Highway to DRSSTC

[Laci]

Here are some pictures of the driver.I put it into an ATX PSU case,added to the front side a 7 segment voltmeter to measure the OCD voltage,and a potentiometer to set it.
Now I have problems with the GDT waveform.It looks almost OK without load,but extremly bad with a single IRFP460 as load.
The bridge is not done yet,I am thinking to use 2xMUR1560 and 2xBYT108/400

[Laci]

A halfbridge is done.By removing the two 10uF tantalum capacitors from the driver(which are in parallel with the 10 ohm resistor),the signal gets better.I have no experience with IGBTs,so maybe not good enough,what do you think?How can I enhance the signal to more like square?Also can I connect the secondary base to mains ground if I am running the halfbridge from a transformer?

The screenshot with the white waveform is the comparison with(blue)and without(white) the 2x10uF capacitors in parallel with the 10 ohm resistor.

[Mads Barnkob]

I like the addition of a OCD setting display.

In https://highvoltageforum.net/index.php?topic=165.msg1273#msg1273 I told you to use polypropylene or ceramic capacitors for the DC blocking capacitor, not tantalum, they are not suitable for that application.

I am not a fan of using mains ground as it is terrible as a RF ground, either make your own separate ground rod for RF or put out a artificial ground mesh/plate underneath the coil, with some fence or aluminium foil.

To troubleshoot your GDT waveforms take a look here: http://www.richieburnett.co.uk/temp/gdt/gdt2.html

[Laci]

Without capacitors it looks like the "Slightly overdamped gate drive waveform".I will try to decrease the 10 ohm resistor to something like 4.7,but if that will not help,probably have to buy some good capacitors.The IGBTs switching can be checked from high side E to ground,and low side C to positive?

[Garyf]

I had the "slightly overdamped" problem until I rewound my GDTs, going from 8 turns to 15.

I notice the various driver designs all have a capacitor in the GDT drive, but the configuration differs a little.

Steve 2005 has a 3uf cap only.
Steve 2010 has two 1uf and two 33ohm in parallel
Screamer has 10uf and 10ohm parallel
Loneoceans has 2.2uf in parallel with series combo of 1uf and 15ohm
OneTesla Ts has 3.3ohm series with 1uf

I understand the "flux walking" treatment as a good standard practice, but has anyone revisited this lately? When I shorted out my capacitor I got a cleaner drive waveform with no sag and no evidence of saturation.  But I see lots of warnings about leaving out the capacitor.....

[Mads Barnkob]

You are right that is a engineering best practise, but that is for a reason and the little cost of it does not really justify to remove it. The DC blocking / AC coupling capacitor is needed in a single ended driver. If the driver on the other hand was a push-pull type there would be no need for it as the driver would ensure to demagnetize the core as it switches polarity.

Maybe you used a too little capacitance and had a low frequency ripple from the capacitor and primary winding inductance?

Page 38 and forth in here tells what you properly already knows: http://www.ti.com/lit/ml/slua618/slua618.pdf

[Laci]

Thanks for the ideas!I managed to get a bit better waveform,by using 2 primary coils and a 0.68uF MKP cap in series with them,the 10ohm resistor is still there.The interrupter is almost done.I tested the complete project and it gives some mm arcs at around 25V.I expected more,but at least it works.Some problems are still existing:for example the GDT drive transistors heats up without load,so I think the N-P pairs are both on at the same time.I must fix this problem,but I have no idea what is causing it...Maybe I will try a bootstrap driver.

[Laci]

I made a some things today.The interrupter is almost done,just a lcd is required.The fullbridge too.I used 2X BYT108 & 2X MUR1560 diodes as the HGTG30N60A4's internal diode.Another thing is a bootstrap GDT drive prototype.It seems to work much more efficiently than the UCC3732 1&2 +IRF540 & IRF9540 output stage.Without the GDT the current drawn is really small,and with GDT it goes up to about 600mA,however the IRFZ44 mosfets heats up slowly,simultaneously.The only down side of it is that it requires a lot of space.
The GDT output signal is not that good,pretty much the same as with the halfbridge,but without the 10R resistor;just a 1uF cap in series with teh GDT,but I made a 20V test and the results are around 2cm arcs at 1A.Before the fullbridge I tested the halfbridge at 50V and that was working quite good,pretty much the same as with mosfets,but that is normal.
An important thing would be to make a new primary,with tighter turns.

