Alternative phase lead

[panjaksli]

Hey everyone,

So I have been looking for a proper analog phase-lead circuit, but unfortunately there is not much information out there.
One thing that's commonly used is variable inductor around 10s of uH with a 51 Ohm resistor, and the feedback voltage being measured across only the inductor. Correct ?
For whatever reason, I couldn't get the feedback to work, even with 1:100, 1:500, 1:1000 CTs. Seems like board wouldn't pick up the weak signal (Clouded leopard driver from Ali).

However the reason I'm writing this post, is that I found quite well working alternative. And that is, to use much larger value inductor 100uH - 300uH straight at the output of CT without any resistor. This way I was able to get the desired phase-lead without signal attenuation. But the thing is, it is pretty hard/impossible to find variable inductor of that value.
And by using a ferrite core, it's pretty hard to hit the sweetspot, because one turn could make the difference between having stable feedback and none feedback. Eg. (200-260uH for me)
But lucky me, the 10 piece package of "same" ferrite cores from Ali had anything but the same cores. After some digging I was able to find a core that would have 230uH at 20 or so turns, so the lower "AL" value is better for fine tuning.

Anyways, here is a picture, I have only one small issue, the igbts seem to turn on sligthly before zero crossing, but it doesn't seem to affect anything, since the feedback works perfectly.

[davekni]

The normal UD2.7 etc. feedback signal is CT output current across both the 51-ohm resistor and the variable inductor, not just the inductor.

Without the circuit for your driver, there's no way to know what phase lead options exist.  There may be clamp diodes directly across the CT input terminals.  That makes input impedance vary as current changes.  Your phase-lead solution may change phase as primary current changes.

Voltage switching slightly before current zero-crossing is ideal, better than at the exact same time.  That way the residual current causes a gentle voltage transition during the dead-time when all IGBTs are off.

[panjaksli]

Well, none configuration of R + L worked, but whatever, if the inductor alone works too, don't change it.
I found schematics of the input filter from the clouded leopard driver, it has infact a network of zener/schottky clamping diodes on the input, and then tesistor to the fpga, however adding resistor in series with feedback to reduce load on the CT doesn't really help, I may try lowering the CT ratio even further.

Also found out that the board itself has about 300ns+ propagation delay, which certainly doesn't help.