Author Topic: DRSSTC Performs only with 1cm Arcs - Troubleshooting (Read 9243 times)

davekni · « **Reply #20 on:** March 02, 2021, 10:48:04 PM »

Yes, higher primary impedance is well matched to longer on times. Pulses/second depends on heat sinking and other things. I can't predict the future.

Not sure what your screwdriver test was. Did notice that your MMC doesn't have bleed resistors. Most have a high-value resistor across each section to make sure voltage remains evenly distributed and that the MMC discharges after use for safety.

I can't determine every detail of your circuit. I always add lots of bypass capacitors - more than needed to keep supplies very clean. Capacitor type matters too. Z5U caps are worthless. X7R and X5R can be useful, but many drop by 90% at rated voltage (only 1/10th of capacitance left). NP0/C0G are great, but physically large and expensive.

For SSTC, RMS current rating is the key for the DC-coupling capacitor. That one lists 5A. You could likely get away with 10A. That's 100A at 1% duty cycle, 50A at 4%, 30A at 11%, etc.

I didn't design UD2.7. My guess is 150pF filters out high frequency arc noise. Increase 51ohm rather than removing it.

I use a simple Faraday cage for my DRSSTC driver - open top cardboard box lined with aluminum foil connected to ground and driver ground. Not everyone does. Just make sure there is a grounded strike rail that any arcs will hit before reaching electronics.

ChrisBlis · « **Reply #21 on:** March 03, 2021, 10:43:20 AM »

Is only one bleed resistor on the fullbrdige output beetween 5k-50k enough? Or do I expecially need to add a resistor of a high value acros every cap of the MMC. May the unevenly distributet voltage be a very likely failiure? How big should this Resistor across every cap of the MMC be (1MOhm?) and can I use any standart 0.25W resistor for this task, or will the voltage jump across them?
Do I then still need the 5k resistor on the bridge output?

Will my circuit board be disturbed likely by noise or should this be with my small teslacoil be problem?

Can I also go without strikerail at my small arc lenght?

To which value should I increase the 51Ohm resistor at secondary feedback of 1:50 ? Maybe 1k?

davekni · « **Reply #22 on:** March 04, 2021, 05:08:32 AM »

If you wire your MMC as 6S4P, only 6 bleed resistors are need on the MMC. Wire each set of 4 caps in parallel, with one bleed resistor across each set of 4 parallel caps. For each bleed resistor, either buy HV resistors or use ~three 1/4Watt resistors in series. 3meg made of 3 1-meg resistors in series for each set of 4 parallel caps (600nF per set of 4) is 1.8-second time constant for discharge, so safe to touch in 10 seconds.

Yes, generally you still need a bleed resistor across the bridge output. Interrupter pulses come too quickly for the 1.8-second time-constant bleed to be helpful in oscillation start-up.

1k may be about right for a small SSTC with 50:1 CT. If I guess at 2A peak secondary, that's 40mA peak, or 40V peak. 500 ohms would be 20V peak.

If arcs stay small, then you can get away without a strike rail. Many SSTCs don't include one.

I can't predict disruption of your driver due to noise.

ChrisBlis · « **Reply #23 on:** March 04, 2021, 09:30:38 AM »

Because I glued the MMC up in my hybris I have to go with 3P8S. Should I also be able to use 3x 1/4W 1 MOhm resistors in series for this task?

Does the resonant circuit also give energy into the bridge internal diodes when shutting of, or is this not the case?

Are 5x (470kOhm smd 2512 1w) resistors in paralell also suffiecient as bleed resistor at the full bridge output (resulting in a approx. 100kOhm bleed resistor)? Will the withstand the voltage of 600V peak? And is this value low enough to discharge the mmc beetween the on times?

Is the parasitic inductance of the bus cap desisgn shown in the picture critical?

The cables are of course also soldert on to the back of the doubled sidet copper board, in order to get a healthy copper to copper connection. (Maybe a iron connection only would be not very good because of the ferromagnetic behavior?)

davekni · « **Reply #24 on:** March 05, 2021, 12:37:50 AM »

3 x 1meg sounds good, 8 sets, one for each series stage of your MMC.

If you are running 320V bus, the bridge should generate +-320V between outputs. 600V is a valid consideration for short switching transients. I'd guess generic 2512 parts would be OK there. There are specialized 2512 parts rated to 3 or 4kV. (I use such for bleeder resistors.)

With MMC being reduced to 56nF, 100k makes for a 5.6ms time constant. That should work fine below 200Hz interrupter rate. Above that it may or may not have trouble starting.

Yes, resonant energy is dumped back into Vbus at the end of each interrupter pulse. H-Bridge output bleed handles the remaining voltage once output is below VBus so the diodes don't conduct any longer.

You could minimize bulk capacitor inductance by using two separate twisted pairs, one pair from each cap. You wouldn't need the bus bars between caps, but could leave them if helpful mechanically. Twisted pair has around 660nH/meter, so ~66nH for 10cm. Say 100nH including loop area at the ends. Two in parallel is 50nH. With 1uF snubber, that's 0.22 ohms reactance at resonance, for 44V spike at the end of a 200A burst. Probably good enough as you have it, perhaps 80V spike.