[Mads Barnkob]

Low voltage spark length in a DRSSTC is not really that much of a indicator of how will it will perform at full input voltage, as we always detune the primary circuit to a frequency lower then then unloaded secondary circuit, so that once long arcs are loading down the secondary circuit it will hit the same frequency as the primary circuit and you will maybe first start to see long arcs when above 75% full input voltage, this is however more pronounced with bigger coils.

How high a coupling are you operating with, with that conical primary coil?

Using the prototype board for the bridge can be a very bad idea if you have a voltage spike / transient that is high enough to jump between all the copper points, the creepage distance is really low with these boards.

You should also not use a breadboard for a GDT output stage, it is simply too high current so you risk damaging it

[Laci]

I am testing things as SSTC,without primary capacitor,not using OCD,but I think it should work as well.

I do not know the coupling of the primary,it's heigh is around 1/3 of the secondary,which I think is definitely way larger than it should be.Larger arcs come out at 4-5 turns.

I know the prorotype board is a bad idea for the bridge,but I have no components for etching or anything else.

The breadboard is also bad,however the driver works better with that,putting it together on a prototype board should not be a problem.


While testing I already killed a diode,luckly the IGBTs are OK(hopefully).The bridge was running from the 50V trafo,and the arcs were big enough at CW.When I turned on the interrupter,some clicking sounds came out from the bridge and no arcs,so I turned off the power fast,then the diode was dead.I think it died due to a high voltage spike,which was caused by the unsynchronized switching of the IGBTs.
I am stuck here.If the GDT signal will not get better after removing or adding turns to it,I will have to buy some 10uf caps,like these:
https://www.aliexpress.com/item/McIgIcM-200pcs-50V-monolithic-ceramic-capacitor-multi-layer-ceramic-capacitor-p-5-08mm-510pF-820pF-2/32825899885.html?spm=2114.search0104.3.22.1fafadf0M75SYh&ws_ab_test=searchweb0_0,searchweb201602_4_10065_10344_10068_10342_10343_10340_10341_10084_10617_10083_10616_10304_10615_10307_10301_5920011_10313_10059_10534_100031_10103_441_10624_442_10623_10622_10621_10620_10142,searchweb201603_6,ppcSwitch_5&algo_expid=1aa53206-8ae9-4463-877b-ec8edfd94481-3&algo_pvid=1aa53206-8ae9-4463-877b-ec8edfd94481&transAbTest=ae803_5&priceBeautifyAB=2

Are they good for this application?

[Mads Barnkob]

You could do your bridge in point to point soldering with some thick copper wire, like I did in my SSTC1/2. I am not saying that it is a good layout, its open, prone to failure if the solder melts and properly not low inductance, but it worked fine for 4 kW peak CW



You should get some MKP capacitors for the decoupling, metallized film polypropylene capacitors are superior to almost any other in specifications / price.

[Laci]

The driver is just not working.I tried various GDT turns from 13-17 turns and all combination of resistors+capacitors in series with it.The bootstrap driver is also lost,I soldered it's circuit to a prototype board and after powering it up,an IR2110 died.I tested the dead IC's socket with the other one and that was working there.Then I inserted another IC,then they both died.Now I have 8 dead IR2110.Probably they die,because they are crappy fake ones.I'm planning to replace them with IR2113s.
It's interesting,that the old UCC3732 1&2 SSTC driver can drive a halfbridge with a good waveform at 15V,but the complicated,robust driver can't.Something is really going wrong here...

