Layout for OPA Alice - for electret

[Toobme]

Howdy folks,

I've been trying to DIY an OPA1642 Alice circuit to power a 2-wire electret. I have tried breadboarding it (and heard my voice, albeit with noise due to no body/shielding), and have gone thru a couple layouts now on my homemade pcbs but get nothing but buzz. Full signal buzz.

I was hoping, after a few nights of struggling, that someone very kind could shed some light on my errors in this design.

I'm using the JLI-1813 and the Alice for OPA1642 circuit.
I'm dead set on building my own - to learn; but the brick wall has hit me and I'm seeking some guidance.

I'm pasting the schematic and PCB layout like I saw another user do (but they were coupling this with the hex converter) for help with troubleshooting.

Thank you very much for any advice!!

Layout:



Capsule:


Schematic:

 

[MicUlli]

The pinout of IC1 is wrong. Pin 5 <-> Pin 8 and Pin 6 <-> Pin 7. Change it and have fun...

 

[Toobme]

Wow, @MicUlli thank you!!
I will try rerouting by doing the swapping you suggest.
Good eye and mind set you have, there - many thanks for the diagnosis.


I updated the layout, swapping 5 for 8 and 6 for 7.

Now to make another PCB and test it out!
I'll report back later.
Thanks again!

 

[Toobme]

I tried just cutting the traces between those 4 pins and running jumpers (on my last board), but no joy.
So I'll build another PCB with the new layout (just above) and actually try it out.
Sheesh, this one is getting the better of me, but can't quit!

 

[Voyager10]

Sorry if this is an obvious question: this is properly screened when testing, is it?

As in: both the PCB and capsule are totally enclosed in a metal case which has a good connection to ground (XLR pin 1)?

If it's just lying on the bench you will certainly pick up mains buzz.

 

[Toobme]

A good point, @Voyager10 - It's not enclosed, but I thought I should still be able to pick up sound with the mic, along side any noise.
Now it's just silent, unless I touch the circuit, then I get a buzzing.
Still not hearing anything.
Argh! This has me loosing sleep - any other ideas out there?
Thanks very much, folks!

 

[ccaudle]

The pinout of IC1 is wrong.

Matches the datasheet:


Although I would like to point out to the OP that the schematic should be drawn with logical symbols, not physical. The way that schematic is drawn makes it very difficult to read.

You may need decoupling capacitors physically close to the op-amp. Do you have capability to check for high frequency oscillation at the two outputs?

You did verify you really received OPA1642 and not OPA1641 by mistake? Seems obvious, but sometimes it is good to check for assumptions.

 

[Toobme]

Thanks, @ccaudle - I’m new to this (obviously!!), and will certainly read up more on schematics and logical layout. I admit it’s bad! I want to improve, so thanks for that feedback.

I rechecked the data sheet last night and saw that the pinouts were not reversed as suggested, but thought I’d go ahead and do the trace cutting and jumper-wiring anyway - thinking perhaps the original schematic I referenced had some known-often-seen error that was easily noticed. I went with it.

Since the pin 5 on my 1642 (triple checked and it is in fact a 42) is where the virtual ground ends up (again, I’ll use the VGnd triangle next time), it didn’t make sense.
So I’m going to scrap that version and start again.

I’ll go back to my first circuit layout and try adding decoupling 22nF caps to the xlr area, and physically solder a 100uF to pins 4 and 8.
Then actually put it all into the body and headbasket and do real try before reporting back.

Thanks to all who’ve read/responded - I’m learning and slowly!

 

[Voyager10]

A simple check of DC voltage levels is generally the first thing to look at with a newly built circuit.

Looking at the schematic, IC1 V+ should be +12V, V- should be 0V and OUT_A and OUT_B should be +6V (measured relative to XLR pin 1).

The R4/R6/C3 junction and R5/R7/C4 junction should be about +21V each, assuming standard 48V phantom power.

 

[ccaudle]

read up more on schematics and logical layout. I admit it’s bad!

