VCAs are current devices (and these have a type number, so we're talking on the same subject). For now assuming, your audio-VCAs might be THAT2180 or THAT2181 in grade A, B or C.
Next step is checking the audio path for unity gain in bypass condition. Power down, pull your audio-VCAs out of their (hopefully fitted) socket and temporary jumper VCA sockets pin1/8 with a piece of wire, a paperclip or whatever else handy. Power on again, feed your left side Ch.input with a maybe +4dBu (1.228Vrms) sine signal at a frequency, your multimeter can handle (probably below 1kHz, 100Hz seems a safer bet), and measure voltage of your left Ch. audiosignal at input (between XLR-pins2/3). Same level should arrive at left side Ch.output between XLR-pins2/3 (but with inverted polarity because of the now missing audio-VCA).
If not, double check (22K) resistor values at input stage, (10K) at output inverter stage, the (27K) resistor in front of the audio-VCA and the (15K) feedback resistor in the I2V stage. With from schematic 15K/27K fitted, your output level will read a little (times 1.111 or +0.915dB) higher. If this buggers you, a 5K trimmer/rheostat in series with this 27K, trimmers initial value 3K, would give you the oportunity to tweak this for same i/o level. Keeping the 27K and substituting the 15K with 2* 27K in parallel (=13K5) would have the same effect without the trimming ability.
Repeat this procedure for the right side Ch. with same signal source (just switch the XLRs over) for same readout.
Power down, pull out the temporary jumpers and refit the pulled VCAs with their correct orientation and power on again.
This already might have fixed your level differences. If not, report back with a little more info (type of VCAs used, value of fitted SSCF input resistors, Aarhus/Oxford switch wiring, ...)
