The filter was designed to have a centre frequency of 9Mhz, but these things are getting on a bit now and may have drifted a bit one way or another.  After you have adjusted the carrier oscillators and peaked the IF, take a look at the 'S' meter.  If it shows a reading with no antenna connected, flip the sideband switch and see if the reading changes.  One I was working on recently, showed S5 on normal and S0 on reverse even though the carriers were spot-on.  This is the result of the IF filter no longer being exactly 9mhz, so a few tweaks are needed...  The best approach here is to sweep the filter and find its exact curve, then shift the carrier oscillators accordingly so that they are just off the skirt of the curve.  But, if you can't sweep the IF, you could try the following method..

  Set the bandswitch to 7Mhz, and the SSB switch to normal.  Adjust the 8998.5 carrier trimmer to a point just BEFORE the S meter starts to rise.  Flip the sideband switch to reverse and adjust the 9001.5 carrier trimmer to just BEFORE the S meter rises.  This puts the carriers right on the base of the filter curve, wherever it has drifted too!!  If you now tune to one of the calibrator markers and tune for a pitch of about 1.5kHz (S meter peak) the pitch should be very close to the same when the sideband switch is flipped back and forth.

This set of adjustments has to be done carefully and slowly, as more than one frequency WILL pop up in the output.  With the rig turned off, pull V1 (6U8) and also pull V106 (12AU7A).  You only need V1 out, but this will cook V106 with the series/parallel heaters (V101 heater cooks V106 if it's parallel (V1) is missing.. got it??).  Turn the power back on and allow things to stabilise again.  Hook a CRO together with the counter (Spec-an would be nice here..) and probe it into V1 pin 9 which is the RX mixer injection.  Set the bandswitch for 3.5 and dial up '0' on the VFO BLACK scale.  You should see a healthy carrier on 5.5Mhz.  Now set 7Mhz band and look for 16.5Mhz, adjusting L5 for maximum. If the frequency is a bit off, tweak the heterodyne osc TC201.  I found that I had to tweak L202 a bit as well to get a signal.  Skip the 14Mhz band (check for 5.5Mhz again if you wish) and go to 21Mhz.  Now adjust L4 for maximum at 30.5Mhz. (this one IS touchy and goes everywhere).  Re-check all results again and touch-up if needed.  Now go to 28Mhz band, and set 28.5Mhz.  You guessed it.. adjust L3 and maybe tweak L202 again and go back to re-check..  about 6 laps through the settings ought to do it, checking and tweaking.  If you have the other crystals installed, use the trimmer caps on the crystals to set the right frequency.  You shouldn't need to touch L3 again.  Also beware that L4 & L5 have 2 slugs... one topside and one lowerside of the formers.

Heterodyne Oscillators.

  Place the bandswitch in 28.5Mhz position and hook the CRO to V3, pin-2 (6CB6 Cathode).  Adjust L202 for Maximum signal.  Now connect the counter to the same spot (V3 pin-2) and adjust TC204 for 43.0Mhz.  Select 7Mhz and adjust TC201 for 11.0Mhz.  Select 21Mhz and adjust TC202 for 35.5Mhz.  If you have the optional crystals fitted for 10M band, select 28Mhz and adjust TC203 for 42.5Mhz, then 29.0, adjusting TC205 for 43.5 and 29.5, TC206 for 44.0Mhz.  You may have to give L202 a slight tweak to get the best signal across all frequency settings, but it shouldn't need that much tweaking.

VFO..

  The output of the VFO is 5 -> 5.5Mhz aligning with 0-500 on the RED scale.  Hook a counter onto the output at the filter board and make sure the heterodyne oscillator is off by selecting 20M band.  Adjust TC401 for correct frequency.  There is a piston type trimmer on the VFO also, but this is for the temperature compensation, so only adjust this if the VFO drifts as it heats up.  If this is tweaked, TC401 will require adjustment also to put it back on frequency. The VFO should only be adjusted if all covers are in place as this will affect the stray capacitances and hence the operating frequency.  Check the tracking with the dial and if this requires adjustment, the rear rotor plate is slotted into chunks and these will require CAREFUL bending to get it linear. TAKE YOUR TIME as over-tweaking gets frustrating and can lead to damage!.  Hook the CRO onto pin-1 of V3 (6CB6 Grid-1) with a x10 probe and make sure the VFO injection is >0.5v when swept end to end of VFO travel. 

The Carrier oscillators..

  Locate tube V106 (Carrier Osc) and wind a piece of insulated wire about 6" long around the tube envelope.  This is a convenient pick-up for the counter without loading the circuit and possibly throwing the tuning out.  Connect the counter probe to the bared end of this wire and stoke up the rig. Set the controls for normal sideband, 20M, SSB receive and leave it run for about 1/2 hour to warm up.  Measure the frequency of the carrier and adjust TC102 for a reading of 8998.5Mhz, then flip the sideband switch to reverse and adjust TC101 for 9001.5Mhz.  You should be able to get it spot-on and it should be stable.  If not, try another tube (V106 and let it warm up again) then check the components around V106 for burned resistors, leaky caps, junk trimmers etc.  Last resort would be to replace the crystal.  Whist looking at the 9001.5, set the function switch to AM and make sure the frequency shifts close to 9.000Mhz.  Make sure it returns properly when SSB selected again.  Don't be too fussy as they may have to be adjusted a little again later to compensate for a filter that's not exactly 9.000 MHz centre frequency.