Is feedback necessary?

[faraday]

Hi,
Is current feedback really necessary in a DRSSTC (or voltage FB in a SSTC)?  Or can it be run without any feedback by just hitting it with a fixed oscillator at the frequency that both secondary and primary are tuned to?

I understand that streamers will result in a lower secondary resonant frequency, and that feedback somewhat helps get around that.

And I get that having feedback provides the "kick" just at the right moment.  But, how important is it to have that kick just at the right moment (with FB) vs almost at the right moment (without FB)?

Will a coil run okay without it?

[TDAF]

The might run Ok(For an SSTC that is)
But, in a DRSSTC soft-switching is the reason why these IGBT's switch that kind of huge currents
Since, DRSSTCs are very dynamic systems
manual tuning just doesn't work
If the circuit is even slightly out of tune the results can be catastrophic as the IGBT's will then have to hard-switch

[Garyf]

My drsstc runs ok from an oscillator substituted for the feedback ct. Its useful to see how the currents and voltages behave as you adjust the frequency. But you may want to do this at a reduced bus voltage to avoid hard switching stress. I start with the oscillator set to the primary resonance and adjust a little up and down from there. It works if you don't get too adventurous.

[Mads Barnkob]

The spark is a very dynamic load, varying RC depending on branch size, this means that the secondary circuit is never operating at the same resonant frequency. So you might have the primary and secondary circuits tuned perfectly to each other, but as soon at the load conditions change, its all over the place and you are hard-switching current, which will be fatal if you are relying on soft-switching currents higher than its peak rating. 

[Steve Ward]

A bit of DRSSTC history:

The guy that made the first "impressive" DRSSTC (and who coined the acronym) was Jimmy Hynes, and his coil never used feedback.  It was simply a microcontroller with a POT feeding into an ADC that would adjust the PWM output frequency to the IGBT gates, and he would fiddle the knob to get the biggest primary current and sparks.  His machine made ~8 feet of spark with some CM150s until it blew out the MMC capacitors, which was something like 2 series x 10 parallel of the 0.15uF 2kV CDE rolled caps that were so popular for spark gap coils at the time.  The caps did not last long at all when subject to AC voltages that were far in excess of their DC rating!  However, the IGBTs managed to put up with the abuse of >1000A peak and non-ideal switch timing.  I think one reason the "fixed" oscillator approach worked OK on this coil was because of the very low primary impedance and relatively low coupling coefficient (guessing it was about K=0.1).  These 2 things mean that 1) the "burst" length to make a good spark was short (<10 cycles) and 2) the change in primary resonance frequency due to spark loading is less severe.

Hard switching currents that exceed the "Safe Operating Area" (SOA) for the device can trigger immediate failure.  To save $$, many DRSSTCs push the IGBT current beyond the Switching SOA, in which case you really want to be sure your driver maintains correct switch timing near the zero crossing rather than somewhere near the peak.  However, it's fundamentally difficult to achieve both bad switch timing and huge resonant currents, because the resonant rise depends on the driving source to be near Fres.