SRAM testing program?

[+intvsteve]

Aside from test cartridges, is there a testing  program that can exercise the Intellivision’s SRAM (RA-3-9600), and give it reasonably accurate good/bad result?
 

I recently repaired a system that had a bad SRAM but the test carts were not reliably detecting or identifying that chip as the culprit.
 

Thanks!

 

[+intvsteve]

A follow-up question, too. Is there a difference between RA-3-9600 and RA-3-9600-1 chips?  The wiki mentions a RA-3-9600A variant. So far I haven’t been able to locate any data sheets or other info on the -1 chip. 

[intvnut]

I wasn't aware of the RA-3-9600-1.  Where did you encounter it?

 

I suspect it's a later mask revision, similar to RA-3-9600A.  From what I recall, there were two main differences between the RA-3-9600 and RA-3-9600A:

• RA-3-9600A didn't have a heat sink.
• RA-3-9600A responds with device-specific garbage on addresses $360 - $3FF, while RA-3-9600 did not respond to those addresses at all.

In any case, I could try to dust off my RAM tester and modify it to push on the RA-3-9600.  To truly be thorough, though, it'll need to run from a flash-cart or other cartridge with 16-bit RAM (e.g. Intellicart, CC3, LTO Flash, Hive, JLP, Bee3, 5-11under's cart w/ RAM option, T-card w/ RAM installed, etc).

 

That'll test the RAM array itself.  Testing the video FIFO will be a little trickier since it's not directly addressable.

[+intvsteve]

30 minutes ago, intvnut said:

I wasn't aware of the RA-3-9600-1.  Where did you encounter it?

Got a batch of RAMs online and noticed a few have that part number. They almost appear to have two date codes on them. Then again, the ones without -1 also are marked with two such numbers.

about half are 8225, the others 8226, all 8237. 

None came with heat sinks.

 

 

[intvnut]

Fun fact: 32040 is the mask revision ID.  For example, you can see that the AY-3-8914 has mask ID 32022 in this picture (middle, bottom).  There was at least one procurement specification on PapaIntellivision that listed specific mask revisions for each of the main ICs.

 

 

[intvnut]

Here, could you give this a try?

 

I just kit-bashed my existing mem tester to test the RA-3-9600.  Since that includes display memory, you'll have to hit an input on the controller to see the status screen.  It'll also display the status screen if it detects a failure.

 

It's not perfect.  But, it may be useful.

mem_test_ra39600.zip

[+intvsteve]

Thanks! Tried it out tonight and did not see any noticeable differences testing the -1 chips vs. the others.

 

The system I have with a known bad SRAM is buried in cold storage so it may be some time before I get a chance to try with a known bad chip.

[intvnut]

1 hour ago, intvsteve said:

The system I have with a known bad SRAM is buried in cold storage so it may be some time before I get a chance to try with a known bad chip.

I'd love to know how that turns out if/when you try it.

[+intvsteve]

On 12/2/2019 at 11:51 PM, intvnut said:

I'd love to know how that turns out if/when you try it.

Here you go!

 

[intvnut]

Found it right away, I see.  Awesome!

 

It's fun to watch, isn't it?

[+intvsteve]

Hypnotic.

[+intvsteve]

I've been letting this run with the bad RAM and it seems to be slowing down. (About 40 minutes in now.)

 

Initially, the errors all seemed to happen in the 'Fill Const' test, with 26E! as the value. I've now seen 'Walking 1s' and 'Rand Value' fail the same way as well. I.e. the failures are not consistent.  I think those other failures also displayed red 26E! but wasn't watching the entire time. While it is hypnotic to watch, it's not quite can't-peel-my-eyes-away TV.

 

It does seem to be getting progressively worse. Just after initially posting this, another 100+ errors happened (while away, natch).

[+intvsteve]

So... about the new heat-sink-less chips.... I'm still tempted to add sinks to them, but have not checked to see if they are used or would fit in an Intellivision II.

[intvnut]

On 12/5/2019 at 12:12 PM, intvsteve said:

I've been letting this run with the bad RAM and it seems to be slowing down. (About 40 minutes in now.)

That's actually by design.  As the RAM test goes on, it increases the number of iterations of each test.  The net effect is that it appears the test sequence slows down.  Really, it's iterating each test in the sequence more times.

 

On 12/5/2019 at 12:18 PM, intvsteve said:

So... about the new heat-sink-less chips.... I'm still tempted to add sinks to them, but have not checked to see if they are used or would fit in an Intellivision II.

I'm not sure that's necessary.  And I'm almost 100% certain they won't fit in an Intellivision II with heat sinks, as the trays for the controllers nearly rest on top of the chips.

 

The chips that were intended for use only with heat-sinks tend to have their pertinent information printed on the underside, and came with heat sinks epoxied on the top. 

[+intvsteve]

8 minutes ago, intvnut said:

The chips that were intended for use only with heat-sinks tend to have their pertinent information printed on the underside...

And there’s the rub. These have the part numbers printed on the underside. Some info on the top. Perhaps heat sinks are optional and going forward, when using these, the application of a heat sink can be decided in situ.  If fixing an Intellivision II, none, else, why not? 
 

based on a recently acquired cheapo IR thermometer, the heat-sink-from-the-fab version stays in the ~82 deg F range (outside the console) which seems pretty cool.

[intvnut]

11 hours ago, intvsteve said:

And there’s the rub. These have the part numbers printed on the underside. Some info on the top.

Iiinteresting.

