I’ve seen this before, and the problem was related to the inductor. Try getting 2.2mH inductors from a different source, or ,as a quick dirty hack, put two 2.2mH inductors of the kind you have now in parallel.

We might get over this by changing the frequency the inverter operates at – drop me a mail so I can send you a test firmware that uses a lower frequency (that might get noisy).

We tested a lot of transistors from different sources. The best Voltage/Duty ratio, we got it with the transistor (T2)
http://www.reichelt.de/BF-299/3/index.html?ACTION=3&LA=446&ARTICLE=5464&artnr=BF+299&SEARCH=bf299
BF 299 :: Transistor HF NPN TO-92 300V 0,1A 0,625W
and
http://www.reichelt.de/L-XHBCC-2-2M/3/index.html?ACTION=3&LA=446&ARTICLE=138556&artnr=L-XHBCC+2%2C2M&SEARCH=2.2mh
L-XHBCC 2,2M :: Festinduktivität, axial, XHBCC, Ferrit, 2,2 mH

best regards.

Tuxik, could you please elaborate on the problem? Ițm having the same situation here (15V on LCD, 8V measured across the tube).
Thank you!

Hi Wolferl!

My mistake… I meant the internal voltage measurement is 15V, as it is shown on the LCD.
The LCD is powered at 3V (2xAA batteries).
I have replaced the small axial inductance (2.2 mH, 32 ohms) with a larger, ferrite-core one (2.2 mH, 3 ohms),
but the voltage remained the same.

Hi,

You can’t measure at M3 directly. It has a very high impedance, a multimeter or an oscilloscope probe loads the voltage down so much. You would need a voltmeter with a input resistance of 100 MOhms at least to get meaningful results. And you’d need to measure not at M3 (anode) but on the cathode of the diode. You can instead use scope or multimeter across R4 and multiply the reading by 214 to get the real high voltage value.
That M1 measurement looks weird. It has only 2,7 volts peak-peak. That should be more like 3.3 to 3.6 volts. Frequency looks good.
What you can try: with the KIT1 turned off, check the resistance of the connection R3 to R3 (the 2 10Mohms resistors) to ground. It must read 10 MOhms.

I hope I can find some time for your firmware tomorrow.

Cheers,
Wolferl

Hi folks,

Since I had my KIT1 apart, I took some measurements with the scope for comparison.
The yellow curve (lower trace, 2 volts per division vertical, 20µs per division horizontal) is the PWM output of the MCU. We see it is running at about 50% duty cycle, which is good.
The green curve (upper trace, 1 volt per division) is probed at the cathode of diode D7. We see some ringing when the current in the inductor is turned off.
That’s normal. HV is 380 volts (displayed).
If you have no ringing, you have no high voltage spikes. Loading the collector of the MPSA42 with a 10Mohms probe completely kills the HV (duty goes up to 90% which is the maximum).
If you probe there, be careful, there are 200 volts peaks present! Do not kill your scope 🙂

I have the original parts for transistors and inductors as the kit came from factory.

The only thing I modified are:
R3 is 47Mohms instead of 10MOhms
R4 is 220kohms instead of 47kohms
That yields the same divider ration of 213.
These modifications creates less load on the high voltage and reduced current consumption dramatically, plus reduces stress on the parts.

Cheers,
Wolferl