[Hifihedgehog]

It would work, but it would probably be pretty noisy.

Might be better to at least power a seperate linear PSU from the 12v rails of the computer's PSU.

I have no issues with power supply noise or noise on the output when using IEM's even. I don't quite understand where you are coming from as far as the computer power supply is concerned unless you a suggesting a true dual supply with a real ground.

 

[tangent]

if all it is is only around 5 mV, I shouldn't really be too worried about it. I mean, it is still three quarters of the way from the 20 mV mark.

 
This is true.  However, it does seem to be a legitimate mystery, rather than something like some meaningless process variation.  I recommend that you pursue it only if you would be gratified to know the answer.  If you don't care why it's happening, don't waste your time.
 
Rotating the buffers was a good test.  I should have thought of that myself.
 
You've already rotated the op-amps, which means you're now down to the passives.
 

 I plan on going through the Trace a Signal through the Amp section next.

 
That's only likely to find problems in the signal channels, not in the virtual ground.
 
I was hinting that you might check the amp's quiescent current.  It even walks you through how to add the BUF634 contributions to the supply draw.

 

[Hifihedgehog]

Thanks! I will do that--I especially don't like this being left unanswered, since electronic analysis and diagnosis and all should be my forte as a computer engineer. Once I have college finals in my wake, unraveling this mystery will be my next task at hand. It may just be a matter of a bad solder joint somewhere down the line, so I will reheat and redo the joints to cover my bases there. The resistors measured okay, but I will double and triple check everything in the ground channel to see if I may have past over something.

 

[Hifihedgehog]

To those who have eyed the recently released PIMETA V1 design in this thread: if you were leery of using this design with the LMH6321 on account of the short circuit issue while disconnecting headphones that would burn up those buffers, worry no longer. MAGNEZ on Kickstarter is a new kind of audio jack adapter that solves not only the breakaway issues that jacks have but this shorting issue as well. 
 
Link: https://www.kickstarter.com/projects/magzet/magzet-the-audio-jack-reinvented-with-the-power-of/

 

[tico420]

Had a v2 board laying around for years, finally got around to finishing it. Sounds great, thanks.
 
OPA637 on left right channels, AD8065 on ground.
 
Alps big pot
 
24V tread

 

[whitelabrat]

I've had a board laying around too and I'm just waiting on four resistor values to show up.  Accidentally ordered 0402 package sized resistors.  Ha!  So small.
 
But yeah, hoping to brown dog some OPA637 as well.

 

[whitelabrat]

Got it working, and it sounds really good. So happy. The OPA37 opamp for the ground channel is crazy hot though so I'll need to tinker with that. Running about 0.60 mA for the class A bias.

I really like having the option to do SMT or through hole.

Photos: (my soldering is not super)

 

[tangent]

The OPA637 isn't unity-gain stable; the ground channel is unity gain. That's why they make the 627.

 

[whitelabrat]

Heh. I put the opa37 (not 637) in backwards. Perfect now.

 

[tangent]

The OP37 is also not unity-gain stable. From the datasheet:
 

 the design is optimized for circuits with gains greater than five. 

 

[whitelabrat]

WAT?  looks at datasheet.  almface:
 
Ok.  I do realize I'm one of the knuckleheads often referred to in the instructions.    Popped in an OPA27GP.  Because I have some.
 
@tangent nice.

 

[Hifihedgehog]

Keeping this thread alive and sharing a hot deal for makers! I simply joined PCBWay and I recently received a $45 credit that was enough to cover both the fabrication and shipping and handling costs of 20 100 x 80mm 2-layer PCBs, absolutely free for me! For potential makers of the Pimeta V1 and he-who-shall-not-be-named's dioxygen amplifier, I cannot recommend PCBWay enough for their quick and impeccable service.
 
http://www.pcbway.com/

 

[spbkaizo]

Firstly, Apologies for dragging up a rather old thread - but I'm out of ideas.