 

I'm trying to remember whether I've seen Inty IIs with just the mask-number on top.  I think I have.  (That's that 32040 number you mentioned previously.)

 

11 hours ago, intvsteve said:

based on a recently acquired cheapo IR thermometer, the heat-sink-from-the-fab version stays in the ~82 deg F range (outside the console) which seems pretty cool.

Sounds positively chilly compared to this:

[intvnut]

BTW, this is the doc that has all the thermal testing stuff: http://papaintellivision.com/pdfs/CCF10222011_00011.pdf

[intvnut]

Aha... I figured out what's up with RA-3-9600-1.  Basically, it looks like they speed-binned these turds, and the -1 parts aren't fast enough for PAL systems.  See the clock period spec at the top of the page, and Note 2 at the bottom, on page 68 of this PDF:  http://papaintellivision.com/pdfs/CCF10232011_00012.pdf

 

 

And if that wasn't clear enough, on page 74:

 

I'm sure I've read this text before.  I must've forgotten about it.

 

 

[intvnut]

Finally found the mask number table, by the way.  Page 197 of the same PDF.  This confirms 32040 is the mask number for the RAM.  I wish I could find a CPS for the "40-1/4K Exec. ROM", as it might tell us what pin selects Sears vs. Mattel.

[+intvsteve]

AWESOME! Clearly you finished the Evelyn Wood speed reading program in record time!

 

1 hour ago, intvnut said:

BTW, this is the doc that has all the thermal testing stuff: http://papaintellivision.com/pdfs/CCF10222011_00011.pdf

I think in this doc is also where it *sounds* like the IMI Tester may be mentioned on page 6:

CPD  IMI Testers                                  28
GI   IMI Testers (to be replaced by GI)           40
Mattel Test Cartridges (for CPD Consignment)      60
Reserve                                          134
                                                 ---
Total                                            162

It seems that CPD is GTE/Sylvania's 'Consumer Products Division' (though 'Circuit Products Division' is written on the front of this packet IIRC).

 

Now I can tag these chips in the "bin" for 'NTSC REPAIR ONLY!' … not that I'm equipped to repair PAL units anyway.

 

As to the temperatures... Running those test in an actual console with thermocouples mounted is so very, very different! With the heat shield RF shield installed, the way these things were built there was zero active airflow, and likely an infinitesimal convection type airflow possible.  The chips are upside down and completely enclosed. In this case, the chips are upright in the wide open, with zero obstruction -- so hardly fair.

 

In any case, should I end up using these chips in any repairs, they're gonna get a heatsink just to be cautious.

[intvnut]

29 minutes ago, intvsteve said:

AWESOME! Clearly you finished the Evelyn Wood speed reading program in record time!

 

I've never been great at reading linearly.  I do more of a page level non-linear grep.  Maybe that's why I'm better with data sheets and could never really grok fiction narrative.

[HunterZero]

Very interesting stuff here!

 

So the RA-3-9600-1 could work as a drop-in replacement for NTSC systems, but not PAL due to the slightly higher CPU/bus clock? And the RA-3-9600A should work as a drop-in replacement on both?

 

The SRAM is such a common failure I always wondered if a modern RAM chip could be adapted to fit the RA-3-9600 pinout with a small adapter board. It may need some logic for the buffering registers, and address bus logic, but can't imagine it being too hard.

 

The STIC chips (CCIR and NTSC) and CP-1610 CPU would be good candidates for replacement with a FPGA drop-in replacement. This would be an interesting project for someone...

 

- James

[intvnut]

26 minutes ago, HunterZero said:

The SRAM is such a common failure I always wondered if a modern RAM chip could be adapted to fit the RA-3-9600 pinout with a small adapter board. It may need some logic for the buffering registers, and address bus logic, but can't imagine it being too hard.

 

The STIC chips (CCIR and NTSC) and CP-1610 CPU would be good candidates for replacement with a FPGA drop-in replacement. This would be an interesting project for someone...

 

I've thought about doing it.  You could either do FPGA or maybe a microcontroller. 

 

Microcontroller is a bit trickier because you'd likely need to write very carefully threaded code.  It'd be rather close to what it's like to code on the Atari2600, I'd imagine.  ie. Set up a DMA engine to DMA a line buffer to the GPIO at a particular rate for the "pixel out", while the MCU has to balance time between MCU bus cycles and rendering to the line buffer.

 

I'd tackle it, but I've got too many higher priority things ahead of that.

[Eisengrim]

On 12/1/2019 at 5:22 AM, intvnut said:

Here, could you give this a try?

 

I just kit-bashed my existing mem tester to test the RA-3-9600.  Since that includes display memory, you'll have to hit an input on the controller to see the status screen.  It'll also display the status screen if it detects a failure.

 

It's not perfect.  But, it may be useful.

mem_test_ra39600.zip 10.9 kB · 21 downloads

I'm trying to troubleshoot a graphical glitch. I tried running this from my LTO Flash but it just quits to the main menu immediately. Is there another version, or something special I need to do to get it to run?

[YannAros]

I’m into Intellivision only for a bit more than 6 months and I already encountered 2 bad ram chips in about 10 inty that I’ve seen.

It seems to be a common failure with not very obvious effects, the first one was crashing some games especially sword and serpents, the second one had a strange behavior in astromash, some meteors were jumping up from time to time.

 

Having a prog to test this is very nice.

One thing I’d like to know is if there is a source of chips somewhere? As being able to identify the issue is one thing but being able to replace the bad part is another.