AD converter oscillating?

[living sounds]

I'm trying to fix an old Black Lion Sparrow converter. One of the two channels is noiser than the other and both suffer from oscillation, sweeping tones visible in the DAW via an analyzer above the noise. I cannot see the oscillation with a scope. It's not just spurious tones etc., since the oscillation "jumps around" in a way typical of unstable circuits. The LM317 powering the

I disabled the analog input stage (textbook CS topology) by desoldering the two regulators powering it, but the problem persists.

The unit uses a CS5381 chip for conversion. All DC power lines measure OK. The two +5VDC lines needed for powering the AD chip and the rest of the digital parts are generated by two LM317. The unit is powered by an external 24V trafo (15V works just as well) and there are chokes before all regulators.

I replaced all the electrolytic caps, which didn't make any difference. They are at least as big as the chip documentation stipulates.

The LM317 generating +5DC for all the digital lines gets really hot and is soldered to a plane. It has turned the PCB a little browner via the heat.

I'm at a loss where to check now. There's nothing I can measure with at scope or multimeter showing malfunction. Could the AD chip have been damaged? Would it be any use trying and replacing one or both regulators? Anything else to check? Thanks!

 

[scott2000]

I chased my tail with weird ground/clock/whatever  noises on a couple of real old Myteks (ADDA) before and it was the DA on both and the AD as well on one...... But these were older 5394 and 4390.....The 4390 was a dip and in a socket thankfully.....

I did caps , relays. regulators, pulse transformers......a couple of other chips....lol

I think the AD was goofed by a previous attempted repair so I'd be inclined to say the DA chip...... But these were old so who knows....


and it's weird that I left the cs5394 (AD) out on one and the noise remained until I swapped the DA chip....All better even without the AD installed...

No schematics makes it tough....... There were no measurements I could make that showed anything wrong but I'm not super experienced

So I'd say yes, the chips can be wonky??


and one of the units had a heat sink on the 5v reg and the other (newer one) didn't...guess it wasn't needed....or previous owner/repair forgot it???///... maybe you could slide one on???

 

[JohnRoberts]

may be beat frequency interference between some other HF oscillation  and the A/D's clock.

JR

 

[living sounds]

may be beat frequency interference between some other HF oscillation  and the A/D's clock.

But what would be the source of that HF oscillation? There are only digital parts left, and they must be working ok, otherwise there would be no clock signal, would there?

Could the regulator have a problem? These old NPN Darlington based linear regulators are usually very stable, and I've replaced the low ESR electrolytics after the regulators with normal electrolytics (damping), which didn't make any difference at all. So I don't expect it to be the regulator.

Looking at the design of the converter it's funny the manufacturer claimed it has a special input stage with 6000 volts per microsecond - since it is pretty much the textbook differential design with an OP275 for input and a NE5532 as a buffer for the bias voltage...  :

 

[living sounds]

So I'd say yes, the chips can be wonky??

Thanks. I think I've got a replacement I ordered as a sample a few years ago. Desoldering a 26-PIN surface mount chip ain't much fun though...

 

[scott2000]

Thanks. I think I've got a replacement I ordered as a sample a few years ago. Desoldering a 26-PIN surface mount chip ain't much fun though...

I hear you.... I've got some pretty rough looking things because of my inexperience with it.  I know in a recent thread about an RME converter Volker here recently said he's pretty proficient with it and is in Vienna if you felt like not taking any chances.....

I don't know your level with that stuff but I'm still pretty intimidated by it.... I did find that lots of flux and that pre-fluxed skinny solder braid made things a lot easier for me when removing the cs5394-ks...but it wasn't the super tiny smd stuff..... still a pain though......

What converter is it???I just saw this on youtube....

Black Lion Audio Sparrow Micro Series ADC From The Inside

https://www.youtube.com/watch?v=32WOChm9RBw
@ 4:30 he mentions
"We suppress the mechanical noise of all the transistors switching on and off at high rates of speed"

Wonder how they do that??

