UD2.7c adjust self 7M3 feedback (oscillation ?)

[JCF]

Hello, with my new coil, I have strange oscillation who appear for certain plunger settings when I want to make the minimum of phase shift . In this case the sound of sparks are not regular and an oscillation appears at the lower signal output of full brige (vdc = 50V)
Measure of resonnance frequency of primary alone 157khz, secondary alone 153khz

have you ever met this





tesla Detail 80 180.txt

[davekni]

Nice scope plots, but a few questions about them:  Is channel 1 (bridge output) from a differential probe across the two bridge outputs?  If so, I'd expect +-50V (discounting spikes).  Looks more like +-25V.  Or, is it of one bridge output?  I'll guess so based on amplitude.  Also implies channel 1 vertical is AC-coupled.  Changing to DC coupling will make traces easier to interpret.  (Edit:  Or, perhaps channel 1 is differential and Vbus has sagged from 50V to ~25V at the point of the scope capture.)  Is channel 2 of a separate CT, or of UD2.7 R1 (feedback 51-ohm resistor)?  What is the CT ratio and burden resistor for channel 2?

Observations about your scope traces:

Duty cycle of bridge output is a bit over 50%.  Non-50% duty cycle is normal for UD2.7 at low current.  However, at +-30V on channel 2, I'm guessing current isn't too low.  You may want to scope through UD2.7 to find the asymmetry cause, as it could be related to other issues.

Phase lead requirement can be a bit different at 50Vbus vs normal higher voltage.  Doubt this is too significant here.

have you ever met this

Not exactly the same, but I can tell you roughly what is happening.  The second scope capture has sufficient phase lead for rising edges, but not quite enough for falling edges.  (With duty cycle over 50%, falling edges are delayed relative to half-way-between rising edges.  Thus falling edges require more phase lead.)  On falling edges, one pair of IGBTs turn off before zero-current.  Remaining current causes bridge output voltage to transition.  Proper so far.  Then current crosses zero, causing bridge voltage to transition part-way back.  Then opposite pair of IGBTs turn on at non-zero Vce.  That turn-on initiates the high-frequency ring.  The ring itself is a resonance of parasitic VBus inductance with IGBT Cce and/or snubber capacitance.

Improving VBus construction could reduce ringing, as could slowing IGBT torn-on speed.  This would help with robustness in the event of insufficient phase lead, but not fix the phase lead itself.

So, either find and fix the duty-cycle issue or add sufficient phase lead for falling edges (as in your first scope capture).

[JCF]

thanks for all. these last weeks I could not connect to the forum anymore, thanks to Mads, all is ok now

Nice scope plots, but a few questions about them:  Is channel 1 (bridge output) from a differential probe across the two bridge outputs?  If so, I'd expect +-50V (discounting spikes).  Looks more like +-25V.  Or, is it of one bridge output?  I'll guess so based on amplitude.  Also implies channel 1 vertical is AC-coupled.  Changing to DC coupling will make traces easier to interpret.  (Edit:  Or, perhaps channel 1 is differential and Vbus has sagged from 50V to ~25V at the point of the scope capture.)  Is channel 2 of a separate CT, or of UD2.7 R1 (feedback 51-ohm resistor)?  What is the CT ratio and burden resistor for channel 2?

channel 1 is THE full bridge outputs with differential probe X50X AC coupled. VDC +/- 25v
channel 2 is separate secondary current Transformer OPCT16AL 2000:1 with resistor 100  ohms AC coupled. The diodes TVS Bidi VD1 are not cabled

It's all I can do to minimize oscillations. more I don't understand the curve.

I believe that I will stay with the least oscillations, I do not see what to do more. Is dangerous for IGBT ?

  noyau full out

noyau full inside

noyau middle (for less oscillation)

gate of one IGBT

[davekni]

It's all I can do to minimize oscillations. more I don't understand the curve.

Inductance (and therefore phase lead) is maximum when the ferrite adjustment slug is centered inside the inductor winding.  Your adjustable inductor (as common) allows the ferrite slug to continue below the winding center as well as above.  Thus full-inside and full-out are both lower inductance than centered inside the winding.   Your middle position looks good for phase lead.  Waveforms look reasonable with middle position (high phase-lead).

Gate oscillation may actually be VBus ringing coupling into scope probing of gate signal.  Presuming so, it is probably not a serious issue.

GDT leakage inductance looks high, with some overshoot on Vge rising edge even with 5 ohm resistors.  Falling edge would have even more undershoot, but is being clamped by Vge TVS diodes.  Make sure the Vge TVS diodes don't overheat at high duty cycles.  There are several discussions on the forum about GDT construction and minimizing leakage inductance within the GDT and GDT lead wires.  May not be critical, but I suggest taking a look.  Pictures of your build (bridge and CTs and GDT and wiring to driver) would help.

Remaining VBus ringing is likely OK with sufficient phase lead (middle adjustment).  As with GDT, minimizing interconnect inductance is key to reducing ring amplitude if you decide to pursue such.  Again, pictures are helpful.  For an extreme in inductance minimization, here's my parallel-copper-foil VBus construction:
https://highvoltageforum.net/index.php?topic=1324.msg11785#msg11785

BTW, DC-coupled scope traces are often more useful.  DC information is lost with AC coupling.

channel 1 is THE full bridge outputs with differential probe X50X AC coupled. VDC +/- 25v

Scope traces suggest VBus is 25Vdc total (0 to +25Vdc).  H-bridge output is +-25V.

[JCF]

Hello,

After having reduced the large Vbus wires to 15cm, the oscillations and the phase shift are improved.



to calculate the current, is it the voltage across the 100 ohms = 96V CC or 96/2 = 48V. (Current sense is 1/2000) ?

[davekni]

Thank you for clarifying current scale.  Scope traces (current and bridge output) look good.