1A/1B reading 100v [resolved]

[Paul Birkeland]

I would be sure that the 10,000uF caps up front and the 100uF caps on the front PC board are well soldered, as a loose joint on one of those will leave you with perfect DC voltages but a large difference in gain.

[Brooklynbottle]

Thank you.
I can't imagine what it must be like diagnosing the myriad difficulties created by an army of amateur assemblers of the Bottlehead line of products!
I'll fire up the iron.

[Brooklynbottle]

I hate our diagnostic session has hit the second page.
I re-flowed the 100uF caps and removed, swapped sides and re-soldered the 10,000uF caps. No improvement on the R side gain.
Full fidelity, just an annoying deficit in comparative volume.
Any other suspicions?

[Brooklynbottle]

Reading through @caderette, perhaps my 6922 is the source the channel imbalance? Switching the 86's didn't change the discrepancy.

[Paul Joppa]

Ten dB is about 70% less voltage, not 20%. The gain of the two triodes in a 6922 might possibly vary by as much as 20%, but not 70%!

I see two choices - either gather more data (measure gain at  several frequencies), or just try refreshing all the solder joints. Personally I'd start with those on the eq network tag strip.

[Brooklynbottle]

Fair enough. I'm sure there has to be fault in my soldering somewhere!

[Brooklynbottle]

Quick clarification. Is there a concern that solder flow from an upper terminal strip position to lower position may cause problems? I figure they occupy the same terminal, but I want to make sure excess flow isn't contributing to my channel imbalance in the EQ circuit.

[Paul Birkeland]

Quick clarification. Is there a concern that solder flow from an upper terminal strip position to lower position may cause problems?

No, no concerns with that.

[Brooklynbottle]

Well I took the C4 board off. Re-soldered the whole thing again, touched up my t strips on the R side. Checked voltage, fired it up and same diminished signal on the R.
Took the C4 board off again l, touched up the pad and wanted to share the current situation before I put everything back together for another test. Perhaps there's something I missed?

[Paul Birkeland]

I'd be a little suspicious of either lead free solder or a bit of a cool iron on that one.

[Brooklynbottle]

Reflowed C4s board. Touched up the t strips. Re-assembled, checked resistance and voltage. Everything in spec. I also checked continuity of all the wiring to the board which was also normal. 
Plugged the Eros in and I'm closer. I've got a 7-10dB deficit in the R channel based on my crude sound meter app. This is an "improvement" from the 20dB difference as measured by the same phone/app.
I'm apprehensive about burning up my board at this point.
I'm planning on leaving the Eros powered up for a while in hopes I can get things evened up.
I hate to admit defeat, but I think fully correcting this imbalance may be beyond my capabilities.

[Brooklynbottle]

Update on an update. Soaked the Eros for 24 hours.
I used the averaging function on my sound level meter and we're at a 4dB deficit on the R channel.
I haven't re-touched the front, A/B board. Is that a potential source for imbalance?
BTW using standard Kester solder w/ the iron at max.

[Paul Joppa]

Normally there is a large variation to acoustic measurements, due to room acoustics. So I have to ask, how exactly are you measuring this 4dB difference?

[Brooklynbottle]

Fair question. Just trying to quantify what my ears tell me when listening.
I'm just using my phone app while capturing a dB average reading at approximately 12" from the face of the speaker.
I know it's far from precise, but I was hoping the information would aid in the diagnostic process.

[Brooklynbottle]

Final dispatch. My Eros is functioning as intended! Channels are balanced (as confirmed by my crude app based sound metering) My vinyl has never sounded better. I can only postulate that I just needed to sit down and spend an evening using the Eros for everything to even up. Thank you for the support and urging to double check my work. Consider the case "resolved".