OK, my new theory is that the 6922 is not performing at full emission - either it's a bad tube, or it is not yet fully broken in (the cathode is not fully "formed"). Sometimes it takes several tens of hours to get the cathode working properly - it does seem to be a problem with modern-production tubes. I came to this conclusion after I reviewed my detailed analysis. But having spent some time working it out, I've left the analysis below. You can let it run and check voltages as suggested below, or try another 6922 if you have one. Or ask Eileen for another tube.
Notice below that I originally concluded the tubes are OK, forgetting that the servo bias will force an under-performing tube to get the right voltages, but it may only happen when the bias is too small - which leads to the 6922 grid loading down the RIAA network, giving loss of gain.
======== here's the original analysis ==========
Yes, you may lose the 2200uF cap. But the alternative at this point is to build a new board.
Since you have swapped the boards, we know that all the tubes are OK, hence the problem must be on the board. (I am assuming all the connections are done correctly, because you have been careful about checking and re-checking them. There is always a chance that something has been consistently missed, but I can't do much with that possibility so I won't try!)
So, the logic I have is the following:
* Inititially the shunt reg board LEDs are out because all the transistor bias current is going through the base-emitter junctions, which have a lower voltage than the LEDs. Once the shunt reg tube cathode starts emitting (around that 10 second point), then the shunt reg tube draws enough current to allow the LEDs to light. This drops the output voltage to the regulated 225 volts.
* Initially the applied voltage to the audio circuit is higher, say 350v. That's enough to allow both C4Ss on the servo boards to operate, since they have resistors from plate to ground.
* When the shunt regulator kicks in, it drops the voltage to 225v, which is no longer enough to drive the full 7.4mA until the 6922 cathode starts emitting. This could cause the LEDs on that side (the A side) to drop out until the 6922 finishes heating up.
* If the 6922 cathode (C3 or C8) is not drawing current, then the servo transistor 2N2222 does not see any base voltage, and is an open circuit. When the EF86 cathode starts emitting, its current charges the 2200uF capacitor. This creates a bias voltage that shuts off the EF86, allowing its plate voltage to be high. Once the 6922 conducts, that high voltage at the grid of the 6922 will cause it to conduct current through the 27K and 174 ohm resistors, providing bias to the base of the 2N2222 which will pull down the voltage on the 2200uF capacitor.
So, some speculations - perhaps the 6922 heats up slowly on this side, and if you let it run, it will stabilize at more normal voltages. Or, perhaps the servo feedback resistors (27K and 174 ohms) are not providing the necessary continuous path - but since the steady state voltages were correct before the board swap, we know it was working correctly before. That leaves a possible error in connecting to the tube at C3 or C8 (not sure which one goes with which channel, but C1-C2-C3 goes with one board, C6-C7-C8 with the other board.) You should be able to measure 27K from that pin to ground.
That's why I suggested letting it go - if the voltages stabilize correctly and the LEDs come back on, then the slow-heating theory is correct. If not, then something that was working is now broken, and the voltages will give us a clue. I have given a detailed reasoning in hopes that it might stimulate you or another Forum denizen some idea which has not yet occurred to either of us - this is a tough one.
For what it's worth, on my own Eros the servo board LEDs come on immediately and never waver in intensity; the regulator board LEDs come on at about 10 seconds and remain on thereafter. I have an EH 6922, which I think is still the stock tube. So there is something different with your Eros.
I have said slow-heating, but it may be the cathode is not well formed yet - sometimes it needs several tens of hours before a new tube is emitting at full performance.
