[Abyss Audio] (2) PCB Rescue Operation — OPA Bi‑Variable Push‑Pull Composite Amplifier (Headphone Amp) DIY

Good evening. A cold wave has arrived — and my PCB completely self‑destructed! ><

It was a “steaming hot bath” incident!

Seriously, it was rough.

But I learned a lot… really. Sweat

In the end, after repairing the first‑edition PCB, I managed to complete the assembly, and at least for now, listening tests are OK.

As designed — great sound! I did it. ^^

And the schematic in the photo… yes, that’s the “accident property” circuit I posted in the previous article. Cold  

Two major disaster points are clearly visible… Sweat

Here’s what happened:

After simulating in LTspice, I manually re‑entered everything in KiCad and generated the PCB.

And yes… without checking for catastrophic mistakes, I went ahead and ordered the PCB. ^^;

(I sacrificed six transistors before noticing.)

(There was white smoke…)

(It smelled terrible!)

So the photo shows the corrected version.

The first edition is now covered in scars. Sweat

Even though I checked the schematic and double‑checked everything… the moment I powered it on, steam rose like a hot spring.

After two days of inspection, I finally realized the op‑amp and the pull‑side PNP were both reversed. Sweat

(Yeah… that was dumb.)

Corrected Circuit

Here is the corrected schematic.

I also added a few improvements beyond the major disaster fixes.

One is in the input section:

After the RF noise‑reduction high‑pass filter, I added a DC‑cut low‑pass filter.

It worked fine on the breadboard, but during PCB accident handling, the circuit started oscillating.

Since it was mid‑disaster, I also became concerned about input DC offset, so I added countermeasures.

Another improvement:

I added resistors (R9, R10) between the OPA output and the VBE multiplier.

Once I finally got sound output and started adjusting the semi‑fixed resistor in the VBE multiplier, the current fluctuated more than expected during adjustment.

So these resistors help suppress that and prevent the op‑amp from overworking.

The last improvement is around the ground layout.

I refined the separation and connection point between PGND and AGND for better stability.

However, even with cutter‑knife modifications on the first PCB, the sound quality and stability were already more than sufficient to my ears, so I kept the PCB revision only as a backup. ^^;

In the current repaired first edition, I didn’t include the DC‑cut low‑pass, but with the PopoDAC combination, the input DC offset was practically 0.001 mA or less, so the first edition is now complete.

Next, I’ll put it in a case to prevent further accidents, and spend some time listening and collecting data.

I briefly tested NE5532 and OPA827.

As intended in the design philosophy, the op‑amp doesn’t drastically change the sound quality, so the approach seems correct.

The concept of this headphone amp is:

“Deliver the PopoDAC’s sound quality to the ears as faithfully as possible.”

Separation and soundstage also matched the DAC’s intended presentation.

Top 3 Lessons Learned from the Steam‑Smoke Accident

“Do you like accident properties?”

Well, well… deep down, most people are fascinated by accident properties — gossip is delicious, right? Sweat  

And that interest probably comes from the valuable lessons you can learn. Cold

So here are the top three lessons. ^^;

1st place: Manual re‑entry from simulation to KiCad makes you drift

2nd place: Breadboards are “weaker than a loyal Hachiko,” so don’t trust them too much

3rd place: VBE multipliers are trickier than they look

How is it?

“Come on, you should already know that!”

Yeah… that level. Sweat

Next time will be casing, listening samples, or measurement data.

If possible, I’d like to hear the differences between PopoDAC’s MC profiles!

Stay tuned♪