Weird ringing

[buchtawill]

Hello,

I've been busy the past ~month building a DRSSTC for Worcester Polytechnic Institute (I am a freshman studying EE and computer science). This is my second large DRSSTC, here's a quick run down:

*Two 15,000uF 350V aluminum electrolytics in series for 7,500uF @ 700 VDC rating
*Power thru variac for up to 400V on the bus. Could possibly use higher bus voltage in future
*FF450 series IGBTs, 1200Vce. See datasheet folder.
*15 strings of 4 942C20P15K caps in series for ~550nF @ 8KVDC, 1.4KV rms derated.
*Resonant frequency of around 65kHz. Still have to tune primary.
*Secondary coil is 2600 turns of 26AWG wire on 6" PVC, ~43" long. Coated with 2 layers of epoxy
*Primary is 5-6 turns of 3/8" copper tubing. Connections to primary and MMC are 2 strands of 6 AWG wire
*Primary circuit is mounted on large heat sink, all connected by 4mm copper busbar
*Driver board is UD2.7c, phase lead not yet implemented. OCD set to a conservative 600A for the time being.

Folder with all relevant datasheets:
https://drive.google.com/drive/folders/1kL6LuwodEBO9mZbCrnn2QW135SiPJtNz?usp=sharing

Here's a folder with all photos of the build:
https://drive.google.com/drive/folders/1Itj2b1XsIh36L6cnCYV5a5Gj6Og6ui5g?usp=sharing

Here's a video of it in action:
https://drive.google.com/file/d/10YgtUomzpNwaoSLM_fcaC_5-Yb45gqDK/view?usp=sharing
And here's another:
https://drive.google.com/file/d/1Ism8dJ6ExHD3cvlw9vcu9pmDup8iqSFr/view?usp=sharing
You can hear me say "uh oh" at the end of the second video. I thought I blew an IGBT, but really the fuse on the variac went out! It's nice to have an issue with power supply instead of the bridge.

Anyways...
When I was scoping the primary, there was some strange ringing:
https://drive.google.com/file/d/1zkfeNz5XmHUd1OsSY2roKvLLWx_6c68M/view?usp=sharing
https://drive.google.com/file/d/1VsXrqr0qZbCER665NyauXxkr8RvLvbRJ/view?usp=sharing

Could this be coming from parasitic inductances in the bridge? I've only gone up to 325V on the bus, I'm afraid that any higher could cause this ringing to be an issue.

Any other thoughts/questions/comments?

Thanks,
Will

[Weston]

Nice build!

The first check is if the ringing is real. Whats the probing setup? I don't see any differential probes in any of the pictures. If you probe the DC bus instead of the switching node is the ringing still there? If you have both leads of the probe on the negative side of the bus do you see nothing?

A possible mechanism is that the snubber capacitors on the IGBT terminals are ringing with the inductance of the bus capacitors + bus bars. The big electrolytic capacitors are basically a low inductance short at high frequencies.

If this is the issue, if you increase your snubber capacitance (it looks like you might have space to double up the snubber caps?), it should shift the resonance away from your operating frequency.

[davekni]

Could this be coming from parasitic inductances in the bridge? I've only gone up to 325V on the bus, I'm afraid that any higher could cause this ringing to be an issue.

A possible mechanism is that the snubber capacitors on the IGBT terminals are ringing with the inductance of the bus capacitors + bus bars. The big electrolytic capacitors are basically a low inductance short at high frequencies.

I believe this ringing is real.  Bottom half of H-bridge output waveform is relatively clean 0V.  Probing issues usually show ringing on both high and low states.

Yes, this is a ring of the snubber capacitance with bus-bar inductance.  Bulk capacitors are an AC short-circuit, so inductance is from snubber capacitors to bulk capacitors.  Unfortunately, the ring frequency is very close to twice your coil operating frequency.  H-bridge current load on its DC supply input (VBus) looks like a rectified sine-wave, so is dominated by 2x frequency, right where your ring is.  I do suggest fixing before increasing power.  Tuning the coil could cause operating frequency to hit exactly half of ring frequency, increasing ring amplitude yet farther.  Of course, your 1200V IGBTs give you margin for large ringing.

Lower VBus impedance is generally better, so best fix is to increase capacitance or decrease inductance.  (Even if you did both and hit the same frequency, ring Q would be lower, so lower amplitude.  One or the other fix is likely safer, however, moving ring frequency either higher or lower.)

One simple fix for lower inductance is to add some insulated copper or aluminum under or above or wrapped around the bus bars.  This restricts the magnetic field, which passes between the two bars and loops around the outside of the bars.  Insulated by a layer of something of course to avoid shorting bus bars.  If the added copper or aluminum is thin enough (ie. thin foil), it will add damping besides lowering inductance, but the foil will heat up some too.  Not sure if the heating would be enough to be problematic.

No need to go to these extremes, but for reference, here's my thread on low inductance bridge design, half-bridge first, then a couple full bridges:
https://highvoltageforum.net/index.php?topic=1324.msg9795#msg9795

[buchtawill]

Hello,

Thanks for the replies and your ideas. I ended up doubling the snub capacitance by placing two more caps on the bus, as seen here: https://drive.google.com/file/d/1AZ76udfhaEa-8PIddRayiyQVBYZEMVU7/view?usp=sharing. You can also see the probe connections for the following scope shot:
https://drive.google.com/file/d/1I0mnVANob44GsXKjn4vJSQVXD2BWrxFR/view?usp=sharing

That seems to have taken care of the issue well enough. The transients seem high, but I don't think I need to worry because they're well under 1200v. Heating also isn't an issue; I ran the coil for a decent period of time (~8 minutes with short breaks in between songs) at full power, and the heat sink was only warm to the touch.

As a side note, when I was testing the bus at 350v (not pictured), the OCD tripped after just 2 cycles. Scary in my opinion.

I would also like to thank everyone on this forum for their detailed builds, especially Mads for his DRSSTC 1 write up. I really learned a lot following your build on my previous Coil. All this has kept me entertained over the past 2 years . It's safe to say that I wouldn't have been able to make this coil without this community.

Here are some more videos:

https://drive.google.com/file/d/1KyCgH2_GW8i_lfpE901m7jdYXsrRBR48/view?usp=sharing

https://youtu.be/jtbsVDrVPQY

https://drive.google.com/file/d/1wynRvVx7Rr0uifXccZ6RAO8KxCkHofqi/view?usp=sharing

https://drive.google.com/file/d/1uOeRaNgaVVidVXlBwK4LK9k_m7skJI0F/view?usp=sharing

[davekni]

Looks good for lowering Vbus ring frequency.  If you go for really big arcs, which lower frequency due to arc capacitance, you could get close to 2x again.  1200V IGBTs should survive anyway.  Worst-case, ring amplitude is clamped to be from 0V to 2x DC bus voltage.  IGBT diodes prevent Vbus from going significantly below 0V.