I decided to build some of these, so I ordered some boards which arrived and I started to populate 3 of them.

One board became leader, and once I got to the section "check power" and everything was fine (voltage split +/-V correctly) on the op-amp pins, I then proceeded to populate the buffers, op-amps, etc as instructed. First test after that revealed there was the following voltages on the following pin combos (18v bench supply):

1. PG/OPALR V- = -9V
2. PG/OPALR V+ = +9V
3. PG/OPAG V+ = +9V
4. PG/OPAG V- = -9V

So, all still ok. Moving onto check the IG to Outputs revealed some scary results:

1. IG-OG = +8.5v
2. IG-OR = +7.4v
3. IG-OL = +7.4v
4. OG-OR = 110mv
5. OG-OL = 110mv

OK, so clearly not right. Been unwinding the build and rebuilding it, checking as I go. I was stumped.

As always, try to reproduce. So I got one of the other boards, and started to build slowly and checked everything as I went along. I got as far as installing the Op-Amps, and checked the input signal of a sine wave on a scope - perfect, on both pins 3 & 5.

Moving on, I checked the output of pins 1 & 5 - again, perfect, and absolutely in line with my gain settings. So, I decided to solder 1 of the buffers into place.

And this is where things get weird.

Pin 2 of the output buffer goes to R11 - I carefully made sure that Pin2 did indeed go to R11 (currently unpopulated), soldered down the legs and re-checked. Now, output from the corresponding OPALR Pin 1 is near zero, but the other output on Pin 7 remains fine. I believe something to be amiss with the buffers - but would like support in this conclusion, as well as any other pointers you may have.

https://photos.app.goo.gl/RRdrY7HU1VgUvSiQ9
https://photos.app.goo.gl/6YTiRAXQCXPXUd8cA

 

[tangent]

I ordered some boards

Do you mean that you found a store selling simple copies of my board, against my wishes, or that you took my Gerbers and had a board house make some for you, in accord with my wishes?

If the latter, did you tell them it needs to be a 4-layer board? If you made it as a 2-layer copper board, that'll explain a whole lot of trouble.

Your close-up photo of one of the buffers (BUFR?) shows a number of problems with the physical realization of these boards relative to an original I've got sitting around here still:

1. Pin 4 appears completely flooded, not fenestrated into the plane.
2. The originals don't have those thin fingers between the pins, most egregiously between 2&3, and between 6&7.
3. Whoever ordered the boards elected not to include the bottom side silk layer, which means you might not be populating the components in the right spots or in the right orientation

Observations 1 & 2 call into question whether these boards do in fact express the schematic correctly. What other creepage and tolerance problems are there, which could cause shorts, opens, and parasitics where we don't want them? I've made a close-in screenshot of what my Gerber viewer shows for BUFR in the bottom copper layer, which I believe is what you're showing in the second picture. Quelle différence!

Observation 3 means we're reliant on your photos to tell whether your buffers are oriented correctly on their pads. It looks like two of your buffers are oriented on the pads correctly; you don't show the third.

Moving onto check the IG to Outputs revealed some scary results:

What op-amps are you using? Maybe post pictures.

I checked the output of pins 1 & 5 - again, perfect, and absolutely in line with my gain settings. So, I decided to solder 1 of the buffers into place.

I don't see how your op-amp tests gave sensible results without the buffers in place, especially the gain, since that means you'd be running on the inner Jung loop, which has a gain far higher than you'd be expecting.

Unless you shorted the buffer positions, you should not have gotten sensible results from that test.

Pin 2 of the output buffer goes to R11 - I carefully made sure that Pin2 did indeed go to R11 (currently unpopulated), soldered down the legs and re-checked. Now, output from the corresponding OPALR Pin 1 is near zero,

I can't tell whether this is a grammar error — should these be two separate paragraphs? — or you're trying to get some point across to us that I'm missing. There's nothing about sentence 1 that leads us to sentence 2, and sentence 2 doesn't in any way clarify sentence 1.

