http://auroraaudio.net/forums/topic/32264-calibration#post-6601 (See: blakeyboy) 33609 is the same principle.
" Put in +20, trim gain ( lower rear trim) so same out as in irrespective of bypass on or not.
Select 'limit', set to +8 ( count clicks from fully anticlockwise in case knob is on wrong).
Trim top rear trimmer for +8 out.
Keep in this mode, and trim small pot on pcb so meter reads '12'.
That's that.
Now. Check compression. Keep +20 in, compress only in, no gain make up, threshold set to '0'.( count clicks down from+10)
Set compress ratio to '2' and see how near to +10 is coming out. Also set to ratio '4' and see how close to +5 is coming out. This will tell you whether you have a pair that will match easily.
If they are very different,and 2-3dB is not uncommon,the only thing you can do is to check diodes. Check you have the same ones for the diode bridge or the same to5 can bridge in each unit.
The diodes on the rest of the board can be different too. This is where the differences creep in. I found this the hard way- trying to get a pair to match. For more hours than I care top remember. If you have a problem, by some diodes on a bandiolier (easier to handle), 1N914 or 1N4148 etc, leave them on the bench and DONT touch them. Let their temperature settle. Measure the .6v forward voltage accurately, and sort them into sets that have the same voltage- ie .596v or .587v etc. This will take ages. I lay the bandiolier over a sheet of A4 and write the voltage next to the diodes, only touching them with the meter probe tips.
Touching the diodes will warm them and change their voltage _radically_. Once you have a pile that measure the same that is big enough to do both units, fit them! You will have to do the bridges if they are still a bit wrong."