is the card jumpered to enable memory upgade, MFBEN jumper closed and MPU-401 emulation checked in windows control panel?

auron wrote on 2020-08-02, 16:27:

is the card jumpered to enable memory upgade, MFBEN jumper closed and MPU-401 emulation checked in windows control panel?

Three times "yes"

crvs wrote on 2020-08-02, 15:57:

I recently acquired AWE32 CT3980 good in all aspects besides working with ext RAM modules […]
Show full quote

I recently acquired AWE32 CT3980 good in all aspects besides working with ext RAM modules

• "DIAGNOSE.EXE" in pure DOS is not detecting installed DRAM at all
• "AWE Control Panel" in Win98 is reporting installed memory size corectly
  Sound bank loading also works, properly changing "Available" kbytes number
  However, MIDI played with any non-default bank sounds awful, with all instruments distorted

This sounds very much like RAM not working properly, but good enough for the windows driver to recognize the size. You are most likely aware that you need to insert two identical modules into the two slots. The AWE32 does not work with a single memory module. In that case, the issue sounds like broken traces or solder connections between the EMU8k chip and the DRAM sockets. You could try to buzz out that all relevant pins at the memory sockets (parity-related pins are not relevant) are indeed connected to the synthesizer chip.

mkarcher wrote on 2020-08-03, 07:30:

This sounds very much like RAM not working properly, but good enough for the windows driver to recognize the size. You are most likely aware that you need to insert two identical modules into the two slots. The AWE32 does not work with a single memory module. In that case, the issue sounds like broken traces or solder connections between the EMU8k chip and the DRAM sockets. You could try to buzz out that all relevant pins at the memory sockets (parity-related pins are not relevant) are indeed connected to the synthesizer chip.

Yes I know RAM modules should be identical, tried only 2x1, 2x4 and 2x16 mb combinations, with all SIMs good and tested.

Thank you for idea, missing contact looks plausible, could you please give me a hint which connections to check? I'm not that good in the electronics, and didn't find EMU data sheet or AWE schematics over the web 🙁

What are the specs of your system? If I remember correctly, Diagnose.exe has issues with faster computers.

crvs wrote on 2020-08-03, 20:08:

mkarcher wrote on 2020-08-03, 07:30:

In that case, the issue sounds like broken traces or solder connections between the EMU8k chip and the DRAM sockets. You could try to buzz out that all relevant pins at the memory sockets (parity-related pins are not relevant) are indeed connected to the synthesizer chip.

Yes I know RAM modules should be identical, tried only 2x1, 2x4 and 2x16 mb combinations, with all SIMs good and tested.

Thank you for idea, missing contact looks plausible, could you please give me a hint which connections to check? I'm not that good in the electronics, and didn't find EMU data sheet or AWE schematics over the web 🙁

The AWE32 should indeed work with all of your combinations, although I never tested 2x16 because I don't have this kind of modules. As I'm currently on vacation, I can't check what connections to check, but it is very likely that all address and data pins, as well as /WE, /RAS and /CAS are directly connected to the EMU8K chip. As I own a quick-reacting continuity tester, the way I would test my card is putting one probe at a SIMM pin (you can omit the power pins 1, 9, 22, 30, as well as the parity-related pins 26,28,29) and gently brushing over the EMU8K pins with the other probe. If the connection is working, you should hear a short beep.

Typically, connections are grouped (like all data pins next to each other), possibly interrupted by ground and power pins, so if one data pin is missing, the interrupted trace should be obvious just from the neighbouring connection - and even if not, just knowing which RAM pin does not connect to any AWE32 pin is a great help because it points to a specific trace that you can inspect more closely.

As the AWE32 has 16-bit memory access, the address and control lines of both SIMM sockets should be paralleled (a very simple thing you can quick-check for broken traces between the sockets), whereas the data lines should be separate for a total of 16 individual data lines. If you happen to have an ESR meter (to check capacitors), you could also check ESR between +5V and GND at the SIMM sockets. High values (>2 Ohm) point to bad decoupling caps near the socket. As I don't have my SB32 at hand, I can't give you a reference value from my card.

Don't disregard the software hint, though. While I never had issues with my SB32 in a Pentium 133 machine, it is possible that DIAGNOSE indeed fails on faster machines.

cyclone3d wrote on 2020-08-03, 20:29:

What are the specs of your system? If I remember correctly, Diagnose.exe has issues with faster computers.

K6-III+/450 + ASUS TX97-LE + 384 Mb RAM

Yesterday I have temporarily moved the card into 486 (DX4-120), and tested it with both Turbo On / Off = DRAM had been also not detected in both cases.

BTW, for some reason DIAGNOSE.EXE has been less stable with slower machine, with 50% chance of freezing in the middle of diagnostic, and "AWE Synthesized Music" didn't work properly (producing noise). I observe no these issues after returning AWE32 into faster PC.

Have you tried cleaning the edge connector on the awe32 yet?

mkarcher wrote on 2020-08-04, 07:15:

The AWE32 should indeed work with all of your combinations, although I never tested 2x16 because I don't have this kind of modul […]
Show full quote

The AWE32 should indeed work with all of your combinations, although I never tested 2x16 because I don't have this kind of modules. As I'm currently on vacation, I can't check what connections to check, but it is very likely that all address and data pins, as well as /WE, /RAS and /CAS are directly connected to the EMU8K chip. As I own a quick-reacting continuity tester, the way I would test my card is putting one probe at a SIMM pin (you can omit the power pins 1, 9, 22, 30, as well as the parity-related pins 26,28,29) and gently brushing over the EMU8K pins with the other probe. If the connection is working, you should hear a short beep.