OPALR pin 1 is a signal output pin, and BUFL pin 2 is a power modifier. You would not expect any connection between them, nor is there supposed to be.

I believe something to be amiss with the buffers

Did you get them from a reputable source?

Maybe someone who's bought some of these recently could tell you whether they're still marking these chips with the Nat Semi logo so many years past their absorption into TI. Mine are so-marked, but they're from before the merger.

any other pointers you may have.

The board could use a cleaning. It wouldn't be the first time I saw that solve a problem.

 

[spbkaizo]

Would just like to say thanks for your time, in writing your response, as well as for the original work you did.

Do you mean that you found a store selling simple copies of my board, against my wishes, or that you took my Gerbers and had a board house make some for you, in accord with my wishes?

I had the boards made, from your Gerbers.

If the latter, did you tell them it needs to be a 4-layer board? If you made it as a 2-layer copper board, that'll explain a whole lot of trouble.

Was done as a 4 Layer Board.

The order artefacts are available for download from them, https://jlcpcb.com/file/download?uuid=135a5870c7d84874a24e2abef588ca54&businessType=example

Your close-up photo of one of the buffers (BUFR?) shows a number of problems with the physical realization of these boards relative to an original I've got sitting around here still:

1. Pin 4 appears completely flooded, not fenestrated into the plane.
2. The originals don't have those thin fingers between the pins, most egregiously between 2&3, and between 6&7.
3. Whoever ordered the boards elected not to include the bottom side silk layer, which means you might not be populating the components in the right spots or in the right orientation

Observations 1 & 2 call into question whether these boards do in fact express the schematic correctly. What other creepage and tolerance problems are there, which could cause shorts, opens, and parasitics where we don't want them? I've made a close-in screenshot of what my Gerber viewer shows for BUFR in the bottom copper layer, which I believe is what you're showing in the second picture. Quelle différence!

Observation 3 means we're reliant on your photos to tell whether your buffers are oriented correctly on their pads. It looks like two of your buffers are oriented on the pads correctly; you don't show the third.

I've done a side by side image of the viewer in Eagle, vs the Board - I'm not clear how I could have gone wrong.

Here's a better picture of the board: https://photos.app.goo.gl/7g3BySSVFQnZqyQy8

What op-amps are you using? Maybe post pictures.

For testing, an OPA2134, OPA2132, OPA134.

I don't see how your op-amp tests gave sensible results without the buffers in place, especially the gain, since that means you'd be running on the inner Jung loop, which has a gain far higher than you'd be expecting.

Indeed - I meant to demonstrate the opamp was in fact increasing gain, and therefore was alive.

I can't tell whether this is a grammar error — should these be two separate paragraphs? — or you're trying to get some point across to us that I'm missing. There's nothing about sentence 1 that leads us to sentence 2, and sentence 2 doesn't in any way clarify sentence 1.

OPALR pin 1 is a signal output pin, and BUFL pin 2 is a power modifier. You would not expect any connection between them, nor is there supposed to be.

Apologies. I meant that I output on PIN1 of opamp drops to 0 after soldering the buffer to the board. Removing the buffer causes the signal to reappear.

I was referring to BUFL2 in that statement as a chip orientation hint.

Did you get them from a reputable source?

No, I'd have to say I didn't. They're hard to track down, at a reasonable price so I ordered 10x from a Chinese supplier. Caveat Emptor.

As it stands, I have 10 boards. I've traced the schematic through the board, and all seems reasonable. Some questions if I may:

• Is there a way to test that the inner layers are in fact present, and correct?
• I have a "reusable" SOIC8 - DIP8 adapter with more reasonable pin distances for testing. If I pop a buffer in there, is there a way to test the operation of it you can recommend?

In addition, I'm going to order another single buffer from a reputable supplier. However, if you think the boards are a lost cause then maybe I won't

TIA,

S.

Thanks for replying.