First DRSSTC 1.3b DRIVER

[Alex_1996]

Hello! This question may be too quick to answer and at the moment it is the only one I have. Maybe in a few days I will have more questions, you know how it works 

I am building a small DRSSTC, and for this I am using a half bridge from FGH60N60SMD and a driver and switch from Aliexpress (I attach picture of them).

I have tested the system with a dummy load (a metal can) and the system oscillates when I test it at 50VAC input. The can gets very hot. The system consumes at that voltage about 350W with the pulse width at maximum.

To cut to the chase, I was interested in measuring the voltage of the OCD resistor, which in this case is 10 ohms, and thus know how much current was flowing through the circuit using Ohm's Law. My TC's are 1:25:25, total 1:625. I limited the OCD comparator voltage to about 4.5V so that 300A max. When I tested the system at 50VAC and measured with my multimeter the voltage on the 10 ohms resistor of the OCD, the multimeter on DC voltage scale marked 0 volts and the voltage on the comparator also marked 0 volts, something very strange, because if the diodes at the output of the CT of the OCD rectify the AC current, my multimeter should throw some voltage value from the 10 ohms resistor of the ocd and it does not do it.

That has led me to think that this driver, which looks like a 2.7 (the manufacturer hardly gives any data), is in bad shape.

Should the OCD resistor of a DRSSTC driver give a DC voltage reading? And this could be measured with a multimeter / voltmeter on DC voltage scale?

I hope you can help me, so I can implement another driver or revise the one I have  :'(

Thanks a lot

[davekni]

Should the OCD resistor of a DRSSTC driver give a DC voltage reading? And this could be measured with a multimeter / voltmeter on DC voltage scale?

Meter is measuring average voltage.  OCD trips on peak current (converted to voltage across 10 ohm resistor).  For a full wave rectified sine wave, average is 2/PI times peak.  If interrupter duty cycle is 1%, meter should read 0.01 * 2/PI of peak voltage.

[Alex_1996]

Thank you very much for your quick response

It is a subject that I was curious about, so I really appreciate the explanation

Best regards 

[davekni]

Thank you very much for your quick response

You are welcome.

BTW, It is also possible that the driver board OCD circuitry is not just like UD2.7.  I've seen designs, at least one from EVR, that place the burden resistor directly across CT secondary, with bridge rectifier after burden resistor.  Minimum current limit is then 2 diode forward voltage drops across burden resistor.  And of course, voltage across burden resistor is all AC, no DC.

[Alex_1996]

Hello, you are right! It had passed me by that there are two types of OCD circuits

I will keep all this in mind for this DRSSTC design and for other future designs 

Thank you very much for your explanation

[Alex_1996]

Hello, I have not found another post similar to the question I am asking.

I am building my first DRSSTC. Right now I am using the 1.3b driver schematic as is, without making any modifications. The only modification, so to speak, is the use of UCC27425 instead of UCC27423 as originally used in this driver.

When I feed the driver I can't get it to work. The GDT does not hum at all. In testing, I found that it only works if I electrically connect pins 3 and 4 of the 74HC14N externally. If I do the latter, the GDT does emit a hum. I have tested the OCD with this mode of leaving pins 3 and 4 of the 74HC14N electrically connected externally and when I supply a voltage higher than the OCD limit voltage, the red LED lights up, so it seems to work.

I don't know what is happening. The integrateds are new and original. I don't understand what could be happening.

I have followed the 1.3b driver schematic as originally designed.

I hope you can help me. Attached is an image of the driver

[klugesmith]

Did you pay attention to logic polarity difference between UCC27423 and UCC27425? 

[Alex_1996]

Did you pay attention to logic polarity difference between UCC27423 and UCC27425?

Thank you for your quick response

As I saw the internal diagrams of both UCC's I didn't seem to see any difference between them.

Attached is an image of the diagrams of the UCC27423 and UCC27425.

If you can explain a little more I would appreciate it

[Alex_1996]

Did you pay attention to logic polarity difference between UCC27423 and UCC27425?

Thank you for your quick response

As I saw the internal diagrams of both UCC's I didn't seem to see any difference between them.

If you can explain a little more I would appreciate it

Okay. Checking as you told me, I have seen that in the UCC27425 pins 4 and 5 are connected by an IF gate, while in the UCC27423 they are connected by a NOT gate.

That is why when I connect electrically externally pins 3 and 4 of the 74HC14N the driver works completely, since I don't deny the signal coming in through pin 3 of the 74HC14N. Not being denied, it enters without problems through pin 4 of the UCC27425 and continues without being denied, leaving through pin 5 and going to one of the Mosfets.

I hope I understood it well

[Alex_1996]

Hello everyone!

After several months of troubleshooting, learning about relevant parameters in the construction of a DRSSTC and asking silly questions, I have managed to complete my first DRSSTC.

The design parameters are as follows:

Secondary frequency with toroid: 147.5 kHz.
Primary frequency: 120 kHz
Primary inductance: 12.59 uH
MMC bank: 143 nF
Detuning: 18.6%.

I am using a full bridge with 4 TO-247 FGHL75T65MQDT IGBTs, whose ICM is 300A. The capacitor capacitance of the DC bus is 2000uF (I use two 450V / 1000uF capacitors in parallel).