Typically, connections are grouped (like all data pins next to each other), possibly interrupted by ground and power pins, so if one data pin is missing, the interrupted trace should be obvious just from the neighbouring connection - and even if not, just knowing which RAM pin does not connect to any AWE32 pin is a great help because it points to a specific trace that you can inspect more closely.

As the AWE32 has 16-bit memory access, the address and control lines of both SIMM sockets should be paralleled (a very simple thing you can quick-check for broken traces between the sockets), whereas the data lines should be separate for a total of 16 individual data lines. If you happen to have an ESR meter (to check capacitors), you could also check ESR between +5V and GND at the SIMM sockets. High values (>2 Ohm) point to bad decoupling caps near the socket. As I don't have my SB32 at hand, I can't give you a reference value from my card.

Don't disregard the software hint, though. While I never had issues with my SB32 in a Pentium 133 machine, it is possible that DIAGNOSE indeed fails on faster machines.

No luck with slower PC (see my previous message). I will try to inspect the connections on the weekend, will post again in case any findings.

cyclone3d wrote on 2020-08-05, 17:15:

Have you tried cleaning the edge connector on the awe32 yet?

Yes of course.

Sounds like I need to pull out my AWE32 cards and do some testing. On the DX4-120 setup, what speed is the ISA bus running at?

Would you be able to post some hi-resolution pictures so we can take a look? Probably best to not attach them to the forum since it absolutely destroys image quality since the forum software update.... Used to just limit the file sizes I uploaded and it worked just fine... Now if you post pics it compresses them and kills the quality.. especially bad if you are trying to look at small stuff.

No.. I'm not bitter at all.

mkarcher wrote on 2020-08-04, 07:15:

The AWE32 should indeed work with all of your combinations, although I never tested 2x16 because I don't have this kind of modul […]
Show full quote

The AWE32 should indeed work with all of your combinations, although I never tested 2x16 because I don't have this kind of modules. As I'm currently on vacation, I can't check what connections to check, but it is very likely that all address and data pins, as well as /WE, /RAS and /CAS are directly connected to the EMU8K chip. As I own a quick-reacting continuity tester, the way I would test my card is putting one probe at a SIMM pin (you can omit the power pins 1, 9, 22, 30, as well as the parity-related pins 26,28,29) and gently brushing over the EMU8K pins with the other probe. If the connection is working, you should hear a short beep.

Typically, connections are grouped (like all data pins next to each other), possibly interrupted by ground and power pins, so if one data pin is missing, the interrupted trace should be obvious just from the neighbouring connection - and even if not, just knowing which RAM pin does not connect to any AWE32 pin is a great help because it points to a specific trace that you can inspect more closely.

As the AWE32 has 16-bit memory access, the address and control lines of both SIMM sockets should be paralleled (a very simple thing you can quick-check for broken traces between the sockets), whereas the data lines should be separate for a total of 16 individual data lines. If you happen to have an ESR meter (to check capacitors), you could also check ESR between +5V and GND at the SIMM sockets. High values (>2 Ohm) point to bad decoupling caps near the socket. As I don't have my SB32 at hand, I can't give you a reference value from my card.

Don't disregard the software hint, though. While I never had issues with my SB32 in a Pentium 133 machine, it is possible that DIAGNOSE indeed fails on faster machines.

Finally I got time to tackle with my faulty CT3980, after a long break. On the good side, during this delay I was lucky to obtain exactly the same sound card, which is fully working, so I could make a comparison.

First and foremost, I could confirm that the issue is definitely with the card and isn't driven by the PC speed - another CT3980 is successfully passing all tests on the same PC and has no sound issues in both DOS and Windows 98. Then I checked SIMM sockets wiring with the probe, as you recommended. Data pins are really unique and connected to EMU, while all address/signal lines are indeed parallel, but they're not going to EMU8K (A0-A9 and CAS can be traced to HY514260B, with other lines connected elswhere). Nonetheless, it is all the same as on the working card, if I didn't do any mistakes while probing.

Then I noticed another difference in behavior between cards: if I remove SIMMs and switch JP2 to use card's internal RAM, DIAGNOSE.EXE doesn't detect onboard 512k on the faulty one. Working card is again passing this test properly. May it give any additional clues? I would still be happy to find a fix, if it's possible.

cyclone3d wrote on 2020-08-05, 17:54:

Sounds like I need to pull out my AWE32 cards and do some testing. On the DX4-120 setup, what speed is the ISA bus running at?

Would you be able to post some hi-resolution pictures so we can take a look? Probably best to not attach them to the forum since it absolutely destroys image quality since the forum software update.... Used to just limit the file sizes I uploaded and it worked just fine... Now if you post pics it compresses them and kills the quality.. especially bad if you are trying to look at small stuff.

No.. I'm not bitter at all.

In addition to findings from my previous post, I also did additional tests in 486 - now with my both CT3980's, faulty and good. Eventually I was able to identify and address another issue with this PC (bad cache), afterwards the results became the same as in other computer: with the good card - all tests pass successfully, with the faulty card - both external and internal RAM is not detected in DOS.

PS: answering your questions anyway - bus speed is 40 MHz, after visual inspection of the cards (with magnification) I couldn't find any obvious defects or differences with my working card, what pictures you'd like to see?