OPA Alice Failure

[Rusan]

I just built a pair of Alice OPA mics, using the stock boards ordered from PCBWay, the standard Jules Ryckebusch BOM and TSB2555B capsules in BM800 bodies. I triple-inspected every detail of the build during and after, and had meticulously removed every trace of visible flux residue immediately after populating and soldering the board.

When I connected them to my Focusrite Scarlett, one didn't work. A quick check showed that all the circuit voltages were extremely low, including the 48V coming from the Scarlett. With the mic unplugged, the voltage on pins 2 and 3 of the cable was back to a normal 47.7V. Uh oh, something's drawing waaay too much current. So, once I got the mic connected to my bench power supply, I saw that the circuit was drawing a whopping 34mA, instead of the expected <5mA. I replaced the OPA1642, and the circuit voltages tested pretty much perfect according to Jules' stated voltages.

VCC_1 =12.35V
VCC_2 = 11.52V
R8/R10/R11 Junction = 5.81V

Awesome, we're back in business, or so I thought! I plugged it into my Apollo and it worked great for few minutes, then silence. I checked the voltages, and they were again similar to the previous extremely low voltages. Back on the bench power supply, it was drawing the same 34mA of current! I don't know which amazed and dumbfounded me more, that it was somehow drawing that much current in the first place, that it had repeated itself, or that the OPA1624 didn't explode.

I examined and re-examined every millimeter of everything, and couldn't see anything wrong anywhere. I laid it aside and went back to playing with its functioning brother, which in a few minutes stopped sending audio as well. Guess what? Same symptoms.

After inspecting everything until my head hurt and my brain started shutting down, there are no shorts or mistakes anywhere in either mic that I can see, and well, the fact is the same thing has now happened three times on two different mics. A bad new opamp is rather rare anyway, but two and especially three is impossible, and certainly not by failing in the same way in mere minutes, when connected to two different interfaces. I very closely examined and traced a blank board from the same batch to ensure everything was correct, and it indeed seems to be. All the electrolytics are brand new Panasonic that I bought a few weeks ago from Mouser, not 20 year old caps dug out of my boxes of unused junk. The OPA1642s likewise came from Mouser, not eBay or Amazon.

I've successfully troubleshooted circuits of immensely greater complexity, and certainly this is waaay too simple of a circuit to defeat me that easily, or at least I thought it was... but, I'm perplexed. Any ideas?

 

[pucho812]

https://www.jlielectronics.com/content/TroubleshootingOPA-Alice MicRev1.pdf

 

[Rusan]

The Troubleshooting OPA-Alice pdf was actually where I got the reference voltages from, and they do correspond to the proper 1.8mA per opamp channel if you Ohm's Law them with the circuit resistor values. Which, said proper voltages I did have for a few minutes until the OPA1642s went berserk.

It's a proven design that's been built using the exact same boards probably hundreds of times without issue, plus the fact that I've obsessively checked every minute detail about a dozen times now. I've never seen anything like this before, especially in such a simple circuit, and all reasonable logic says I got an entire bad batch of opamps. But, OTOH logic says it's also impossible to get an entire bad batch of opamps, especially when they came from Mouser and not a dubious source.

I guess at this point, I'm gonna remove the boards from the mics and replace the opamps, and test them open-air. However, there are definitely no shorts anywhere to the metalwork or between things that shouldn't be, no solder bridges or flux on the opamp legs, no misinstalled componentry or anything else physically amiss after repeated examinations under bright light and strong magnification. The most perplexing thing is of course the fact that both mics worked perfectly for a few minutes before failing in exactly the same way.

 

[Khron]

All three mics behaved this way with the capsule connected?

And do you happen to have (or can borrow) an oscilloscope? The opamp could conceivably break into oscillation, possibly.

 

[Voyager10]

I'm a bit puzzled by 34mA current from a bench supply. How are you connecting it to the circuit, exactly?

If you were providing phantom power by setting the bench supply to 48V and connecting via 6k8 resistors to XLR pins 2 and 3, you wouldn't get 34mA even with a dead short to ground.

