SURGEON 25: Mastering EQ Build (Part 7) – Audio Troubleshooting, Testing, Completion, and Reflection
- Eitan Brown
- Feb 24
- 7 min read
Updated: Feb 25
Audio Troubleshooting: Electrical and Mechanical Faults
By the end of Part 6, the unit was assembled, calibrated, and behaving mostly as expected. Mostly.
What followed was not a single fault, but several. Some were electrical. Some were mechanical. Each demanded a different kind of attention.
The schematic was never in question. The power rails were stable. Offsets were trimmed. Unity gain had been verified. And yet, something was not right.
On the electrical side, one channel measured about 1 dB lower than the other with the EQ engaged. Not catastrophic, but unacceptable. On the mechanical side, a few switches misbehaved and failed to function properly.
I began with the electrical side.
The R18 Fault
The level discrepancy did not announce itself dramatically. It was subtle, about 1 dB lower on Channel 1 when the EQ was engaged. In bypass, both channels matched. With EQ engaged at unity, the difference appeared.
My first assumption was calibration drift. I rechecked offsets and trim points. Everything lined up. The discrepancy remained.
I spent a long stretch measuring. I traced the schematic node by node, comparing Channel 1 and Channel 2 through the entire EQ-IN path. Supply rails were stable. Reference voltages matched. Resistor values measured correctly in circuit. Nothing obvious was wrong.
Only after exhausting those checks did I suspect an active component. The op-amp in the low-band EQ-IN circuit was the most complex element in that area. If it were slightly off internally, it could explain a small but consistent level drop.
Replacing it was escalation, not instinct. I removed it carefully and installed a replacement. I powered the unit. No change.
I returned to the schematic and the meter. I measured under load and at idle. I probed surrounding nodes, trying to justify another replacement. Eventually, the trimmer became suspect. Not because it clearly failed, but because cleaner explanations had run out.
Again, replacing it was a last resort. Again, careful heat. Again, power up. No change.
At that point, the margin for error felt smaller. Each part removed carried risk. I began reflowing nearby joints methodically, following shared nodes in the schematic.
Then something shifted.
While probing the low-band EQ-IN node, I noticed the resistance reading move depending on probe pressure. When my hand brushed C3, the value drifted more dramatically. I repeated it deliberately. Slight pressure, release. The reading moved predictably.
C3 shares a node with the top of R18. They sit side by side. If that node were unstable, mechanical influence would explain the drift. After confirming the pattern multiple times, I replaced C3. The discrepancy remained.
That left R18 and the solder joints tying that node together. Visually, the solder at the top of R18 looked perfect. Smooth. Properly wetted. No visible fracture.
But the meter told a different story. Channel 1’s resistance-to-ground at that node drifted under light pressure. Channel 2’s did not.
I reflowed the R18-top joint and the corresponding node at IC6 pin 3, allowing the solder to fully wet and settle. I powered the unit again.
The resistance stabilized immediately. The 1 dB discrepancy disappeared.
After hours of tracing and escalating only when necessary, the fault resolved into a single marginal solder joint. The relief was quiet but real.
Mechanical Fault Example: Gain Switch
The high-band gain switch on Channel 1 began behaving strangely.
At first, the detents felt one position off. The mechanical click suggested one setting, but the measured gain indicated the next one. It was odd, but the unit still passed signal, so I kept working through other checks.
Then I attenuated the gain toward minimum.
At the final position, instead of reaching maximum attenuation, the output jumped to maximum boost.
That was not a subtle error. That was a serious fault.
My first suspicion was a short between pins 1 and 24. On a 24-position gain switch, that would explain the extreme jump from minimum to maximum. I checked continuity. There was no short.
The symptom was consistent. Every position was shifted exactly one step. Cleanly. Predictably. That consistency made it more confusing.
I opened the switch.
Inside, one of the pins had shifted slightly out of plane. The misalignment was small, but enough to interfere with the wiper’s movement. The wiper spring had deformed as a result. It had bent close to a right angle and was reaching directly across to the adjacent pin throughout the entire rotation range. That is why the switch tracked one position off every time.
The mechanical fault had created an electrical one.
I realigned the pins carefully and replaced the damaged wiper using the unused pole of the same switch. I reassembled it and mounted it back in place.
This time, each detent corresponded exactly to the expected gain step. Minimum meant minimum. Maximum meant maximum. The tracking was correct.
Other mechanical issues surfaced during this process.
Two low-frequency switches had gotten stuck because their pins had shifted just enough to interfere with rotation. On Channel 2, the high-band gain switch had no effect at all because its body was touching the chassis floor bracing and routing current directly to ground. In another instance, a gain switch would boost but would not cut due to a severed wire inside a crimp housing.
