A strange LSB demodulation problem on my Yupiteru MVT-9000

I bought my Yupiteru MVT-9000 from an on-line auction at what seemed quite a fair price. I was a former nostalgic owner of the MVT-7100 model, that I had sold when I thought I wasn't interested anymore in using a scanner receiver.
The radio was in very good functional and cosmetic conditions, but I soon realized that it was affected from a strange LSB demodulation problem, that I documented in this short video:

At a first impression, it seemed that the unit worked in USB mode even when the LSB mode was selected.
Some researching on the web and questions posted on specialized forums produced no solutions in terms of unit configuration, reset, etc.. Also, it was clearly not a common issue for the radio: no trace of a similar problem experienced by others.
Finally, I decided to inspect the schematic diagram. Quickly enough I found the area of the circuit where the BFO was located. It was apparent that the selection between LSB and USB was performed by slightly modifying the output frequency of the BFO, as described in diagram below:

The BFO section of Yupiteru MVT-9000

There was two control signals from the CPU that seemed to play an important role: first, CARSEL (CARrier SELect) was in charge of adding/removing an additional capacitor in parallel with the crystal X3, thus causing the BFO frequency to switch between the USB value (lower) and the LSB value (higher). In fact, the BFO of the MVT-9000 is actually a VXO, with two selectable output frequencies depending on CARSEL value.

Second, the SSB/CW control signal from the CPU (together with signals AM/W-FM and N-FM, which have a similar function) was used to switch the input of the AF (audio) amplifier onto the proper demodulation path and bandpass filter. When SSB/CW is active, the audio amplifier takes its input from the mixer stage (transistor Q12 in previous image), which combines the BFO signal with the received RF spectrum (that is the output from the 455 kHz IF filter).

Well, it was time to dismount the radio and to inspect things directly during its behaviour.

Disassemblyng the Yupiteru MVT-9000 was fairly easy (below some photograps taken during the process).

Then I gained access to some important measurement points:

The measurement of the DC levels of control signals CARSEL and SSB/CW in modes AM, LSB, USB showed that they was working as expected.
On the other hand, it seemed that the BFO output frequency was changing only by a very small quantity when the operating mode was switched between USB and LSB. Respectively, 453.65 kHz in USB mode (which is a pretty good value, that is about 1.5 kHz below the center frequency of the IF filter) and 453.77 kHz in LSB mode, which is wrong. A value of about 456.5 kHz would be expected (about 1.5 kHz above the IF frequency).

Even if I was measuring the BFO output frequency with a tiny, very old and cheap frequency counter (a little SOAR FC-841), the results matched well with the observed erratic behaviour of the receiver.

Well, so it resulted that the VXO was actually switching between two frequencies, but one of them was out of specs. This seemed to suggest that one at least of the trimmer capacitors (TC2 and TC3), in parallel with the X3 crystal in the BFO circuit, needed to be realigned.
So I needed to furtherly dismantle the radio to reach the involved devices.

In the image below, you can see the other side of the printed board which hosts the BFO circuit, with trimmer capacitors TC2 and TC3 and the test point TP1 where to check the VXO output frequency. Also visible is the Murata CFJ455K device, which is the IF filter.

Exactly when I was ready to try a realignment, my old SOAR FC-841 frequency counter decided to stop working (to be honest, I broke it accidentally, by connecting a power supply with inverted polarity). So I had to buy a new frequency counter.

When it arrived, some days later, I re-checked the BFO frequencies in LSB and USB and (surprisingly) they both looked good! I was measuring about 456.55 kHz in LSB mode and about 453.67 kHz in USB mode, as recorded in the short video below.

I really couldn't explain what happened. After some additional verification (like connecting a temporary antenna to verify that the LSB demodulation was now acceptably clear), I decided to reassemble the radio and to make a live test. This confirmed that LSB was now a properly working mode on my Yupiteru MVT-9000 scanner, as documented in the short video below: