SSTC (Not) Working Correctly

[Laci]

I'm also planning to buy some IGBT's.
I've seen which ones are good for tesla coils so I have a collection to choose:
HGTG30N60A4
IRGP50B60PD1
IGW60N60H3
and IGZ100N65H5XKSA1 which have very promising properties of 650V,100A,500W+,but I don't know its switching speed.

Which IGBT would be the best at >150kHz frequency in full bridge  configuration?

[Mads Barnkob]

Laci: Could you also measure your power supply voltage for these experiments, because I think that you are loading that down so much that its voltage drops a lot and that is why you have no output.

Lets look at some parameters on the different IGBTs

IGBT	Vce____	Vce(sat)125Cmax	Icm____	Gate charge___	Combined switching time___	Switchng losses
HGTG30N60A4	600 V	2 V	240 A	300 nC	225 ns	1 mJ
IRGP50B60PD1	600 V	3.6 V	150 A	250 nC	180 ns	1.25 mJ
IGW60N60H3	600 V	2.2 V	180 A	375 nC	350 ns	3.5 mJ
IGZ100N65H5XKSA1	650 V	1.9 V	400 A	210 nC	480 ns	2 mJ

Vce(sat) is the voltage drop across the IGBT and the conduction losses are equal to Ohms law, so if this is 2 V and you have 100 A, 200 Watt losses, there is ofcause some derating in the link below that can be done due to short on-times. The lower Vce(sat), the better.

Icm is a good measure on hard you can push the coil, how much current you can conduct in the primary circuit.

Gate charge is a figure for the energy needed to turn the IGBT on, but also the same amount that has to be drained when turning it off, the large number, the better driver is needed.

Combined switching time is how fast you can switch the current on and off, the lower number, the high frequency is possible, for more details see here: http://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/igbts/

Switching losses is also very important, here you have to calculate how much power you have to dissipate from the IGBT to the cooling system (heat sink, water cooling, etc) in order to keep the die temperature rise within a acceptable limit, again see the above link for details.

IGZ100N65H5XKSA1 does indeed look very good, so does the others, but not IRGP50B60PD1, it has a too high Vce(sat). They will all be able to switch up to 300-1000 kHz(Fmax1), the limit you have to find is the power dissipation(Fmax2) to keep temperature rise within your design goal.

[Laci]

I made a driver,with PLL halfbridge(irfz44),running everything on 12V,so I can tune the frequency and measure it.At a range of 150-160kHz,a fluorescent lamp is lighted up at almost  max. brightness,current drawn is at around 0.1-0.2A at 5+ primary turns and under that the current increases rapidly and does not increases the output power.I did not expected anyting better than this,because the setup was just made in order to check the secondary's resonant frequency.
The feedback for the PLL is also working,it adjusts frequencies as even high as 180kHz down to the resonant frequency (150-160kHz).

The previous slayer exciter test loaded the 55v PSU to 50V,so that was not the problem.Even if I shorted the collector to the + rail,no current was flowing,so I think that the 47k base resistor was too big.In this case without a signal from the base of the coil,the transistor never turns on,and with less turns,the induced current into the secondary coil is not big enough,to trigger the transistor I THINK...

My new question is how and what kind of varnish can I use to insulate the secondary coil?I visited some nearly electronics and hardware  stores and nowhere could I find varnish for magnet wire.I have nitro based laquer,but I do not know if I can use it for this or not.Actually I want to protect the coil,and prevent the pretty lose windings from overlapping,not to prevent it from too high voltage.

[Mads Barnkob]

Try to look for a clear non-coloured polyurethane varnish, I have used some that was meant for boats or floors, just be sure it does not contain oil, some for boats does that and it has problems hardening up completely when there is nothing to absorb the oil (it properly would if it was used on wood). You should be able to find polyurethane varnish in all home improvement stores and paint stores.

I would like to use epoxy, but that is hard to come by in Denmark.

[Laci]

This evening I ordered 4xIGZ100N65H5XKSA1 IGBT,calculated the maximum switching time which is around 800kHz so the switching should not be problem.Also bought some 74HC109N's to maybe upgrade to DRSSTC,if it will work as SSTC
I will look for the type of varnish you said,and hopefully I can make the job.

[Laci]

The secondary got 3 layers of acril-pulyurethane water based eco  varnish.It dried very fast and isolated,fixed the windings very good.I added an interrupter to the circuit,increased the voltage to 30V and got pretty good results,2-3cm arcs.
Here are some photos about the setup,and the GDT,which I'm planning to use.What is your opinion about it?

[Mads Barnkob]

Good that you found a varnish that worked for you, 3 layers is good, normally I am also somewhere in that region with 2 to 6 layers of varnish, depends how much time I want to spend on it, and higher power Tesla coils also needs more layers, so a SSTC will do just fine with 3 layers.

Your breadboard setup can give you trouble when you start making longer sparks, with many long wires and leads to pick up the energy from the secondary coil, if you have problems with unstable control, try to move the electronics away or shield it with grounded aluminium foil.

Do you know what material type ferrite core you used for the GDT? At-least it does not look like a powdered iron core from a computer power supply (the yellow/green ones), but the windings itself look real good, for a half-bridge!

[Laci]

I do not know the material of the GDT...It came out of an old pc.My friend found it in his pc,and he said,that some wires were twisted around it(probably the ATX PSU's wires).
The circuit is made on a breadboard,because I do not have any perf board.It will be upgraded soon.

[Laci]

The 4 transistors have arrived.I'm a bit worried about the 4th kelvin pin,because I do not know what is it,but it's shorted to the emitter and the datasheet says that they are not exchangeable.
An oscilloscope is getting pretty necesarry,and I'm going to buy a cheaper,because it should be enough for me.I found the 3 cheapest scopes in the EU which are two hantek:DSO5072C and DSO4072C and UTD2102CEX.
I think that the DSO4072C would be the best choice,due to its frequency generator function,but its also a bit more expensive than the two others.Which one of the other two would you recommand?

Also,is it important to mount so high power transistors like these on big heatsink,while running SSTC?

[Mads Barnkob]

If the ferrite core came from a ATX power supply, there is a good chance that it is powdered iron and NOT useful for a GDT working at hundreds kHz in a SSTC.

Just bend the kelvin leg the opposit way or snap it off completely if you are not going to use it. If you want to learn more about kelvin connections: http://www.powerguru.org/kelvin-emitter-configuration-further-improves-switching-performance-of-trenchstop-5-igbts/

How are the prices on DSO5072P and UTD2102CEX compared to a Rigol DS1052Z? The first two has remarkably low memory record depth of only 40k points and 25k points. Hantek gives a lot more detailed technical specifications than Uni-Trend, so just on that point I would choose Hantek over Uni-Trend.

[Laci]

Well,I bought the DSO4072C 
The item's location is in Germany,but the seller is based in CH,so I hope,that they will send the scope from Germany....
Hopefully I will get it in 1-2 weeks,so I will have the chance to finish the project before nov.10 for a physics demonstration in my school.

Until then,thank you very,very much for the help!

[Laci]

I'm back again
I have some good results at a halfbridge of IRP240's,70V.The full bridge of the new mosfets is also made,i will test it and upload pics tomorrow.
One thing that is interesting is that without resonance cap on the primary,the results are not even nearly as good as with it.2x0.15uF 250V capacitors are connected to 11-13 turns primary gives the best results.I'm thinking why not buy some components to upgrade to DRRSSTC? So I have the following questions:
1.Would be good 2x 942C20P15K-F capacitors in series as resonance capacitors?
2.What type of ferrite ring should i buy for the current transformer and for the GDT?
 If you can,please send the accurate models of the ferrite rings,so i can buy them on mouser.com.

[Mads Barnkob]

You do not have a resonant capacitor in a SSTC, you have a DC blocking capacitor.

If you upgrade it to a DRSSTC, be sure to use a DRSSTC driver, else it will be a short life for your IGBTs, do not use MOSFETs in a DRSSTC, they have a much too high conduction loss at those peak currents.