 

[living sounds]

I hear you.... I've got some pretty rough looking things because of my inexperience with it.  I know in a recent thread about an RME converter Volker here recently said he's pretty proficient with it and is in Vienna if you felt like not taking any chances.....

I don't know your level with that stuff but I'm still pretty intimidated by it.... I did find that lots of flux and that pre-fluxed skinny solder braid made things a lot easier for me when removing the cs5394-ks...but it wasn't the super tiny smd stuff..... still a pain though......

What converter is it???I just saw this on youtube....

Black Lion Audio Sparrow Micro Series ADC From The Inside

https://www.youtube.com/watch?v=32WOChm9RBw
@ 4:30 he mentions
"We suppress the mechanical noise of all the transistors switching on and off at high rates of speed"

Wonder how they do that??

Thanks, I had been trying to locate that old video. Yes, that's the converter. There really isn't anything special inside of it. But the clock is better than all the other converters I've got here, most more recent and more expensive. Clocking my other gear and even travelling through an RME sound card that somehow is very much and verifiably true. Exactly as he describes the sujective quality of it.

As for the supression of switching noises I have no idea. There really isn't anything fancy inside apart from the four chokes, the Os-Con-caps and the high-precision oscillators. The 24V going in is also rather excessive, as it gets regulated to +5V in two places. But I've got an idea this kind of brute-force approach to power is also what gives the EMU SP-12/1200 its beefy sound...

I now wonder if it was wise to replace all the caps (all the ones around digital non-PSU stuff are now Panasonic FM/FR and the rest is run-off-the-mill types). But at least one of those Os-Con caps (on the inputs) tested practically dead, and replacing it fixed a problem of low end loss on one input...

I'll desolder the converter chip by first removing the electrolytics around it, adding lots of solder to both sides and then heating both sides at the same time with two soldering irons. This should do it, I think I've done it before with 20-pin ICs.

 

[scott2000]

But I've got an idea this kind of brute-force approach to power is also what gives the EMU SP-12/1200 its beefy sound...

Interesting.....

I now wonder if it was wise to replace all the caps (all the ones around digital non-PSU stuff are now Panasonic FM/FR and the rest is run-off-the-mill types). But at least one of those Os-Con caps (on the inputs) tested practically dead, and replacing it fixed a problem of low end loss on one input...

I'd be more inclined to replace caps before messing with any chips because of my inexperience but, if you have to remove them again to get at the chip I guess you have to make that call...... it's amazing  how long ago the 2000s started in regards to aging......you're saying you already did replace all of them???

Keep us updated.... Not much info out there on repairs of various converters...I think the only schematic , outside of some random diy ones, I've ever seen was for that old Symetrix 620 I put in the docs....

.Makes it tough it you don't have an understanding of the way they work... It sounds like you have a pretty good understanding though......

 

[living sounds]

Replaced the ADC chip. Was surprisingly easy to unsolder it, had no problems installing the new one either. But: No difference. Replaced the regulator getting hot - no difference.

With the ADC and the analog input stage out of the way, there should at least be no difference from the component side. There is only a single clock input on the ADC, so the only thing left is circuit/layout related.  The side of the ADC on the inside of the PCB (closer to the clock and PSU section) shows higher oscillation levels (both track at the same frequency) as well as higher noise.

Grounding problem?

 

[scott2000]

Have you tried disconnecting pin 1 to  see if that does anything????

"The unit is powered by an external 24V trafo (15V works just as well) and there are chokes before all regulators. "

What does this mean exactly????

 

[living sounds]

Have you tried disconnecting pin 1 to  see if that does anything????


What does this mean exactly????

It means they went to unusual lenghts to clean up the power.

I now found out the datasheets has quite specific local deocupling requirements, using a mixture of ceramic and electrolytic caps. The BLA converter only has the electrolytics, so I added more capacitance per the datasheet. No difference!

I'm really at a loss now...