ChrisBlis · « **Reply #25 on:** March 13, 2021, 11:47:19 AM »

Do I need to ground shieldiet cables in order to get the shielding effect?

May my electronics get disturbed by the RF of the cables to the rectifier?

davekni · « **Reply #26 on:** March 13, 2021, 06:02:13 PM »

Cable shielding is most effective when grounded to the local circuit-board ground at each end. One possible exception: If one end of the cable is connected to just an optical-coupler input, then it might be better to leave that shield end open or connected to the negative return side of the opto-coupler input.

Rectifier wiring should have little RF energy, so isn't a concern.

ChrisBlis · « **Reply #27 on:** March 14, 2021, 11:54:35 AM »

Should it otherwise also be possible to add artificial grounding plates at each side of the cable?

davekni · « **Reply #28 on:** March 14, 2021, 05:55:56 PM »

Grounding/shielding/isolation can be a complex subject. I can't know enough about your setup and surroundings to say what will help.

ChrisBlis · « **Reply #29 on:** March 15, 2021, 08:48:21 PM »

If I want to use a micro controller as interrupter can I then add a zener diode on the GPIO Pin or will this not be sufficient enough. Should I otherwise use another optocoupler?

davekni · « **Reply #30 on:** March 15, 2021, 09:12:02 PM »

Either is likely fine. Try something. If the micro gets confused from coil arcs, then add an opto or fiber connection.

Plastic fiber isolation can be made with standard 5mm LEDs and phototransistors. Drill ~2.2mm diameter holes in the ends and insert fiber ends with glue (epoxy or cyanoacrylate). For this fiber scope project I did that with the LED end only:
https://highvoltageforum.net/index.php?topic=1263.msg9353#msg9353
I've glued both ends (phototransistor and LED) for other projects.

ChrisBlis · « **Reply #31 on:** March 16, 2021, 02:31:50 PM »

Would it be critical to extend the lead wires of the bus caps to the h-bridge to approx. 20cm in total, still twisted pair.

May the occuring transients be problematic?

Can the transients distroy my bus caps (400V) or my rectifier (800V) on 325V bus voltage

I just realised I not groundet the input pins of the unconnected gates is this a big issue?

davekni · « **Reply #32 on:** March 16, 2021, 06:23:23 PM »

H-Bridge voltage peak at the end of bursts will increase at 20cm. Scope it and see that the peak remains well below 600V. If not, then add a second twisted pair. Bulk caps and rectifier don't see the peak - only the H-Bridge (IGBTs).

Yes, connect floating IC inputs. A solder bridge from input to adjacent output of another inverter is an easy fix. (Don't bridge an inverter input to its own output.)

ChrisBlis · « **Reply #33 on:** March 18, 2021, 04:56:11 PM »

What do you mean with driving the oscillator out of phase? Why does changing the polarity of the primary coil fix this problem?

How can I calculate the peak current of the rectifier diodes.

ChrisBlis · « **Reply #34 on:** March 19, 2021, 08:47:39 AM »

Has this to do something with the parasitic inductance of the loop to the primary coil. Or do you just mean swap the polarity of the whole LC primary circuit, because then would everything be clear for me.

davekni · « **Reply #35 on:** March 19, 2021, 06:05:18 PM »

I do not understand what you are asking.

I suggest that you look at working examples on this forum, and those posted on popular HV sites such as Mads Kaiser Power Electronics, Steve Ward's, etc. Update your build based on that and all the answers above, then see what happens. If you continue to have problems, post pictures and scope captures of your updated build.

ChrisBlis · « **Reply #36 on:** March 29, 2021, 12:52:23 PM »

Hopefully things work out better now.

ChrisBlis · « **Reply #37 on:** March 29, 2021, 10:10:58 PM »

Are things looking out fine?

My concerns are, that the parasitic induktance at the gdt may be problematic.

davekni · « **Reply #38 on:** March 30, 2021, 01:06:00 AM »

I think it looks good. Yes, the GDT output wiring could be twisted a bit better, especially the brown wire to the lower-left IGBT. However, I've seen worse examples still function.

ChrisBlis · « **Reply #39 on:** March 30, 2021, 09:51:12 PM »

Today I testet the teslacoil.

Vbus was at 50V.

The thes went as last time. Only 1cm arcs to a screwdriver.

The tests were always performed with one pulse per second.

I addet a scope shot of the primary current with a artifcial aluminum load. Rf was later adjusted to 90kHz.

The aluminum plates were bzzing at every cycle of the on time.

What may be problematic. Still parasitic inductance?

High Voltage Forum · « **Reply #39 on:** March 30, 2021, 09:51:12 PM »

Author Topic: DRSSTC Performs only with 1cm Arcs - Troubleshooting (Read 9243 times)

High Voltage Forum

Recent Topics and Posts