Well, yes, I think it is something to do with that, but the delay must be quite large for it to cause a flicker. Normally we should get all that I2C stuff over in less than a millisecond, so it should not hold things up. I'm going to have to do some tests.
OK, now a real problem:
When switched on, the clock emits a high pitch sound which increases in frequency. During this phase, the tubes start to light with increasing brightness. However, after maybe 2-3 seconds, the sound stops, everything goes dark (also the LED in the power-regulator stage) and after another second everything starts all over again with the high pitch sound. MOSFET and 100µH inductor get hot after a while in this mode. I think the wall-power supply goes into over-current protection and switches off (specifications [email protected]!).
What I tried (and what didnt work):
-Disconnect everything but the button
-Switching power supplies
-Pulling tubes (that only prolongs the duration of above sequence)
-Replacing MOSFET (got very hot after a while, but was not the cause)
-Replacing capacitor C3 (220µF 35V) (only because it looked suspicious, but obviously was nothing)
Any ideas? Maybe the 100µH inductor? (Don't think that it would show up like this). Something with the switching frequency of the HVD not being high enough? What would that frequency be normally? I have a multimeter which is supposed to measure up to at least 200 kHz.
Last edit: 1 week 1 day ago by Phil_Mod3. Reason: Added wall-power supply specs. Added more speculative causes.
Hmm, is this a new controller? Did you set the fuses to use the 16MHz external clock? If not, it will be using the internal RC clock, and that won't provide the "pump" to get the high voltage up high enough.
-Controller: The current controller ran the clock for a few weeks. Additionally, I still have the original µC that I got from your kit, which I did not modify. The problem is almost the same, old or new controller. I think fuses are OK.
-Vcc is 5 V stable, as long as the external power supply does not switch off. When I remove the µC (no HV generation) and switch on, a stable 5V results (also the external supply doesnt switch off in this mode).
-Any ideas about how to measure the current from the external power supply in the least "invasive" way? I don't have an additional barrel plug, don't want to cut any cords.
-My multimeter thinks the HV drive runs with ~9kHz (but I am unsure if this is a result of a long duty cycle which is provided as HV target is not reached, doesn't get above 160V)
-What I did before the problem started: I soldered the ISCP header to the board. However, the clock ran fine until about 48h later.
-Currently I think that there is a problem with the crystal oscillator and the hot inductor/mosfet and to high power draw are only results of a long duty cycle?!
I guess you don't have a scope? Not sure how crazy the duty cycle needs to be to confuse a mutlimeter's freq counter. It's certainly a consideration though, especially since the multimeter may actually be expecting sine wave AC.
I wonder what HVS signal (Controller Pin 23) is doing while the HV is squealing. Incorrect feedback voltage is yet another thing that might make the HV generator run wild. I read approx. 2.2VDC at Pin 23 on mine - approx. because it wanders around +/- 0.1V as the display digits change.
I don't think it's real important to measure the input current. You've pretty well proven that excessive current is being drawn and that the HV generator is what is drawing it. I got a big bag full of barrel connectors, males and females, with screw terminals (eBay form China) and just use those to make up a current testing "fixture" when needed.
Look into it later when the dust is clearing off the crater.