DRSSTC not working

[Merces]

Hello,
I m 16 years old and I made my first DRSSTC. The problem is that it doesn't produce big arcs and its pulling kinda lot of current It uses ud2.7 that drives an full bridge made out of 8 40n60 igbts (2 in parallel for each part of the bridge). the MCC has 150nf. The secondary is wind on a 110mm pvc pipe with a wire of 0.2 - about 1800 turns. Primary is made out of 10mm coaxial cable - 7 turns. The scoped signal is the current through feedback transformer with a can as the primary load . the phase lead its not enabled yet. the second photo with scoped signal its the current through feedback transformer with secondary installed.

[oneKone]

I'd say it's your primary being so detuned. 8 primary turns @ 150nf is about 82khz, I'd guess your secondary is about 200-250khz unloaded (really really rough guess).

[Merces]

I was thinking about this too. My secondary resonant frequency is about 130khz.
Here is the javatc file SavedCoil_2 (1).txt

[MRMILSTAR]

From your picture it appears that your primary coil has a fixed number of turns and cannot be tuned. How did you expect to adjust the resonant frequency?

[davekni]

That metal can on top of the toroid may be helping lower secondary frequency a bit to match primary frequency (adding capacitance).  The second scope trace seems to show some energy transfer to the secondary, as current does not continue to rise as in the first trace.

BTW, one way to help verify your JavaTC secondary turns count is to measure secondary DC resistance with a meter to see if that matches the JavaTC output (332 ohms in your case).  If DC resistance is lower, then turns are likely lower, so secondary frequency higher (farther from primary).

Do you have two probes for your scope?  Much more trouble-shooting can be done with two simultaneous traces.  One probe can be left on the same node such as one of the GDT input wires and used for trigger.  The other probe can be moved around for multiple measurements, current (as you have), H-Bridge output (but don't connect scope ground lead to H-Bridge when powered directly from mains), etc.

I'm guessing your scope horizontal sweep rate is 20us/div.  What is the vertical scale?  What is your CT turns-ratio and burden resistor value?  What H-Bridge bus voltage was being applied for those scope traces?

Thank you for posting all the detailed pictures.  That is quite helpful.  I don't spot any big design issues.  Hopefully you can get to your expected performance.  Sometimes it takes a while to debug electronics and find the more subtle issues.  It is all valuable learning, so don't give up.

[Merces]

My secondary resistance is about 315 ohm

I don't have real probes, i have the probes that last owner of the scope used. Just an alligator clip connected through a wire to an BNC connector, soo no components inside the probe and stuff.

I fallowed this video to make the current transformers:

/>
The current limit is set to 160A for now.

I think my phase lead its not tuned good yet, because I don't have a good inductor.

(Sorry for late response)

[davekni]

My secondary resistance is about 315 ohm

315/332 * 1800 = 1707 turns.  That likely makes your secondary frequency a bit higher than your JavaTC results.  The can on top likely lowers frequency by at least as much.  The exact secondary turns count is not likely any problem.

I think my phase lead its not tuned good yet, because I don't have a good inductor.

For the scope input showing current, are you probing the CT output?  The phase-lead inductor L1 will change phase of the CT voltage too.  To measure current accurately, connect the scope across R1 (the 51-ohm CT burden resistor on UD2.7).  The voltage across R1 is proportional to primary current without being affected by L1.

The GDT input scope trace shows parasitic oscillation (multiple transitions) on each edge.  That is likely leading to the high line current you mention at the start of your first post.  Such oscillations may cause IGBT failures when running at higher power.  Seeing edge oscillations lead me to examine your initial images more closely.  I think reconstructing your GDT and connections to IGBTs may fix this, or at least help.

The best image I can find now for good GDT construction is in this post:
https://highvoltageforum.net/index.php?topic=588.msg3779#msg3779
The upper GDT connected to a driver is the one to copy.  It uses a CAT5 cable, all four twisted pairs.  (There is no need to leave the CAT5 outer jacket on.  The four twisted pairs alone is fine.)  All four pairs are wound around the core.  One half (one wire) of each pair connects to an IGBT (or pair of IGBTs in your case).  The two ends of one wire going to an IGBT pair should be twisted together as they leave the GDT.  The other half of each pair connects to the driver.  (It would be even better if the wires to the driver were also twisted.)

Can anyone else help?  I thought I'd seen a good video or instructions for GDT construction using CAT5, with all four primary wires in parallel to driver.  Can't find that now.  If none exist, perhaps I'll make such and post.

After updating the GDT, also change GDT winding connections to the IGBT emitters.  The GDT output should connect to the emitter leads as close to the IGBT body as possible, not farther down on the board.  Move your emitter jumper wires out to be over the IGBT bodies.  That minimizes coupling between high-power emitter current and the low-voltage gate drive.

What voltage are you feeding to your UD2.7?  The scope shows about 11V on GDT primary.  Most IGBTs are specified at 15Vge.  Most DRSSTC builds run higher, 18-24V, in order to get more current out of the IGBTs.

Finally, I cannot tell from images the shapes of H-Bridge power traces/planes.  May be fine.  Top-level layout looks great.  For reference, here is my tutorial on low-parasitic inductance H-Bridge construction.  (It shows my GDT constructed from four twisted pairs.  Different cores, but the wire connections are the same as I tried to explain above.)
https://highvoltageforum.net/index.php?topic=1324.msg9886#msg9886

[Merces]

ok sooo, I manage to make it work, messed around with the current limit and set it to 280 amps. it makes like 50 cm arcs wile playing C4 note at 240v DC wile pulling max 3A.

But now i think my control circuit died. and probably the gate driver ics are broken. it drives the GDT for a second and then the current that driver board is pulling goes up.

I ll later investigate what it happened and come back with an update tomorrow probably.

This is how it worked (sorry for low quality video) : https://imgur.com/gallery/tDIYTgY

[davekni]

Great to hear of some success.

The parasitic oscillations on GDT input (gate driver output) are a likely cause of driver chip failure.  Multiple transitions increase power dissipation of both the driver chips and the FETs between the driver chips and GDT.

[Merces]

Well.. Nothing was broken. I think the board just crashed from the interference generated by the coil. I will try to shield it better.

I have a question: Should I implement an soft start for charging the main caps? I m thinking at something simple like this.