Highway to DRSSTC

[Laci]

After a short pause I'm back again
The concept is simple:make a working DRSSTC.

I've analyzed some DRSSTC circuits.The first question is about the CT feedback.Is this simplified schematic capable of doing a CW mode operation?Or it cannot start the oscillations without an interrupter?Then a simple 555 oscillator connected the the UCC's 3'rd pin will work?

[futurist]

Primary current in "normal" SSTCs is limited by the primary reactance which allows for CW mode operation. DRSSTC in CW mode would cause primary current to rise until switching transistors fail

Schematics you drew needs at least 555 interrupter connected to the enable pin to work. However, the interrupter can disable gate drivers when current flowing through IGBTs is not zero, so it would be best to use original schematics with flip-flop added to the interrupter logic

I suggest you read excellent DRSSTC guide written by Mads
http://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/

And short DRSSTC theory of operation
http://stevehv.4hv.org/drsstc_design.htm
http://www.loneoceans.com/labs/drsstc2/

[Laci]

What is your opinion about the HFBR-0500Z series optical transmitters & receivers?Which type would you recommend?The optical fiber cable: MIKROE-1473

[Mads Barnkob]

Good to see that you take on building a DRSSTC 

CW mode is completely out of the question for a DRSSTC, the closest you can come at the moment is properly the very long on-time type of QCWDRSSTC, where 5 to 20ms on-time is normal, but these are a bigger engineering challenge than a DRSSTC.

Regarding the driver, Steve Ward started out simple, as you can see on his schematics, but he also warns against using his older designs as they are not stable and lacks proper IGBT protection. Instead I really advise you on using the UD1.3, it is not as scary and complex as it looks, once you get to test it. http://stevehv.4hv.org/new_driver.html

The fiber optics, in the HFBR-0500Z series, avoid the HFBR-1523Z/HFBR-2523Z as they are only 40kbps, the others are 1Mbps and upwards.
The plastic fiber wire looks exactly like the plastic fiber I am using myself, its easy to cut with a sharp knife and handles easy without damaging it.

[Laci]

What kind of mosfets did you use for GDT output stage in the Kaizer DRSSTC I?

And the IC8 driver chip?Is the UCC27425P or the UCC27423P any good?

[Mads Barnkob]

What kind of mosfets did you use for GDT output stage in the Kaizer DRSSTC I?

And the IC8 driver chip?Is the UCC27425P or the UCC27423P any good?

It is IRF540 and IRF9540, I should add that to my website

The UCC MOSFET driver ICs are very good, sturdy and hard working, just do not ever run them without a load connected! That will risk destroying them. UCC ICs can however be a little expensive, so you could find cheaper MOSFET driver ICs with a lower peak current like MAX4420/29, but it all comes down to if you need a enable input on the driver.

[Laci]

Is that a problem if I use different cores for the CT and for the OCD?
2x B64290L674X830
2x TX42/26/13-3E25

[futurist]

It's completely fine 

[Hydron]

Is that a problem if I use different cores for the CT and for the OCD?
2x B64290L674X830
2x TX42/26/13-3E25

Those should work very well. I have used the first one you mentioned as a GDT core for my QCW coil - it has good properties for driving a high frequency coil (~350kHz) without too many turns (I used 5-6). The current transfomer core also should be fine - it's difficult to get them wrong.

[Laci]

So I've got through some bad days...Received some components,including 2x TX42/26-3E25 cores(like the other ones) for the OCD,some logic IC's + a MCP1403 driver,optical fiber cable(the transmitter & receiver is coming from ebay:HFBR1531Z + 2531Z)and 3xIGZ100N65H5XKSA1 IGBT replacement.
The IGBT fullbridge was made in a couple of hours.At one IGBT I've used 2xMUR1560's instead of the 2x 1.5KE440A-E3/54 TVS diodes,since one pair died previously.To test it,I hooked it up to the SSTC driver.The phasings looked correct,so the fullbridge was connected to a power supply of 18V,and to the TC primary.It was working quite well even at such a low voltage.The current drawn was about 0.6A-1.5A depending one the primary number of turns.Then I give it some more juice with the big 55V trafo.Guess what?The multimultimer's current display immediately started to show like 20A.I disconnected everything fast,but two IGBT died..The other two also not worked not for long;at 1Hz interrupted on rectified mains after some massive arcs the whole room lighted out.Another two died IGBT...Still not learning about my bads..Idiot!

Today I started the making of Steve Ward's universal driver.The logic part is done,the OCD looks working,because the LED lights up when the voltage reaches a higher level then the reference at the LM311.I'm interested if this driver can drive a SSTC,since I don't have more IGBT...I would like to change the primary CT to secondary base feedback CT,without using primary resonance capacitor.What do you think?And also what is your opinion about the OCD CT and about the primary CT?At the end same cores were used(because I ordered the parts from TME),wound the same way both.

[Laci]

So I received a HFBR-1531Z and a HFBR-2531Z optical sensors with MIKROE-1473 cable.My question is what kind of connector can be used to connect the cable's ends to the sensors?