 

[living sounds]

Here's an (amplified) sample of the noise:

http://www.dropbox.com/s/prqhkkm4hkbupwv/Sparrow_oscillation.mp3?dl=1

 

[living sounds]

Have you tried disconnecting pin 1 to  see if that does anything????


What does this mean exactly????

Pin 1? I have disabled the entire analog input section. Didn't make any difference.

 

[scott2000]

Oh right....Forgot to back up......

So what is this external transformer???

 

[living sounds]

I did it! I now added low ceramic value decoupling to the digital side, and the problem is gone.

I cannot believe how they could release such a deeply flawed commercial problem. Especially considering how easy and cheap it would have been to simply impliment the datasheets decoupling strategy. Standard practice, really.

Forget about all the marketing jargon in that video, they obviously had no clue what they were doing...

 

[scott2000]

Interesting! 

So this junk is that far down where the average consumer wouldn't know about it???

 

[JohnRoberts]

I did it! I now added low ceramic value decoupling to the digital side, and the problem is gone.

I cannot believe how they could release such a deeply flawed commercial problem. Especially considering how easy and cheap it would have been to simply impliment the datasheets decoupling strategy. Standard practice, really.

Forget about all the marketing jargon in that video, they obviously had no clue what they were doing...

I was thinking maybe a cap deteriorated over time but didn't consider missing from the original design.

Good catch, never assume the design is perfect (unless it was one of mine and even I make mistakes.)  : 

This raises the question, did it ever work right? A popular ASSumption when troubleshooting mature products.

JR

 

[living sounds]

I went through 3 of these converters within a couple years or so after they first came out 10 years ago. They broke down quickly, one way or the other. BLA replaced them until the unit was discontinued. I don't it ever measured that well, but I don't remember oscillation like this either. Not sure about the average consumer, but this was marketed as an inexpensive few-features-but-sonically-beats-much-more-expensive kind of gear. And I think this actually held and still holds true. Much of it has to do with the clock. I bought another one last month used because I wasn't happy with any AD converter I had since.

The PCB of the Sparrow is not equipped for any of the required ceramic decoupling caps around the ADC chip, so the design is marginal at best. The datasheet calls for 6 ceramic caps close to the respective pins each going to the correct ground. They added exactly zero. They must have read the datasheet, since the correct electrolytic caps were fitted.

I replaced the OP275 handling the input with a much better part (OPA2211) and got no improvement in THD. Have not yet compared it directly to my Lake People converter, which uses the same ADC chipset and a generally questionable op amp (OPA2604), but achieves much lower THD (I think).

 

[scott2000]

Thanks for the info! Very informative......

I actually got to play around with some LT1358 recently and was pretty impressed.... I know it's all subjective in my case and is circuit dependent but, I actually liked it in the end in  my scenario....

What kind of AD have you left behind in favor of the Sparrow if you don't mind me asking??

 

[living sounds]

Thanks for the info! Very informative......

I actually got to play around with some LT1358 recently and was pretty impressed.... I know it's all subjective in my case and is circuit dependent but, I actually liked it in the end in  my scenario....

What kind of AD have you left behind in favor of the Sparrow if you don't mind me asking??

Op amps are supposed to make very little difference once you arrive at a certain quality level and if implimented correctly, but from my own experience they do differ, in objective measurements as well as sonically. FWIW, I usually end up with NE5534/5532.

Currently in the running (converter wise) besides the Sparrow there is the PCM4222EVM with my own input board (using NE5534) and the Lake People RS 04. I replaced the audio op amps in the Sparrow with new ones (you never know what happened with caps going bad etc.) and added decoupling. This seemed to clear up the sound a little and the Sparrow is currently the front runner. But I have to make more tests to be sure.

Unfortunately, converters seem to be a thing that is extremely hard to get right, and the really good ADCs are very expensive. Plus it's difficult to get in several of these units at the same time for comparison. Or you have to buy them first and check them out later.