[Mads Barnkob]

I hope you get it to work, the best advise I can give you is to take the driver apart, verify every single component is still in working order and rebuild your driver from scratch, there has to be some defect or error that you have not found.

Systematic fault finding takes time, but it also guarantees a much better product in the end.

[Laci]

The driver is not working yet,but I found a dead UCC chip.Since I have no replacement for that and the bootstrap drivers are all dead,I have to buy new component,like TC4420,because that's cheaper than the UCCs.The other questions are still the 10uF capacitors.I found just large polypropylene caps,which are really expensive,but also the C1206C224K5RAC SMD type.I think I will buy some of these,but I still do not prefer SMDs.Do you know if THTs are existing for this purpose for cheap?

[Mads Barnkob]

How about something like these, if you are still going for 10uF: http://www.ebay.co.uk/itm/1pcs-ARCOTRONICS-MKP-1-71-10uF-10-F-160V-10-PCM-27-5mm-Radial-Capacitor/151600383845?hash=item234c164765:g:i60AAOSwqu9U7Gqc

[Laci]

Those should be perfect,but way too expensive.Two of them+shipping would cost almost 10$,which is like 1/3 the price of a commercial drsstc driver.
Yes,I paid no attention to the previously mentioned capacitors,they are 220nf.I have a lot of MKS-4 1uF and other values of those.I thought the small capacitor value causes the problem,but as you said,you also used smaller values,so the problem is something else.Probably the bad UCC37321 driver,which dead,and it was drawing no current,but pulled the outputs high.That caused the N channel mosfet to turn on and that just shorted the other P channel to ground,so the transistors heated up.I am going to buy TC4420 drivers,which are pin compatible with the UCCs and much more cheap.

[Laci]

After days of troubleshooting,I finally could not get the driver to work...The problem is the same as before:unusable GDT output waveform.Yesterday I received some new drivers,4xTC4429 and 2xIR2113,already some are blown,due to my stupidity.These IR2110&IR2113 chips are really unstable chips,because they blow up even if the drive signal is not good,for example if you touch the input pins.

I tried everything to fix it.All combinations of resistors and capacitors at the GDT,the signal at the output mosfet stages drain looks ok,perfectly square,at the gate too.
The only thing I can try is to remake this part of the driver again,but the mosfets seem to work fine.I just wonder if anybody has anything to mention,or some ideas?Any help is appreciated.It would be a huge fail to let this project go here...

[oneKone]

the only advice i could offer and have done in the past is try to clear your mind and remake the circuit on a new breadboard. often than not it could be a simple error that is being overlooked. Passed that i absolutely do not want to sway you from the circuit you are designing.... but i would try a simpler driver with a ucc27425 and a pair of IRF540 and IRF9540 on each output, basically what mads done on his drsstc board.

[Laci]

Cheers for the reply!
Today I removed the P channel mosfets from the board,they looked working for me,but after making some breadboard circuits,they were dead.Since I have no more P channel mosfets,I just tried paralleling a UCC37322 with a TC4429 on one side and 2xTC4429 on the other side,to replicate Steve Wards first DRSSTC design,but with no success.Maybe the 2x6A and 6A+9A drivers are not enough.I will hopefully buy some P mosfets from a local electronics store and try the UD1.3 again.There is also a chance that something else has a problem,because I watched an older waveform(which is uploaded to this topic),where I was driving a fullbridge of IGZ100N65H5XKSA1 with just a UCC3732 1 & 2 and got a really good waveform,much better than now with a halfbridge of HGTG30N60A4 with the mentioned driver.

[Mads Barnkob]

I agree with you that there is some other problems, because these drivers should drive a MOSFET/IGBT fairly good on its own, maybe with some rounding on the rising edge, but still should be okay.

It can be a long and tiresome road to get out of the deep fault finding hole you are in, did you try to start all over from scratch with new components?

Try to only build small parts of the circuit first, power it up, check if all voltages are present as it should be, now build on the next stage, test again and so on until you have a working driver, that should help you with finding out in which stage the fault is.