Ok, thanks for getting back with me on this...
Since this is specific - that helps.
In my repairs, I've found 4 different places that each one interacts with each other and can generate pretty much the same symptoms - two deal with the PLL side of IF and the other two deal with Audio Side.
So to keep it short and sweet - you are given a spot or two to work with and use various values, but still there are reasons otherwise for the offset problems.
(Disclaimer : Graphics used from CB Tricks and featured radios from their respective companies)
The IF from PLL and Clarifier sides of a Cobra 146 ...
PC 122XLand Pro 810e ... But notice the changes to values C96, C97 and C98 used between the two...
You can't really tell the differences between the Pro810e and the PC-122 XL - because both radios used that same chassis - but look harder at the Oscillator parts values used.
Then an oldie but a goodie - TRC-453...note also its' use of cap values in the Clarifier section...
Why the difference, they are subtle changes in the tuning ranging from foil pad thickness onto the coil winds and their position of coil winds in the core of the can. Even the oscillators themselves use various cap values to achieve a level of drive or needed signal quality.
So in each of the above, you can make changes to one mode to change the tuning and where the cores' inductance lands for your passband for audio. I do this to help the operator find "center slot" easier and in some cases for a larger level of clarifier Zero beat - as in once a heterodyne meets the audio signal coming in off of the IF the PLL uses for conversion - the slide itself tends to resonate and offer a better bandwidth where the receiver hears the audio with more fidelity for a greater length of slide allowing the "roundtable" to occur and the operator doesn't need to chase after all of them to listen in.
But why the above? Well it also comes from how the IF the PLL uses - passes the images thru the Crystal Filter and it's number of poles (Xtal elements) and their exactness to resonate a narrow range and only pass a specific range of frequencies. (Photos courtesy of Low_Boy)
The above is a 4 pole unit, and note the use of 4 crystals and series of caps to pare down, narrow and sharpen the bandpass of whatever was being sent into it - note too - the 10.4MHz - that's the "center" peak of the filters' passband.
The "colors" in the peak represent the offset tonal effects of the center IF carrier injection - Grey vertical line - while the pink is the noise and signal EMBEDDED in the image received. Red to Blue signifies the tonal range - Red Bass onto treble high-tones in blue - with most voice range in yellow for speech.
So, if all things are equal - why the tonal change?
Several things come into play, one being the TOLERANCE of the Xtals used. Not all crystals can be made to EXACTLY the same resonate frequency - so with the above in mind, you now have an issue of 4 times the poles and their own resonance affecting the signal. This increases the BANDWIDTH the filter unit now has, plus any overtone shift in frequency passage - lower or higher than expected. That means even the IF carrier is going to mix in with the audio signal generating a carrier and an artifact during the conversion process.
So one mode give you nice sharp tones and clarity, but the other mode is slightly different - almost muddy sound - then that indicates the filter - although sharp - may not be narrow enough to remove any artifacts including it's own IF carrier image generated so the radio can select the signal out of all that noise.
So therefore - you have some choices to make changes and allow the IF "centering" to be shifted to accommodate the less than ideal tonal bandpass for reception.
I hope this helps...
Mostly - due to the artifact, it will require experimentation of your part to find the right balance. They kinda' did the same thing back where the radio was made. At least IMO it's why they have parts like C100 - C102 in the 146 board, onto the C96, C97, C98 mess in others. In my research and repairs, these areas were the most likely to show values not corresponding to the schematic - but to variations in values to accommodate the parts and years these boards were made....