Self Starting Coil (Not)

[alan sailer]

There was a thread a while back that had a cool looking little coil,

https://highvoltageforum.net/index.php?topic=1519.0

It looked fun so I went out to build it. The circuit is the same as the original post with one difference, the gate drive chip is a UCC37321 (the chip used in the original post is hard to find and I had some other gate drivers).



The circuit does not want to start up. I changed the polarity of the primary which made no difference. I attached a signal generator to the
cap/diode clipper going to the gate drive chip and can get action that way. The coil works but no self start.

How coils get started has always been a mystery to me. But I never had to understand the mystery until now. I once saw a Steve Ward design (Micro SSTC) that used a starting switch. And I seem to rember another small coil that used a free running 555 timer coupled to the feedback antenna as a starting signal.

Any enlightenment?

Cheers.

[davekni]

Self-starting depends on the initial gate transition after enable causing enough voltage at the secondary top (top load) to couple into the antenna and cause a second transition.  The 1N5819 diode's leakage current will bias UCC37321 input pin 3 to roughly half supply, ~6V.  This is above the UCC37321's input threshold voltage of 1-2V (per spec).  Being inverting, the output (pin 4) will start low, the same as disabled state, so no initial transition to start oscillation.

The easiest (and I think most reliable) way to get self-starting is with a feedback resistor from output (pin 4) to input (pin 3).  100k to 1meg is likely to get things going.  To make it ideal, remove the 180V supply and adjust the feedback resistor until gate frequency roughly matches coil operating frequency.  Then turn 180V back on and it should self-start reliably.

You will still need to check both primary polarities to see which works properly after making the above change.

BTW, I think this circuit would be better with a 1:1 GDT wound with a twisted pair for minimal leakage inductance.  12V is sufficient to drive a MOSFET gate.  If you want a bit more drive, increase UCC37321 supply voltage from 12V to 15V.

[alan sailer]

David,

Thanks (again). I had noted your comment about the original driver having enough hysteresis so I checked that on the new part and it seemed to have decent margins. I missed the inverting/non-inverting. I have some UCC37322 so will try them out. If this fails I'll mess with the feedback idea.

After I get it running and probe the gate signals I'll see about you modification to the GDT.

As far as the transition, does the step function create a rich brew of frequencies, one of which is close enough to the secondary frequency to start the oscillation?

Cheers.

[davekni]

I missed the inverting/non-inverting. I have some UCC37322 so will try them out. If this fails I'll mess with the feedback idea.

Make sure to change back to an inverting driver if you try output-to-input feedback resistor solution.  Another option instead of changing driver chips is to bias the input below threshold, by a resistor across the lower 1N5819 or by changing the upper 1N5819 to some lower-leakage-current diode such as 1N4148 or 1N914 or ...

As far as the transition, does the step function create a rich brew of frequencies, one of which is close enough to the secondary frequency to start the oscillation?

I'd hadn't thought about it in those terms, but sounds accurate.  An ideal step function (V=0 for T<0, V=1 for T>0) has a continuous frequency spectrum with amplitude proportional to 1/frequency.  Secondary resonance will reject all frequencies except resonance.  If the remaining energy at resonant frequency is enough to cause a second driver transition, then oscillation starts.

BTW, the feedback resistor option improves two things.  First, it provides multiple edges to start oscillation.  That is useful if the feedback resistor open-loop frequency is somewhat close to secondary resonant frequency,  That way multiple open-loop edges reinforce secondary resonance.  The second advantage is independent of feedback open-loop frequency.  It is that the driver input pin is always close to threshold voltage, somewhere between rising and falling thresholds.  Thus even a single edge that creates a week feedback signal is more likely to be enough amplitude to create another transition.

[alan sailer]

David,

The change to the non-inverting was good. The circuit started right up. I may play with the feedback thing. Looking through some old Steve Ward designs I saw him using the same concept with a push button switch to start.

In the past all my drivers were various borrowed UD style and I only had to understand the circuit in vague terms.

Now it's time to figure out what I actually have. The drain voltage looks like nothing I have seen before. I also did
a rare thing (for me) and popped the IRFP460.

Thanks again.

Cheers.