[Mads Barnkob]

It is a shame that you keep blowing up silicon, but did you find out was it wrong? How did you determine that the GDT phasing is correct?

Your CTs look fine and its perfectly good to make exactly the same, you can even have them share just one core that goes on the primary circuit and then have two cores being fed from that one core. Like the red/black one here


I like your simple driver on the protoboard, just shows that easy, cheap and fast can work just as good as Chinese silver screen PCBs, nothing beat building something on the same day that you like to, instead of waiting for 5 weeks for shipping.

I think you need some of these connectors ( https://www.tme.eu/en/details/hfbr-4513z/toslink-transmitters-and-receivers/broadcom-avago/?brutto=1&currency=DKK&gclid=EAIaIQobChMI8-Xqi6iF2AIVlQ8YCh0flAl-EAQYBCABEgKu1PD_BwE ) for the cable, but im not completely sure exactly which fits your transmitter/receiver, as I have used the IFD95T that has a screw clamp quick connect on it.

[Laci]

About the optical fiber connectors:two of them should be bought-one for the transmitter and one for the receiver of the same type?

About the UD1.3b driver:are tantalum capacitors good for the gdt driver section's 10uF capacitors?

[Laci]

The new driver is almost done
The only thing that I am still not sure about is the type of the 10uF capacitors.Most of the drivers uses SMDs,some ceramic or tantalum.The THT tantalum capacitors are cheaper than the other THTs,and I want to use THT.
The driver circuit seems working,I get pretty good waveform out from my bad GDT.
I will buy four HGTG30N60A4 IGBT,then make a fullbridge again.A question about it:should I put two 440V TVS diodes in series,parallel,or a single one is enough?

*Additionally here are two screenshots of the waves,one at 175kHz,the other at around 600kHz.The current drawn is around 270mA with the GDT connected at 175kHz and around 240mA without GDT,so there probably is a problem.Anyways,as you can see it is a pretty good waveform from that poor core,not? Note,that no capacitor is connected from the mosfets to the GDT,just the 10 ohm resistor! *

[Mads Barnkob]

Yes, you need a transmitter opto on your interrupter and a receiving opto on your driver, selecting matching series is not that critical as long as they both can do the job.

The 10 uF capacitors are DC blocking capacitors and needs to be able to work at the frequencies you are driving the GDT at, so these should be ceramic or polypropylene film types.

If you make a low inductance layout of the bridge, TVS might not even be needed, it is more important to have snubber capacitors mounted as close to the IGBT as possible.

If the secondary side of the GDT is unloaded, it does not look bad at all, the first peak is a little large, but since there is no extra ringing I assume it is from being unloaded.

[Laci]

The GDT was unloaded,but it gets better with a small load.
What about the current consumption of the driver?Is that normal if it is driving around 300mA without GDT connected and just a little bit more,around 50mA more with the unloaded GDT connected?

[Laci]

I was working yesterday and today on the driver.Yesterday the only change I made was adding the 10uF tantalum capacitors on the GDT drive PCB.Then I tested it again,but it was not working.Around 600mA was flowing without the GDT connected.After a bit of analyze,the problem was obvious:the 7812 regulator(used 12V for the mosfet driver instead of the schematic's 9V) died,which caused the death of the mosfet driver too...I had that single two channel driver,so today I created a new mosfet driver,by UCC3732 1&2 and tested again.A current rush started,which turned off my PSU,due to safety.This time the 18V regulator died.Replaced,tested,died another regulator again...So I got the scope and started to check the gate of the mosfets.The N channel ones were working,the P channels too,but not together,even at just one side of the driver.I could not figure out what causes the problem,the mosfets are working,and if I make a similar circuit to a breadboard,there works the configuration.On the PCB when the P channel mosfet is connected to the N channel through the 0.1uF cap,the waveform gets distorted so much,that at a ringing the P channel completely switches off,then on.
Probably I will restart the driver PCB completely and make a new one.At least the logic driver with fiber interrupter looks working.

[Laci]

The problem got solved,an IRF9540 was dead.The new driver is working.
A question about the bridge:previously I had IGBTs which had no built in diode,now the HGTG30N60A4 type,which also do not have.Could this cause the failures with the previous ones?How can I protect the new ones?

[Mads Barnkob]

Good that you found the faulty part, I was not sure where to suggest that you started looking for errors on that one.

You could add external ultra fast diodes and they would be better than the internal diodes of IGBTs. The anti parallel diode provides a path for the current to flow when the opposite IGBT is turned on, without this diode that current in the inductive load could generate high voltage spikes, bad enough to kill your silicon.

Your bridge layout and heat sink seems fine. If you want even lower thermal resistance from IGBT to heat sink, cut the heat sink up into 4 separate parts and keep them isolated from each other and other things.

[Laci]

The diodes should be the same at every IGBT?I have some ultrafast,soft recovery diodes:2xBYT108,1xBY359,1xBY229.These are from an old CRT TV,and MUR1560 diodes.What about the TVS diodes?They have different purpose?