Each problem required a different solution. Realignment. Clearance adjustment. Re-crimping and rebuilding a connector.
One by one, the mechanical faults resolved.
Testing
After addressing the faults, I half-expected something new to surface.
With EQ bypassed, output matched input. With EQ engaged and all bands set to unity, output still matched input. Channel 1 matched Channel 2.
So far so good.
The next step was the null test.
A null test works by sending identical signals through two paths, inverting the polarity of one, and summing them. If the signals are truly identical, they cancel completely or “null,” and what remains is silence. Any residual sound represents the difference between the two paths, because the same signal traveled through both. A perfect null is objective scientific proof that the two paths are identical.
I performed multiple null tests: original audio file vs. the same file recorded through the unit in bypass; original audio file vs. the same file recorded with EQ engaged at unity; and the two recorded files against each other.
Each test produced a perfect null. Complete silence.
I was pleasantly surprised.
Beyond null testing, I measured noise floor and total harmonic distortion. The noise floor represents the inherent self-noise of the unit with no signal present, while total harmonic distortion (THD) reveals how much additional harmonic content the circuit generates when signal is applied. In high-quality analog mastering gear, a noise floor below roughly −110 dB and THD well under 0.1% are generally considered more than acceptable benchmarks.
The noise floor sat at approximately −124 dB.
Total harmonic distortion under aggressive boost conditions showed a second harmonic at roughly −77 dB (about 0.014%). Higher harmonics did not register meaningfully above the analyzer’s floor.
Those figures were not marketing claims (although they could be utilized as such).
They were simply what the analyzer displayed.
For an analog mastering equalizer, those numbers sit comfortably alongside top-tier commercial units.
Listening
After all the measurements, analyzers, and objective verification, the real test is how it sounds and how it behaves in practice. Numbers can confirm accuracy and stability, but once that accuracy is confirmed, the numbers can be set aside and we can judge with our ears, as Uroš has said.
At unity, the unit remained transparent. The null tests proved that and my ears confirmed it. Once I started turning the knobs, the character of the EQ began to reveal itself. Not from distortion or coloration, because there was none. It came from the interaction between frequency, Q, and gain, and from the way bands interacted with one another.
Low Q settings are wide and gentle. Higher Q settings become surgical without sounding harsh. The behavior is smooth and musical.
The measurements replaced uncertainty with data, and listening to the unit in action confirmed that Surgeon 25 is ready to be sealed up and completed.
Completion
With the measurements confirmed and the listening tests behind me, nothing remained open on the bench.
I secured the boards. I checked the wiring for strain. I stepped through every switch again before closing the lid. I aligned the knobs and tightened them carefully.
Then I installed the top cover.
The interior, which had demanded months of attention, disappeared behind black and silver aluminum.
I powered it up again in its fully enclosed state.
No unexpected behavior. No new noise. No drift.
I let music run through it.
Eight months of design, solder, revisions, and persistence now lived inside a finished enclosure, ready to be racked and put to work.
Reflection
Surgeon 25 was a project that provided learning and professional development.
It was never only about building an equalizer.
It was about learning the difference between understanding something in theory and understanding it with your hands.
Schematics are clean. Real components are not. Mechanical tolerances matter. Solder joints matter. Routing matters. Patience matters. Small oversights compound. Small corrections accumulate.
Audio troubleshooting forced me to slow down and separate assumption from observation. I had to trust measurements, verify them, and question them again.
There were moments of frustration. There were moments of doubt. There were moments where I thought I had made an irreversible mistake.
But every problem had a cause. Every cause had a solution. Sometimes the solution was technical. Sometimes it was mechanical. Sometimes it was simply persistence.
Surgeon 25 now exists as a working tool. It measures cleanly. It nulls perfectly at unity. It remains transparent when asked, and shapes tone with control when engaged.
I know how it is wired, where it can fail and how to repair it. I know what it does under measurement and under music.
Before closing this chapter, I owe deep thanks to Uroš Đorđević for his endless patience and generous guidance throughout this process. His willingness to answer questions, review decisions, and share his knowledge made this build possible. I came out of this project not only with a mastering EQ, but with a friend.
I also want to thank Shun Yoshino for his help in sourcing hard-to-find parts and for his insights on power supply design and related issues. His perspective saved time and prevented avoidable mistakes.
Finally I want to thank my friends and family who followed the build with curiosity and encouragement. And to you, thank you for following along.
Surgeon 25 is now part of the Clear Echo chain.
If you’re working on a record and want to hear what it can do in practice, or if you have questions about the build, measurements, or troubleshooting process, feel free to reach out.
Surgeon 25 is finished.
And my journey continues.
Comments