Servo problem with J99s

[SoundBase]

So I have four J99 modules that I bought and started a while ago, and am finishing now. They are version 3.1 of the circuit board, but the parts list and schematic I received at the time were listed as version 1. At the time Tim sold them as as circuit boards and I was on my own to source the parts, which I did mostly from Mouser and Digikey. I had built all of them at the time except I only had one set of discrete op amps, so I tested them with the opamps I had then and they seemed ok, but I didn't rigorously check the servos (which was the last step) of the instructions then. I recently bought more opamps and I am now finishing them to get them into my chassis 01 with PS02. I'm paying more attention to the servo readings now and that's why I am posting this problem.

Since I was on my own to source parts I have a couple of differences from the parts. I am using Millenia Media MM-990 op amps (serial numbers 713/715, 2327/23236, 2328/2329, and 2330/2331; U1 and U2 respectively) and the Jensen JT-11-DM output transformer which was spec'ed in my parts list then but now I think Tim suggests the Cinemag equivalent.

The problem I'm seeing is different voltages between pin 4 and 6 of U2 at different gain setting and with different opamps in place. This often takes more than 30 sec for the servos to stabilize, sometimes up to 2-3 minutes, but even when it doesn't settle the servo will still be dropping 0.001 volts/second, I stop looking after 2-3 minutes.

For unit 1:
opamps 713/715-min gain servo is at -1.35 v after 3 minutes, at intermediate gain (8 clicks): -5.1V and max gain -3 mV (so the servo works for max gain but not so well for less than that.)
With opamps 2327/2326- min gain -6.8V, intermediate -6.8, and max +4 mV (so only works for max)
For opamps 2328/2329- min +19 mV, intermediate 1.7V and max 4 mV (so it works at min and max but not in the middle)
For opamps 2330/2331- min +32 mV inter -5.1V max -5.5V (so it works at min)

For unit 2:
opamps 713/715 min -1.2V inter -3.7V max +4 mV (so works for max)
opamps 2327/2326 min -1.2V inter -3.5 V max -1mV (works for max)
opamps 2328/2329 min +8.4 mV inter +2.9V max -2 mV (doesn't work for inter)
opamps 2330/2331 min -1.2V inter -3.2V max -3.4 (doesn't seem to work at all, but is dropping very slowly)

For unit 3:
opamps 713/715 min -1.2V inter -3.8V max 1mV (works for max)
opamps 2327/2326 min -1.2 inter -3.4 max -2 mV (works for max)
opamps 2328/2329 min 7 mV inter 3.3V max 1mV (doesn't work for inter)
opamps 2330/2331 min 30 mV inter -2.6 max -3.2V (works for min)

For unit 4:
opamps 713/715 min -1.2V inter -3.8 mx -2.6 (doesn't seem to work at all)
opamps 2327/2326 min -1.0V inter -3.5V max -4 mV (works for max)
opamps 2328/2329 min 2mV inter 3.4V max 2 mV (doesn't work for inter)
opamps 2330/2331 min 20 mV inter -2.4V max -2.9V (works for min)


So as you can see when I wasn't really paying attention to the gain before I thought the servo was working for opamps 713/715, but now that I bought the additional six opamps and am paying more attention to the min, 8 click intermediate, and max gain, it doesn't always work. The 8 click spot seems to always have the highest readings.

After I first observed this thinking this was some cold solder joints somewhere, I reheated all the solder joints in the circuit for the resistors, capacitors, and press fit sockets and the pins for the LT1013CP socketed U3 (yes I socketed the U3) and U3 voltage regulators for boards 1 and 2. I did not do switch 1 (greyhill switch, but maybe I need to ?) And I have not done this for boards 3 and 4 yet. Do I need to or there some other issue?

After googling servo in other audio circuits some people have suggested different U3 for the servo so I used an OPA2604 that I had left over from my c84s, but it did not make any difference the voltage reading as were still the same. I looked at the linear technologies website and they have dual channel opmp with the same pin configuration that has very low offset voltage (5 uV), the LT 1151, so should I try that? Granted they are $11 at digikey, but I'll pay that if that's what it takes.

When I looked at the more recent version of the J99 (version 3.6, I believe) Tim has incorporated a way to trim the voltage supply so as to minimize the DC offset voltage for each U1 and U2, but this is not incorporated into the version 3.1 I now have.

So maybe this problem with all the servo is just the drift of the MM990s and do I need to figure out different pairing of the eight I have that will work?
Or is there some small board I could make and install above where the U1 and U2 are now that would incorporate the trimming feature that Tim now uses? But my gain setting resistors are different than what is now used in the board, so would it work?

The Jensen is not a gapped transformers and probably shouldn't have DC on the primary like this, as it can't be good for the transformer and reduces the head room of the overall circuit.

Is there something else I'm totally missing here or additional tests I need to make to diagnose the problem?

Thanks for any and all feedback

[SoundBase]

This past weekend I took board 1 and reheated all the solder connections at SW1 thinking I had a cold solder joint somewhere (even though I hadn't seen any). Unfortunately it did not make much difference the readings are the pretty much the same as I posted before. I will do the same for the remaining three boards and see if there is any difference.