It is not bad, it just does not follow the convention for a amplifier symbols, so you have to go look up the part number to verify the connections.

again, I’ll use the VGnd triangle next time

I have no idea what a "VGnd triangle" is, I was suggesting you use the standard symbol for an op-amp instead of a rectangle with pins shown in the physical arrangement.
Wikipedia page on op-amps with example symbols

try adding decoupling 22nF caps to the xlr area

The term "decoupling" refers to capacitors added to reduce the impedance of the power supply connections. That means right next to the components, i.e. next to the op-amp. Caps next to the XLR connector might be useful for RFI filtering of the signal lines, but would be useless as power supply decoupling.

physically solder a 100uF to pins 4 and 8

A 100uF capacitor is going to be an electrolytic, which will have a rising impedance at high frequencies. They are only useful up to a few hundred kHz in most cases, and those are fine to be farther away. On a PCB this size you can place them anywhere, it won't really matter.
The op-amp has a gain-bandwidth product of over 10MHz, which means you need a reasonably low power supply impedance up to that frequency range. That requires a capacitor with good high frequency performance located physically close to the op-amp, so usually ceramic.
This app note from Rohm might be a good introduction:
Rohm capacitor impedance app-note

 

[Toobme]

Thanks for these helpful tips and links! I'm reading now, pausing, and seeing it's going to be a while.
You've given me things to test and to read up on, so I'm very thankful for that.

I'm now going back to my first layout, as there were no pin reversals going on in that PCB.

With that, it's off to re-route, reprint and etch another board.
Thanks for keeping it fun, even for the noobs like me!

 

[Toobme]

Before I etch another board, does anyone see anything wrong with this new layout?
View attachment 116492

 

[ccaudle]

Some of the reference designators have changed in your new layout, it no longer matches the original schematic. Was that intentional?

 

[ccaudle]

I just saw what was wrong with your original layout: the power supply was connected to the op-amp input, and the power pin was connected to the 47k/47k voltage divider.
That does not match the schematic. What kind of tools are you using that will let you make connections that don't match the schematic? Or did you not enter the schematic into a full schematic+PCB tool, and just tried to go directly to layout from someone else's schematic?

 

[Toobme]

Fusion360. Yes, made a fresh new schematic so refs are new.

Scrapped original after re-doing based on above comment about pins 5/8 and 6/7 being reversed.
Going to try this way and see - will report back.

Thanks, folks

 

[ccaudle]

based on above comment about pins 5/8 and 6/7 being reversed.

That comment was obviously incorrect, you can see that in the pin layout I copied from the datasheet in post #7.
Hard to keep track of which layout your comment refers to, since your latest posted in post #11 does seem to have power routed to pin 8, which is correct (and also matches your original schematic, which was basically correct, but the original layout did not match the schematic).

 

[ccaudle]

Fusion360

I am familiar with Fusion360 as a mechanical modeling program, not electrical CAD. Does it have a mode which allows entering a schematic, and then tying that schematic to a PCB layout such that it shows indicators of the logical connections (aka "rat's nest") while you are drawing the physical connections?
That is the most useful feature of combined schematic and layout tools, so that once you have verified the correctness of the schematic it will prevent making a mistake in connection while translating to a physical layout.

I'm now going back to my first layout, as there were no pin reversals going on in that PCB.

There were no pin reversals on the first schematic, but as I noted above the layout did not match the schematic. I think that was what MicUlli was referring to, which I did not catch at the time. When he said the pinout was wrong, I looked at the schematic and saw it was correct, and my experience with electrical CAD tools made me assume that since the schematic symbol was correct the layout must be correct. MicUlli obviously looked in more detail at the layout and caught that reversal but did not say explicitly it was only the layout connection which was not correct.

 

[Toobme]

I think what happened was the schematic (which I entered into Fusion) got un-sync’d from the pcb model as I was doing revisions.
I must have ignored a warning it gave when it got out of sync and blasted past, clicking Ok, Ok, Ok.
Lesson learned - slow down and don’t get physical until I’ve really triple checked everything.
Going thru my original layout would have revealed that - but I’m just too excited to get building, and sped past the important stuff (just like I tended to in math).

Well, thanks for having the kindness and patience to help me out on this one, folks!
I’m going to recook everything and start from scratch again with the layout.

Once I’m sure it’s ok, I’ll post a (hopefully) final version for completeness of this thread.
Cheers

 

[Toobme]

SOLVED - Final post about this little build.
It turns out that conflating my issues was the fact that I've never done double sided boards before, and dolty me neglected to solder the caps on both sides.
I just slammed 'em on like it was a manufactured board, so there was no connection.
I've since reinstalled them (standing proud off the board, mind you) and might remake another board with rivets, so I can get the caps to sit flush.

So, a humble lesson learned, but important and so worth the struggle.
When I finally put the thing into a body and connected it to the 702, it was all smiles.
Now I'm really hooked!

Thanks for the support along the way, as always!!