5W PLL problems with oscillator frequency and tuning voltage

[Mek]

Hello, I built the 5W PLL transmitter on a double-sided PCB.
I used all components as described on the page, no replacements. TSA5511, 2SC1971, BBY40, BFG135. Crystal is 3.2 MHz and appropriate hex file is loaded into the PIC.
The problem is I cannot get the PLL to tune to the specified frequency.
When I connect input voltage, tuning voltage is immediately 8.5 V and stays there (frequency on display is 87.5 MHz). The LED indicating "PLL lock" briefly flashes, stays off for a few seconds, and then lights up and stays lit up - saying "PLL is locked". But it is not. Oscillator frequency measured on emitter or collector of Q1 (BF240) is around 105 MHz, and is not always the same, slowly drifts between 80 and 110 MHz.
Q3 heats up normally, R8 and Q2 are very hot to the touch, and lamps connected to HF output do not light up.
Frequency measured between R5 and C20 is less than half of oscillator frequency, and on RFIN1 input of TSA1155 it is around 2-6 MHz (also not stable). Same goes for base of Q2. I think I can't measure frequency with my frequency counter as it causes the frequency to drift but that is only my theory.
What I tried (the list is not exhaustive):
- check for unintended shorts on PCB
- disconnect Q3
- change TSA1155
- change BFG135
- change BBY40 for KB109G
- change BF240
- change C30 (10n) for 1n as recommended in TSA1155 datasheet of TSA1155
- change C36 (27p) for 15p
- change crystal
- change resistors for green HF ones from Tesla
- rewind the transformer using a different core
- use circuit with recommended values from TSA1155 datasheet (page 12) around pin 1
- disconnect C8
- disconnect R10
- change SMD capacitors for non-SMD ones with legs as short as possible
- various combinations of the above
I always tested tuning voltage on Q4 collector as per the instructions, and measured real oscillator frequency on Q1 emitter.
During my experiments I had varying results, sometimes tuning frequency was around 4 V, changing with frequency, but in other than expected way - higher frequency corresponded to lower voltage and vice versa.
Sometimes tuning voltage was zero, briefly popping up to 0.6 - 1 V when connecting power supply, then it was zero.
It was never possible to tune to 107,9 Mhz and then adjust L1 for tuning voltage of 7.5 V, as given in the instructions.
Real oscillator frequency on Q1 emitter corresponded to what I was able to tune on my radio receiver. There were various whistling noises.
Most of the experiments I did were without Q3, I wanted to make sure that my PLL works before connecting it. Is the circuit testable without Q3 or it must be present?
What is the role of the circuit around pin 1 of TSA1155? It isn't explained in the instructions, nor the datasheet, and it's unclear why there are components of different values than given by TSA1155 recommended values.
Measuring frequency on the crystal gives me nonsense values like 0-4.5 MHz, but I'm not sure I can measure if the TSA1155 is running at all. But since the LED lights on after a few seconds, I suppose it is. Also when frequency is changed via buttons, the LED turns off and then back on after a few seconds, just like it is supposed to, but oscillator frequency does not correspond to what was set.
I have three TSA1155 chips from various vendors and they all behave the same, so they are real (or all fakes which is not probable). With SDA3302 (not 3202) the behavior was different but still not working correctly.
I'm stuck for a few days already and would really like to get this finished, please help.

[Mek]

I changed all smd 10n capacitors as I found that C15 was bad.
With SDA3302, the PLL LED indicator lights up after few seconds of tuning, but tuning voltage is still the same - 8.85 V.
With TSA1155, the PLL LED never lights up and tuning voltage is 0 V most of the time, but sometimes around 4 V, no matter the frequency.
In both instances, LED briefly flashes when power is applied. The 24 V lamps on output do not shine so clearly the output is not of correct frequency.

Today I did an experiment:
I lifted one leg of C8 (the one connected to oscillator Q1), and applied FM output from a small cheap FM transmitter there. I also removed Q1 so that there are no oscillations when that oscillator is now unused.
With this setup, it is working. The 24 V lamps on the output shine and I was able to tune the variable capacitor according to the instructions. This circuit takes around 0.8 A from a 12 V lab power supply. So the power part is OK. (Q2 and Q3).
I think I can live with that. I understand that those tiny FM transmitters for personal use were not available back in the days when the web page of this project was written.
So nowadays you can use them as a source of signal to amplify, and get rid of those exotic TSA and PIC chips. Hope it helps someone.

[RIGMO]

If you want to  PLL lock base oscillation need to be a few mhz above  required broadcasting frequency.  example.. no PLL control oscillator need to be set-up  approx. 100mhz if you want lock 98 mHz spread coil od tight...
FB - k19radio