DAC/ADC problem (E-MU SP12)

[living sounds]

The calibration of this old drum sampler is giving me a headache. It uses a 12 bit AD7541 DAC multiplexed into 8 channels. The DAC output is somehow fed to the ADC section when sampling as well, but I don't really understand it.

Anyway, I'm supposed to set certain maximum DC values via trimmers at several points in a defined sequence. The trimmer RT3 (labeled "1" in red here) can be set easily. It sometimes seems to drift without warning though, out of the spec, but not excessively. The next one is the problematic one, RT4 (labeled "2" in red). The DC value is supposed to be lower than 1mV, but it jumps all over the map when dialing in the value, skips, drifts etc., as much  +/- 300mV. I've exchanged it for a high quality cermet 25 turns trimmer, which didn't help one bit. It doesn't change value linearly, but somehow sometimes seems to allow for incrimental changes while turning but then suddenly changes value drastically. Setting it near the zero point is harder than dialing a given value at va. 100mV or so. Even after setting a value it drifts, or rather skips, massively. This also affects the ADC input, which is tied to the DAC  -> Dacout.V is connected to an LM311 comparators positive input, whose negative  input is fed by a S&H chip fed by the ADC analog input.

My question is: Where to look for the problem? The trimmer is obviously all right, the PSU seems to be stable for all analog and digital voltags as well (new caps there, too). DAC chip? Op amp?


Thanks!

 

[living sounds]

I've put in a socket and replaced the op amp after the DAC. Tried different types, including modern low-offset versions. No difference. Any hint about why offset would fluctuate so heavily? I've resoldered a lot of connections to be sure, but didn't find any faults so far.

 

[dolo72]

Hi
      Sounds very similar to problems I was having with my SP12. Do you have the diagnostics prom ? - there is an option to give a value to calibrate the DAC and it states the value it should be. I was having the same problems - I replaced the trimmer ( can't remember which rt ) I also added a resistor to load the trimmer for more control in the correct range. Similar to your experience the calibration constantly changed everytime I screwed the case together. I had also previously got the machine repaired by a reputable synth repairer in London, to no avail. I ended up sending it to Forat in LA and he said some of the logic control was corrupted which I understand was a b***h to diagnose. He fixed that with some of his custom chips also changed the battery and he cleaned and lubed all of the switches. Forat is the authority on the sp and he has de/reconstructed some of the logic control software I would recommend you send It to him. I have had no problems since I got it back and that was around 5 years ago with constant use. With regards to mod's I would not change the audio op amps because the character is in the design. The only thing I would do is upgrade the PSU - better regulators new transformer etc( mine is getting noisey and hot) - or I have been considering getting a couple of a accopians in an external box. Srry no diy solutions but the sp is a master of custom logic controlled old school ( non PCM ) sampling and Forat is the only person I would trust to fix it.

 

[living sounds]

Thanks!

Yes, I've got the calibration rom, and setting the DAC via the respective trimmer (RT1) actually works. It's the DC drift at the other one that messes up this setting again.
I don't really see how the control logic itself would cause this kind of problem. It must have worked before, so a component is not working as it was/should. I don't see any "custom" chips, there's custom programming of course, but corrupted data and DC drift don't correlate well I think.
There's a "level DAC" used for volume control between the DAC and the demuxer that is connected to control logic, other than that I don't see an obvious connection.

Have you looked inside your unit by any chance to see which chips (or other components) got replaced? Shipping this whole thing to the US and back is a gamble, and a pricy one at that, I'd rather avoid it, of course...

What symptoms did you get? I think the sound looses definition when the DC isn't right, and the ADC threshold automatically triggers at the lowest setting without a sound present. It's still useable, but I'd like consistent sound, of course.

I replaced the LF356 with an OPA827 and the 5534 with LME49710, both high end chips with excellent DC specs - which got me a little more clarity but otherwise it still sounds as beefy and dark as before. There are still plenty of TL084 in the signal path, of course.

 

[living sounds]

Hm, just noticed, there's also low level slightly distorted crosstalk between the channels when one channel is at maximum and the rest are at zero. Does yours do that as well?

 

[tubologic]

Did you checked with a scope  if the upper DAC reference voltage Vref. is perfectly stable and doesn't fluctuate ?
I would also check with a fast scope or logic analyzer if the 3 least significant data bits  (LSB) are at a well-defined stable logic level during adjustment. If they aren't, a control logic problem can not be excluded. If everything above is OK the DAC is the next suspect and I would try a replacement.   

 

[living sounds]

Did you checked with a scope  if the upper DAC reference voltage Vref. is perfectly stable and doesn't fluctuate ?
I would also check with a fast scope or logic analyzer if the 3 least significant data bits  (LSB) are at a well-defined stable logic level during adjustment. If they aren't, a control logic problem can not be excluded. If everything above is OK the DAC is the next suspect and I would try a replacement. 

