have them jumbled in one part container.

[amyk]

or even a custom wafer, but I imagine those are really expensive

Yes, it's in the "if you have to ask, you can't afford it" range. The die layers were reverse-engineered fully several years ago, but I'm not sure if there are any fabs left which can do the old large-size NMOS process it used.

[WyverntekGameRepairs]

Yes, it's in the "if you have to ask, you can't afford it" range. The die layers were reverse-engineered fully several years ago, but I'm not sure if there are any fabs left which can do the old large-size NMOS process it used.

Aye, I imagined as much. Maybe I can consider custom chips from a supplier when I have actually built up and established my business and am generating enough income. But for now, I think I'll rest with these remaining options.

[AE7OO]

I'll have to go digging through the cartons to make sure, but I'm about 95% sure I've got some.  The only problem would be is that they would still be mounted on the boards and I have no way of testing.

Would the same chip be used on a 5200 board?  If so, I know I have some as I'm looking at two of them.

My late father-in-law was an even worse pack rat(my wife calls it hoarding, but I don't agree) than I am.  Even though he passed in 2011, I have still not gotten the office closet cleared, let alone the second hall closet.  As an example of how bad it is, I went looking for a power supply(13.8V) last week and ran across a almost mint condition Bird 43 with three medium power UHF/VHF slugs.  The slugs were wrapped in newspaper and all of that was in a paper bag, that was then placed into a heavy black garbage bag. 
As for all the Atari boards/cartridges/whatever, Andy used to buy broken systems, repair them, and then take them to the local flea market(about a mile from the house) and sell them , back in the 90's.

As I'm not really setup for through hole desoldering(solderpults don't count), about the only way to do it would be to sell the entire board(s).  If this something you might be interested in, shoot me a message.  And if you'll buy two boards, I guarantee that I'll sell each board for less than what some of those places are charging just for the chip...

And if your going to consider do more repairs, let me know.  I'll put together an inventory list and with a fair offer(or something even close  ), I'll box it all up and get it out.  That will get me brownie points with the wife so...


GB - AE7OO

[David Hess]

EDIT - I got curious about your emphatic belief that there is something about the '914 that made it critical in the 2600 and found several versions of the schematic so not sure which one applies, but there is one particular use - ironically with a pair of 1N4148, not '914 - that could be sensitive to the specific diode: when used as a varactor substitute in a variable frequency modulator (shown in the schematic excerpt below). Then again, the junction capacitance - or, more specifically, its variation with reverse bias voltage - is unlikely to be significantly different between even vintage '914 and '4148 diodes.

That would explain it.  Atari must have qualified the diodes based on type and manufacturer and sometimes even that is not enough.

[james_s]

The TIA is a custom chip designed specifically for the Atari 2600, the only place to get one is from another 2600.

The entire A2600 including the 6502 CPU consumes 84% of a EP2C5T144C8 FPGA, this is an old part that is tiny by modern standards so the TIA would probably fit in the smallest FPGA you can get, maybe even a large CPLD. The problem with modern parts is going to be the lack of 5V compatibility so some form of level shifting will be needed.

[Moshly]

You can get all the parts new to build a complete 2600 from ->

http://www.best-electronics-ca.com/