2x 942C20P15K-F is what I used in my small DRSSTC2, so it can be good enough for smaller coils.

This would be a suitable physical sized, N30 material and a permability AL value of 5750: https://www.mouser.com/ProductDetail/EPCOS-TDK/B64290L674X830/?qs=sGAEpiMZZMt1hubY80%2fs8DEvGx1xc7EcUsoMlWTaffs%3d   and a datasheet for it: https://product.tdk.com/info/en/documents/data_sheet/R3600x2300x1500.pdf

[Laci]

Well some things happened since the day before yesterday...The oscilloscope have arrived,full bridge of IGBT's was made.Sadly its not working.I think that the IR2109 can't output enough current,because the waveform at the gate is anything,but not a square wave.I think that I will buy the ferrite rings,resonance capacitors and UCC37322 & UCC37321 chips this evening.
An other thing:I replaced the IRFP240's in the halfbridge circuit with an IRFP350 and a FQA24N50F just to try the circuit on mains.Firstly i connected a light bulb as a ballast to see if it is working.It was and produced great,around 5cm arcs.Then I removed the light bulb.What would you expect?Well...more than 20cm arcs started shooting out from the breakout point.It was exciting,at least for me...Of course the transistors died pretty fast,as always,so I could not make a picture or video.It was all good,and working until I put my hand so close to the interrupted arcs,that I almost touched the topload.At a point I heared a bang,the 3.5A fuse blown up,then the mosfets shoretd out.
It was a great experience,but I wonder why the transistors blown up?Please tell me if you know
Also,I'm not using isolation transformer and the coil is grounded to mains ground.

[Laci]

Here is the photo of the full bridge and the other is the high and low side waveform,the upper is the low side.

[Mads Barnkob]

Congratulations on making long and delicious sparks

The IRFP350 only has a continues rating of 10 A and pulsed 64 A, they properly died from over current. Your hand near the secondary circuit will also change its resonant frequency, but I do not think that should have resulted in phase shift/hard switching and therefor cause the explosion.

You are entirely right about that gate drive not being sufficient to drive anything, but to keep the switches in linear mode where they die from too much heat from conduction losses.

[Laci]

I hade some UCC's from the past,but they are not working.Now I got a single new 37321,hooked it up to a test circuit.I added a 1uF film capacitor to the power rails,and connected it to the GDT,the other end to ground.When I touched the input pin,a large current started to flow,and the chip acted as the others:driving large every time if I connect them to a power supply.I've seen somwhere,that it's not easy to kill chips like these.Am I the master of the chip killing,or they are not dead,just acting weird?Also,what about dead time of the IGBT's at a GDT?

[oneKone]

Hi Laci.

My experience with the ucc3732x driver chips is that they need an input or they'll fail, almost oscillate themselves to death! I've killed a few like this so i've always made sure they've had a driving signal. another thing i use is a dc blocking capacitor but i don't know if this still applies to when only using one chip as you've described. Hopefully this helps in some way. 

[Laci]

As a test,I connected the PLL to a bd139 & bd140 and through a 1uF cap to rhe GDT and the other end of the GDT to ground.It draws around 200mA,and the signal at the IRFP240's gate in halfbridge looks like on the picture.However, if I connect a load to the halfbridge,no current is flowing.Everything looks working,and I don't know what is the problem...

Another thing that I've realised,why the transistors blown up two times,when I connected the DC blocking or resonance capacitor:the resonant frequency of the secondary is 150kHz,but the primary's was a way more higher,but as stupid as I am,I blown up 4 transistors,instead of at least tune a proper primary-secondary resonance.Now I measuerd  the primary's resonant frequency and at a point it's 144kHz which should be good at long sparks right?Also,the previously posted GDT problem is still not solved...

[Mads Barnkob]

How much can we trust the voltages shown in the oscilloscope shot?

Ch1 is 20V/div and Ch2 is 2V/div, so you properly have no voltage at all driving one leg of the half bridge.

You do not tune a SSTC primary circuit, you only vary the amount of primary turns in order to decide/limit the current that can flow in it.