If you're applying power directly to VCC_1 there will be sharply increasing current through the Zener diodes D1/D2/D3 as the voltage reaches 12V.

 

[Rusan]

All three mics behaved this way with the capsule connected?

And do you happen to have (or can borrow) an oscilloscope? The opamp could conceivably break into oscillation, possibly.

Two mics, three times; actually, one was DOA when I first plugged it in, but it exhibited the same symptoms as the later failures. The electret capsules were connected, and both mics were passing audio normally for a few minutes before failing. I have an oscilloscope, function generator, bench and handheld DMMs, all the basic stuff. Just before posting, I was actually connecting the scope when my wife came in to talk, and I'd already been fiddling around with the thing forever so I decided to take a break and then posted.

I'm a bit puzzled by 34mA current from a bench supply. How are you connecting it to the circuit, exactly?

It's connected through 6.8K resistors to pins 2 and 3, with the supply set to constant voltage mode at 48V. Honestly, I didn't check voltage drop across the resistors to verify the current; 34mA and 1.632 watts are what the meters on the supply said was being drawn.

If you were providing phantom power by setting the bench supply to 48V and connecting via 6k8 resistors to XLR pins 2 and 3, you wouldn't get 34mA even with a dead short to ground.

My thoughts as well (at least not without the opamp instantaneously exploding!), but the display on my supply has always agreed with my DMMs and Ohm's Law math. All the circuit voltages test extremely low due to voltage drop across the resistors, so something of that magnitude is indeed going on, even without actually calculating the current. I likely won't have time to revisit it until next Friday, but I'll post my findings when I do.

 

[peterc]

Did actually measure each resistor before you soldered it in? Or each capacitor?
I have found that This simple step has reduced my project fault finding to virtually zero.
Sometimes "not my fault", as on some of the parts I buy, the brown is very similar to the red and on other parts the red is very similar to the orange.
So I would suggest that your step is to verify all of your components with a multi meter in circuit, and those that look incorrect, pull them out and verify out of circuit.
I know it's not what what you want to do, but from experience that would be my next move.
Good luck!

 

[Voyager10]

OK, it's possible then the power supply meters are just a little off at low current. If XLR pins 2 and 3 are close to 0V then that's the immediate problem and the actual current is of little relevance.

Can you measure a short circuit/low resistance between Vcc_1 or Vcc_2 and ground on the failed board? Did you keep the "dead' chips, and if so do they measure short-circuit between pins 4 and 8?

Things which might cause the chips to die could be (a) diodes D1/D2/D3 failing and exposing the chip to a power supply voltage > the abs max rating (40V) and (b) caps C4 or C5 failing and pulling an output pin higher than the supply voltage.

 

[Rusan]

Did actually measure each resistor before you soldered it in? Or each capacitor?
I have found that This simple step has reduced my project fault finding to virtually zero.
Sometimes "not my fault", as on some of the parts I buy, the brown is very similar to the red and on other parts the red is very similar to the orange.
So I would suggest that your step is to verify all of your components with a multi meter in circuit, and those that look incorrect, pull them out and verify out of circuit.
I know it's not what what you want to do, but from experience that would be my next move.
Good luck!

Absolutely. It's my standard modus operandi, that I'm beyond OC about. I always test every resistor and cap's value before installation, and check twice for proper orientation of electrolytics and semiconductors. I've already went back and checked everything many times in the unlikely chance that I did make a mistake, but everything looks fine. And, I tested everything in-circuit that could be. Nothing's seemingly wrong anywhere, except for the opamp frying after a few minutes.

Can you measure a short circuit/low resistance between Vcc_1 or Vcc_2 and ground on the failed board?

No, that was one of the first things I checked. There are no shorts or too-low R to ground anywhere, and everything's fine with the virtual ground and the rest of the power rail, as far as I can tell.

Did you keep the "dead' chips, and if so do they measure short-circuit between pins 4 and 8?

Yes, the power pins of each chip were shorted, which is how I knew they were fried.

