boxpressed wrote on 2020-06-11, 01:20:

I think a CT2760 I have is about to give up the ghost. The EMU8K gets too hot to touch for more than a second or two, and there […]
Show full quote

I think a CT2760 I have is about to give up the ghost. The EMU8K gets too hot to touch for more than a second or two, and there are no digital effects, just music.

This is a card that I recently acquired that I thought tested fine, but when going back to it, I discovered this problem. I had the computer on when I plugged in an audio cable, so maybe there was some ESD, but I don't think so. But there was a loud squealing noise that made me check the EMU8K and then turn off the computer.

The card is recognized by UNISOUND when booted up, and it does play music, but after running a while, I start to get memory errors when I try to load a new DOS game. No squealing sound. Then upon reboot, UNISOUND says it cannot detect a Creative non-PNP card. After cooling off a little while, it can be detected again.

I don't have any soldering skills, so I'm guessing that there's not much I can do for this classic card?

If you can find a donor card and somebody to do some surface mount soldering, there may still be hope .

I am blessed to have a cell phone/computer/TV repair shop in a mall close to me that says they will even do BGA rework (not cheap for that, if memory serves). I have only used their services a couple of times for some surface mount soldering and cell phone repairs that I did not dare do myself (or hadn't the patience to) and was extremely satisfied . I really hope they survive the COVID-19 lockdown/closure , they deserve to succeed .

If you live in or near an urban centre, you may be lucky enough to find such a place as well if you look around .

Could there be a shorted capacitor that is causing the EMU8k to be fed too much voltage?

Also may want to double check to make sure that some of the EMU8k legs didn't get bent and are now shorting together.

Caps don't regulate voltage. 😉

If a chip gets hot, it's normally toast. I'd try the card in another PC. If it acts the same there, trash it.

There are three versions of the CT1971 (EMU8000) chip : one without ICxxx marking, IC405A and IC405B. The IC405A version gets very hot (when made by manufacturer named TSMC) while B versions I know of do not. I'm not sure if verison without ICxxx marking (those are made by OKI) is hot or not, but it is the oldest one of the bunch.

derSammler wrote on 2020-06-11, 07:10:

Caps don't regulate voltage. 😉

If a chip gets hot, it's normally toast. I'd try the card in another PC. If it acts the same there, trash it.

Right... but what happens if a cap has a dead short in it? I would think that would just make the system not power on or cause some other issue with the card.

Thanks for all your help everyone. I don't know why I didn't notice this before, but I did brush some dust away from the card after I noticed the problem and saw this. A pretty heavy scratch that cuts across multiple traces and even the legs on one side of the EMU8K. Anything that can be done here?

That does look like some tracks might be severed or almost so. It certainly is fixable with a bit of effort. Scape off some more of the soldermask and with luck solder can bridge the abrasion over and you won't need to add wire patches over the damaged parts. You'll definitely want to resolder some of the pins whole pads are affected as it looks like damage is worst in those places.

I can confirm that the EMU8K in my CT3930 gets also very hot but it works fine, so your problem is indeed that big scratch near the chip.

Tiido wrote on 2020-06-11, 14:50:

That does look like some tracks might be severed or almost so. It certainly is fixable with a bit of effort. Scape off some more of the soldermask and with luck solder can bridge the abrasion over and you won't need to add wire patches over the damaged parts. You'll definitely want to resolder some of the pins whole pads are affected as it looks like damage is worst in those places.

Well, this might be a task for a later day if I ever learn how to solder. More likely I'll try to sell it for parts because it really does seem like this is the problem rather than a fried EMU8K. Just happy to have gotten to the root of the problem so I don't have to worry about my power supply or anything like that.

Maybe I'm missing something, but the scratch looks very very shallow to me. Don't see how it could have severed a trace.

It would be pretty simple to grab a multimeter and find out

The part that goes in the pads of the chip has a few that look like they got severed or perhaps (almost) touching the neighboring pads with the displaced solder.

I'm also have an EMU8k burning up on my CT3930. It's @ 82.6 C ! Looks like at least 2 other people confirm this IC405A gets very hot. @ TheMobRules, how hot does yours get? Hot or not hot, this card isn't detected by Windows 98 at all. It did get detected for a day or 2 right when I 1st plugged it in, but I never installed the drivers to test because I was overcoming other issues like the graphics card not being able to output a high enough resolution that the monitor supports.

I've tried blowing away any debris that could be causing a short circuit. 1 other thing I noticed are 2 green jumper wires at the bottom edge of the card, suggesting this was an early PCB revision. Does anyone else have that?
Am I out of luck?

I haven't measured temps, but I can tell you it gets scorching hot after a short while, I can't keep my finger on the chip for more than an instant and I can feel the heat radiating to the surrounding area. The card does work fine, including the AWE Synthesized Music, but I haven't used it long enough to test long term stability.

Keep in mind that the CT3930 is a rare specimen of SB32 as it sports a Vibra16 chip, Genuine Yamaha OPL and more importantly is non-PnP, which means that it must be manually configured with jumpers and it will not appear as a PnP device in Windows 9x (and may not appear at all unless you tell Windows to scan for legacy cards). I'd suggest you test on pure DOS just to be sure.

By the way, my card also has the green bodge wires near the chip and it's marked IC405A as well.

I measured the temps on 3 cards I have with the IC405A and they all reached 67ºC when read with my IR thermometer. They should certainly be heatsinked, like Voodoo cards with their scorching hot chips...

Keep in mind that the CT3930 is a rare specimen of SB32...

Yes, this is a very precious card ! I got it off of eBay for no more than the much more common CT3600. But unfortunately, I didn't get to test it until after the return period expired because my motherboard (ASUS CUBX-E) ordered from Germany didn't arrive until 1 month later 🙁

OK, so there might be some hope if it can be detected in DOS if the jumpers are changed. I'll try that

Any electrical engineers here? I still think this is an example of permanent latch up. That happens when you apply voltage to the inputs of a digital circuit before you apply power. I've destroyed lots of microcontrollers by making that mistake. The microcontroller then stops working and becomes very hot just like this EMU8k.

Latchup usually results in (partially) non-functional device but these are having no problems that one can observe, all functions seem to work...

I mean I suspect my EMU8k is permanently latched up, not yours.

This high temperature is retarded. In 1995, Moore's law, and more importantly Denard scaling, was alive and well. GPUs and CPUs didn't even have heat sinks. Leakage power was negligible.

I'm trying to think how many multiply adds the chip is doing. I estimate

4 multiply adds / sample assume 4 tap IIR filter, y[t] = a0 x[t] + a1 y[t - k1] + a2 y[t - k2] + a3 y[t - k3]
* 32 instruments assume EAX effects can be applied independently to each MIDI instrument
* 44.1 KHz
* 2 channels
= 11.3 M multiply adds/s

That's not much for hardware at the time. If they did a sliding window FFT, inverse FFT, that would be a lot more expensive.

OMG, it all works now, but only from DOS. Thanks for the tip @TheMobRules. Big payoff, we got Dick Tracy out of the way... I've put a heat sink on. Temperature is much better now (50C).

I'm really relieved the card was OK all along.

Any updates on this? Did you get it working on Windows?