Crystal failure, and the new ones are resonating at 1/3 or their marking

[Sassa]

Hi.
I am repairing a Korg piano from the late 1990s : https://ia600204.us.archive.org/1/items/korg_CONCERT-XC1000-XC2000-XC3000IF_SERVICE_MANUAL/CONCERT-XC1000-XC2000-XC3000IF_SERVICE_MANUAL_text.pdf

I found that X1 the 32MHz crystal of the NEC V55 CPU was not oscillating, both terminals were flat at around 2.5V.
Since I don't have a spare part of the same frequency, I replaced it with a 27.145 then a 33, but in both cases the crystal oscillates at 1/3 of what it should be !

There is another 32MHz crystal on the board that I measure is working at the correct frequency, so I believe it is not a problem with my scope.
I tried probing with a 1k resistor in series with the probe with no effect.
The V55 also has a CLKOUT pin that is half the oscillator frequency so I believe the problem is not caused by my probing on the crystal terminals.
I changed the coupling caps but the problem remains. They are on the other side of the board, I read it is bad practice to connect them through vias...
There is a 2.2k resistor between the X2 terminal of the V55 and the crystal. It measures OK.

Am I missing something or could it be a problem with the V55 itself ? Or the replacement crystals that I used need different caps and resistor value ?

[wraper]

Those crystals must be rated to operate at 3rd harmonic but were running on 1st harmonic instead.

[TimFox]

Quartz crystals have overtones that are not harmonic:  i.e., close to odd-integer multiples of the fundamental (3rd, 5th, etc.) but not exactly an integer multiple.
This is because they are three-dimensional resonators, as opposed to violin strings or organ pipes that are one-dimensional with harmonic overtones.
Some crystals are calibrated to operate at either the 3rd or 5th overtone (in an appropriate oscillator that works at that frequency).
Such a crystal can oscillate at its fundamental frequency (in an appropriate oscillator for fundamental mode, either series or parallel resonance), which will be approximately 1/3 or 1/5 the calibration frequency.
For fundamental-mode calibration, that can be either for series-resonant or for parallel-resonant mode with a specified parallel capacitance.
Overtone-mode calibration is for the series-resonance at that overtone.

[Sassa]

 
You are great.
I picked a 25MHz one that is running fine.
Thank you for your explanation, I learned something new today and that is worth the effort.

[Kleinstein]

Getting the lower frequency is likely from having crystalls made for a higher tone. So the filter network may be failing (e.g. drifted away from resonance).   Getting 33 MHz fundamental requires a rather thin crystal that was difficult to make in the early days. So they often made frequencies higher than some 20 MHz as overtones.  AFAIK there is a slight difference in the marking of the frequency for the overtone crystals ( like using MHz vs.  kHz).

Today there are also some crystals around to get higher frequencies as base frequency, though maybe with a reduced Q.