Things which might cause the chips to die could be (a) diodes D1/D2/D3 failing and exposing the chip to a power supply voltage > the abs max rating (40V) and (b) caps C4 or C5 failing and pulling an output pin higher than the supply voltage.

The stacked 6.2V zeners in both boards test okay, but I'll probably replace 'em with a single 12V anyway, just to be sure. Good call on the output caps! I never thought to check them for leakage, since they're brand new. But, it happens. It might be a few days before I can get back to the bench, but I'll check all the caps for leakage and see what it turns up. Thanks!

 

[Voyager10]

I'd be tempted to do a burn-in test when you next replace the chip. Apply 11V at Vcc_1, with a small current limit (10 or 20mA), don't connect anything else, and just let it sit. (You can replace the capsule with a 100pF-1nF cap).

If the new chip still dies, I can't think of any circuit fault which would cause it. Either a tube of bad chips or something systematic in the handling or soldering process.

 

[Emil C]

Check your bench power supply and ground/supply wires/configuration in general!

Imax = 48V / 6.8K = 7.06mA

Vcc = Isc x 6.8k = 34mA x 6.8k = 231V !!!

 

[Khron]

Imax = 48V / 6.8K = 7.06mA

... Per each of the two XLR signal pins.

 

[Emil C]

.. or Vcc = 34mA x (6.8k II 6.8K) = 34mA x 3.4K = 115.6V

 

[blue_luke]

Just a suggestion here but are the caps you use electrolytics or tantalums?
Tantalums are very unreliable if their voltage rating is too close from the voltage applied.

 

[berret]

Looks like the 6k8 resistors are in fact 680 ohm...

 

[Rusan]

If the new chip still dies, I can't think of any circuit fault which would cause it. Either a tube of bad chips or something systematic in the handling or soldering process.

Nope, no issues with handling or soldering. I have plenty of experience handling and soldering ESD/heat-sensitive components, and use ESD-safe tweezers and soldering station, wrist strap, etc. As I said, I'm beyond OC about this stuff. Which brings us to:

I now have a working OPA Alice, after I unexpectedly got a few free minutes tonight to spend at the bench. I replaced the OPA1642 with one from a new batch that I purchased about two weeks ago, with the intent of building additional mics in the near future. The ones that failed were from a batch purchased back in November.

All voltages test spot-on perfect, and the mic is functioning flawlessly. I started this thread because I thought it was a literal impossibility to get a package of five bad opamps from an authorized dealer (versus eBay or Amazon), so I disregarded that possibility altogether. I guess it's not impossible after all.

I haven't tried the mic with anything except my voice yet, but I played around with it for a good while and it sounds okay. Not mind-blowing great, but incredibly good for a $10 electret capsule and a total cost of $60 USD per microphone, including sales tax and shipping costs. I intend to use them as cheap utility mics for a Mixlr live stream and non-critical spoken word in a semi-permanent installation, and maybe give a couple to friends. Quite a fun project, were it not for the frustrating opamp debacle. Thanks to all who posted!

 

[Voyager10]

Yes, thanks for letting us know the end of the story. I suppose it's a case of "when you have eliminated the impossible, whatever remains, however improbable, must be the truth."

 

[Toobme]

@Rusan - I have been pulling hair over here with an OPAlice build as well. Tried everything but a new batch of opamps.
Hearing this, I’m going to do another order for a hopefully new batch (assuming this may be my cause as well), pull some more hair waiting for delivery and give it another shot.
I had ordered mine in September from one of the majors (either mouser or digi, would have to look again), so assumed that any problems were with my build or layout, etc.
Thanks for posting this update!

 

[Rusan]

What are the issues you've been having? The OPA Alice Troubleshooting thread's already been started....

 

[Toobme]

Of the batch I got from mouser, only one worked in circuit.
Three passed no signal, two were just buzz.
One has worked.
These were OPA1642AID - couldn’t find just 1642 without the suffixes.
I’m ordering a new batch and will try again and post if successful.
Cheers