Is there a really good RAM tester for 64K/128K?

[Jmk]

Hi, I have a 130XE that now has seemingly endless RAM problems.

 

It is the early type with 16*8K chips in it. I ended up desoldering the 16 chips, putting in sockets and then having to replace all 16 chips with new ones. I broke one of the capacitors, so replaced it with a 104 ceramic disk type I found lying around.

 

Now... things just aren't running quite right. Stuff that needs 64K RAM is a bit unreliable and the built-in memory test is only covering the first 48K... I don't have one of those SALT cartridges that I have read about.

 

Additionally, the memory test does report the odd error in the 6th square after the machine has warmed up for a while. Arghh!

 

If anyone can help with the following question, I'd be grateful.

 

1. Do the squares in the built-in mem test actually an invidual K within the machine (e.g. does the 6th square actually cover addresses 0x1400-0x17FF)?

2. Do each of the chips cover 8K of the RAM space, or is it set something like 1-bit per address per chip?

3. Therefore, can I figure out which chips are bad using the built-in mem test?

4. Was my choice of capacitor OK?

5. Could I have done something else, like bad soldering that could be to blame for the results I'm seeing?

6. Is there a really, really good piece of software I can use (with an SIO2PC cable) that can identify all the problems and make me a cup of mocha at the same time?

 

Thanks!

[Rybags]

1. Each square represents 1K.

 

2. 8 chips contribute a bit each to make the byte for every RAM access. The outer ones for base 64K, inner for the extended 64K. Earlier OS versions don't do anything about testing the extended 64K.

The 14K under the OS doesn't get tested. The 2K under the I/O registers can't physically be accessed, so isn't tested. Unsure here, but the 2K at $5000-$57FF probably doesn't get tested either.

 

3. A failed RAM test with the older OS will indicate bad chip/s on the outer column. With newer OS that represents the extra 64K with 4 larger boxes (about half the size of the ones representing ROM) they represent the inner column of RAM.

 

4. Not sure.

 

5. Dodgy solder job or something else along the line could also cause an intermittent RAM problem.

 

6. There's a few other RAM testers around, not sure how deep they go sofar as revealing problems. I was looking to write one myself, but that might be quite some time away as I've got lots of other stuff I also want to do.

 

 

Sixth square would represent the sixth kilobyte in RAM, so addresses $1400-$17FF. That's a pretty critical area, practically any game or DOS situation would use that RAM and be very sensitive if something changed there.

 

It might be possible to work out which chip/s are bad. Maybe something like:

 

10 POKE 106,48:GRAPHICS 8:POKE 710,0:POKE 709,14

20 GOTO 20

 

Start the machine without DOS, try that. The graphics should map over that part of RAM that fails, if you leave it a while maybe some random dots will appear.

 

If not, try filling the screen so the RAM contains $FF data:

 

15 COLOR 1:FOR X=0 TO 319:PLOT X,0:DRAWTO X,191:NEXT X

[Rybags]

http://www.atariage.com/forums/topic/162718-comprehensive-ram-test-programs/page__p__2007533#entry2007533

 

Here we go... that's a similar thread I started with the same question. Some RAM tester utilities were posted in it.

[+bob1200xl]

You broke a capacitor? (how? which one?) Any other badness? Do you know that the chips you installed are good? Can you test them in another 130XE? (800XL, 1200XL?)

 

If it is a bypass cap (each chip has one) it matters little. You could even leave it out if you like. (just one...)

 

3. I have never seen any value in the RAM test as far as determining which chip is bad. The test displays address blocks and the chips are bit-wise.

 

4. Not to worry.

 

5. Bad soldering would normally cause a solid failure. You could try flexing the board while it's running - just a little. Check the temperature of each chip with your finger. Warmer chips are often bad.

 

Bob

 

 

 

Hi, I have a 130XE that now has seemingly endless RAM problems.

 

It is the early type with 16*8K chips in it. I ended up desoldering the 16 chips, putting in sockets and then having to replace all 16 chips with new ones. I broke one of the capacitors, so replaced it with a 104 ceramic disk type I found lying around.

 

Now... things just aren't running quite right. Stuff that needs 64K RAM is a bit unreliable and the built-in memory test is only covering the first 48K... I don't have one of those SALT cartridges that I have read about.