Thanks! No, haven't checked this at all (you mean Vref on the left side of the AD7541, via the TL084, right?).
I haven't got a fast scope unfortunately (mine is of the museum variety, and doesn't go into Mhz territory)...
There's a test mode for the DAC that cycles through the bits, I had checked it with my sound card and didn't see anything obviously suspicious.

Right now the SP12 seems to be working properly, but after prolonged warming up the ADC threshold may have too much DC and auto-sample even without an input present.

The funny thing is that it is so hard to set the offset at RT4, even with a 25-turn pot it doesn't scale linearly and jumps around a lot of the time. Also, after going into sample mode or turning the machine off and on again the value often was way off.

If more information is needed I could post more screenshots, but I'm not sure if I'm allowed to upload the service manual.

 

[tubologic]

Yes, I mean the reference voltage at pin 17 of the 7541 DAC.  This voltage must be perfectly clean and stable up to the lowest mV range. I don't have the details about EMU's recommended calibration procedure using the diagnostic prom but usually DAC output offset must be adjusted with all data bits at zero logic state. (O V)
Provided that the reference voltage and digital input data's are OK there are not much parts remaining: the DAC chip or the I/V output converter. I would first replace the (C83) 33 pF compensation cap around the output opamp, the slightest leakage in this cap will upset the I/V output converter. Don't try to measure this cap, just replace it. If this doesn't cure the problem replace the DAC. (May be hard to find as it is an obsolete part no longer manufactured by Intersil).
I don't believe your SP12 need a complete rebuild and/or software upgrade:  if it worked well before it must be a minor single component failure. 

 

[living sounds]

I thought about replacing the cap, but didn't have a high quality replacement (NPO/COG) handy. I guess temperature stability is the whole point here. Will order some.

I can get a replacement for both the AD7541 and 7524, but the specs of the AD part are actually worse than the Intersil one (even though it's supposed to be an improved version).

Thanks again!

Yes, I mean the reference voltage at pin 17 of the 7541 DAC.  This voltage must be perfectly clean and stable up to the lowest mV range. I don't have the details about EMU's recommended calibration procedure using the diagnostic prom but usually DAC output offset must be adjusted with all data bits at zero logic state. (O V)
Provided that the reference voltage and digital input data's are OK there are not much parts remaining: the DAC chip or the I/V output converter. I would first replace the (C83) 33 pF compensation cap around the output opamp, the slightest leakage in this cap will upset the I/V output converter. Don't try to measure this cap, just replace it. If this doesn't cure the problem replace the DAC. (May be hard to find as it is an obsolete part no longer manufactured by Intersil).
I don't believe your SP12 need a complete rebuild and/or software upgrade:  if it worked well before it must be a minor single component failure.

 

[dolo72]

Srry for the delay - the Problem in terms of sound were I was always getting clicks after the sample which would get worse when I truncated (dc?). The definition was also not great, not quite the big sp sound and the signal to noise ratio was worse - sounding a bit brittle. In use I would always have to loop the end of the sound and then would be tied into the decay settings as opposed to having the choice between tune and decay. The main chip Forat replaced was (facing the front of the pcb) to the left of the sound rom/ram chips, I will try and get a picture later today.

 

[living sounds]

Srry for the delay - the Problem in terms of sound were I was always getting clicks after the sample which would get worse when I truncated (dc?). The definition was also not great, not quite the big sp sound and the signal to noise ratio was worse - sounding a bit brittle. In use I would always have to loop the end of the sound and then would be tied into the decay settings as opposed to having the choice between tune and decay. The main chip Forat replaced was (facing the front of the pcb) to the left of the sound rom/ram chips, I will try and get a picture later today.

Sounds great about a picture, thanks!

Mine doesn't have much of a problam with clicks (though you sometimes can hear a sample ending, so there is/was DC). But the definition certainly isn't what I want it to be.

 

[living sounds]

So, I replaced the ceramic cap and measured the DAC reference voltage. The latter value (which is derrived from the +15V line) drifts up a little with the unit warming up, from 9.96V to 10.01V, but didn't show erratic jumps like the DAC output voltage at all.

Another thing I noticed was that, even when it was fairly well set, playing sounds over different outputs produced different offsets (quite stable after the sound has played), for example the value was 0.6mV, playing a sound on Out4 would result in the value going to 3.6mV, Out7 would have it go to 1.7mV, Out1 back to 0.6mV (these numbers are made up now but that's what generally happened). Also going into sample more and off again, turning the unit off and on again, closing the lid and letting it run another 10 minutes to warm up more would also result in the values sometimes drastically changing. There also seems to be a change of a fewmF due to mechanical pressure moving the trimmer, though I've carefully cleared the bord of residue.

There seem to be some custom programmed logic array chips with E-MU Sp12 related lables printed directly on the chip. If these are the problem I hope Forat sells them, I send them an email now.

 

[tdbdm]

Hi, I have the same issue, do you remember which IC were involved? Thanks a lot