 

Additionally, the memory test does report the odd error in the 6th square after the machine has warmed up for a while. Arghh!

 

If anyone can help with the following question, I'd be grateful.

 

1. Do the squares in the built-in mem test actually an invidual K within the machine (e.g. does the 6th square actually cover addresses 0x1400-0x17FF)?

2. Do each of the chips cover 8K of the RAM space, or is it set something like 1-bit per address per chip?

3. Therefore, can I figure out which chips are bad using the built-in mem test?

4. Was my choice of capacitor OK?

5. Could I have done something else, like bad soldering that could be to blame for the results I'm seeing?

6. Is there a really, really good piece of software I can use (with an SIO2PC cable) that can identify all the problems and make me a cup of mocha at the same time?

 

Thanks!

[Jmk]

OK, everyone, thanks for all the replies.

 

I don't have any other Ataris to test the chips with, I broke (and fixed) the bypass capacitor connected to U33 (which I now think is the bit 7s of the upper 64K). I don't think anything else is broken, although my problems initially came via the cartridge slot, so who knows. I assume the new RAM chips are fine, as I've just bought them (although that's not 100% guarantee).

 

Regarding the old thread, I had looked at that but everything had been about testing the number of banks above 64K, not the integrity of the bottom 64K/128K. But... I had another look and there was one thing there that did what I wanted: SUPERSAL.COM (for the lower 64K, at least).

 

It told me that bit 6 was suspect in the range $6000-$7FFF, so I swapped U15 with U32 and it reported the lower 64K as all working. I swapped the upper set of chips with the lower set of chips and tried again - all OK... This was a bit surprising, because it naturally contained the chip that was causing the problem before... however, it seems to be holding for now.

 

Fingers crossed that it stays that way!

 

Thanks for all the help.

[+orpheuswaking]

OK, everyone, thanks for all the replies.

 

I don't have any other Ataris to test the chips with, I broke (and fixed) the bypass capacitor connected to U33 (which I now think is the bit 7s of the upper 64K). I don't think anything else is broken, although my problems initially came via the cartridge slot, so who knows. I assume the new RAM chips are fine, as I've just bought them (although that's not 100% guarantee).

 

Regarding the old thread, I had looked at that but everything had been about testing the number of banks above 64K, not the integrity of the bottom 64K/128K. But... I had another look and there was one thing there that did what I wanted: SUPERSAL.COM (for the lower 64K, at least).

 

It told me that bit 6 was suspect in the range $6000-$7FFF, so I swapped U15 with U32 and it reported the lower 64K as all working. I swapped the upper set of chips with the lower set of chips and tried again - all OK... This was a bit surprising, because it naturally contained the chip that was causing the problem before... however, it seems to be holding for now.

 

Fingers crossed that it stays that way!

 

Thanks for all the help.

 

Possibly dirty contacts?

[MEtalGuy66]

HEres a question.

 

You replaced the memory..

 

Which memory chips did you use? 64kx1 or 256kx1?

 

If you used 256kx1, you need to ground pin 1 of all the chips in the main 64k bank (the one to the extreme left of the machine.) On the extended bank, you should connect all the pin 1's together and either hook them to your expansion upgrade circuit, or if you arent expanding them memory, ground them like you did on the main bank.

 

If you dont do this, you are leaving the high adress line floating, and the slightest bit of static electricity can make it sporadically "flutter back & forth" (which effectively switches between the upper & lower half of the chip).. Obviosuly, this is not good..

[Jmk]

HEres a question.

You replaced the memory..

Which memory chips did you use? 64kx1 or 256kx1?

 

I used 64kx1, a mixture of KM4164B-15 and HM4864-3.

 

Things seem to be holding up, perhaps the chip wasn't seated properly?

[erazmus]

I know this is an old thread, but I just spent the evening diagnosing and repairing RAM problems on a new 130XE I just acquired. One technique I didn't see mentioned, but one that works very well for me, is to put a known-good RAM chip piggy-back on top of the suspect chip. Hold the chip in place (make sure each pin is only touching the corresponding one on the chip below) and power on the machine and run your memory test. If the memory test passes (or shows less bad blocks), then this chip is probably bad, and you can remove it and insert a socket and a new chip. I successfully identified one bad chip in my base 64k, and two bad chips in the upper 64k using this method tonight.