I power my DRSSTC with a 3kVA variac. The maximum voltage before OCD is activated is 129VAC (OCD set to 317A). This supply voltage point I reached after performing numerous fine tuning tests. Before, for example, the OCD at 317A would activate at only 100VAC. With grounded discharges, at 129VAC (the maximum before turning on the OCD), the consumption from the mains is 1378W. The maximum length of discharges is 70cm

And, after all this introduction, here comes my main question. Since these IGBTs withstand up to a 300A ICM, would it be possible to raise the current further, for example to 400A? The FGH60N60SMD are known to carry up to 300A, when their ICM is 180A. It is also the same with other TO-247 IGBTs, they go up more current than their maximum ICM.

In my case, for a 300A ICM I am setting my OCD to 317A, which is a conservative point to avoid bridge explosions. I would like to be able to raise the max current further and get 1 meter discharges, would that be possible with my IGBTs?

My interrupter has these parameters: 258uS / 300 bps (max)

Attached is the datasheet of my IGBTs and a long exposure photo of the DRSSTC in action. Thank you very much for your help

[Mads Barnkob]

And, after all this introduction, here comes my main question. Since these IGBTs withstand up to a 300A ICM, would it be possible to raise the current further, for example to 400A? The FGH60N60SMD are known to carry up to 300A, when their ICM is 180A. It is also the same with other TO-247 IGBTs, they go up more current than their maximum ICM.

In my case, for a 300A ICM I am setting my OCD to 317A, which is a conservative point to avoid bridge explosions. I would like to be able to raise the max current further and get 1 meter discharges, would that be possible with my IGBTs?

My interrupter has these parameters: 258uS / 300 bps (max)

Attached is the datasheet of my IGBTs and a long exposure photo of the DRSSTC in action. Thank you very much for your help

Congratulations on getting it to run and to run great! 70 cm is not bad at all for a full-bridge of TO-247 IGBTs.

I have had stable TO-247 based coils run at 300'ish A, but going to 400-500A, I only see burned out IGBTs. While the Icm ratings might be better, the die size and heat sink capabilities remains the same for TO-247 devices. You need to calculate the power dissipation and see if you can remove it fast enough. I walk through those calculations in the IGBT guide: https://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/igbts/

[Alex_1996]

Thanks for the quick response!

Doing some tests I have detected that the OCD resistance had a wrong value and it was due to a failure in the multimeter reading. This has meant that I have been running the coil at 378A, as the OCD had it set to 378A when I thought I had it at 317A!

I have also been putting the coil on a table, which has greatly improved the performance. I think I have reached a better tuning point by raising the coil. Before, at 129VAC it was already turning the OCD on at 378A (and me thinking it was set at 317A). Now, with the coil raised, I was able to go up to 160VAC and the OCD at 378A still wouldn't turn on, the lightning reached the ceiling (80cm)! In fact, at that voltage, the 13A fuse I had to protect the input has blown. I thought the Full Bridge had died. Thank goodness it didn't 

In short, the fghl75t65mqdt will hold 378A OCD. I think after this experience I will lower the OCD to 320A to make it more reliable so I can enjoy the DRSSTC.

[davekni]

In short, the fghl75t65mqdt will hold 378A OCD. I think after this experience I will lower the OCD to 320A to make it more reliable so I can enjoy the DRSSTC.

One more key parameter:  fghl75t65mqdt survived 378A at 120kHz.  Switching losses are proportional to frequency.  Higher frequency coils likely wouldn't achieve 378A, at least not without phase lead.   Under specification test conditions, fghl75t65mqdt turn-off switching loss is about half of turn-on loss.  Adding phase lead (ie. UD2.7 instead of UD1.3) might allow similar IGBT performance at 240kHz as you experience at 120kHz.  Even at 120kHz and 378A, switching loss may be half of total power dissipation.  Changing drivers might provide as much margin as dropping to 320A.

Congratulations on a nicely performing coil!

[Alex_1996]

Hi, thanks for the explanation!

No doubt, when I had the OCD at 378A it was unbelievable: the discharges reached the ceiling! For my first DRSSTC I was very happy. But I understand that maybe I'm pushing those TO 247 IGBTs to the limit.

Today I set the OCD to 320A and the discharges are somewhat smaller, but the Full Bridge will have longer life.

Up to how many amps would the fghl75t65mqdt carry to have safe operation and good life? Granted they have held 378A, but I think every time I set the OCD to that amperage I'm one step closer to the death of the IGBTs.

Best regards!

[davekni]

Up to how many amps would the fghl75t65mqdt carry to have safe operation and good life?

There is no way to know with any certainty.  Depends on many factors including actual use duty cycle, heatsink cooling and thermal resistance of any insulating pads you have between IGBTs and heatsink, delay through UD1.3 (total phase lag).  For my DRSSTC with 40 IGBTs, I tested a couple parts to destruction first to get some idea of margin.  DRSSTC use is not typical, so not well covered by data sheet details.
BTW, download rev3 of the data sheet.  Rev2 that you posted has thermal impedance graph time axis off by 10x.  Makes it look like an incredible IGBT per rev2 spec.
If you want to push IGBTs hard, phase lead as I mentioned, and good heatsinking.  I use electrically-live heatsinks, one per IGBT or set of parallel IGBTs.  That allows use of conductive silver-filled thermal compound and direct contact of IGBT case with heatsink.  And apply mounting force directly over IGBT die, not with just a screw through the hole.

[Alex_1996]

Hello! I bring you news of the coil 

Having put the above tips into practice, I decided to set the OCD to 350A to drive my FGHL75T65MQDT full bridge with some safety. Everything was going well until I started to notice that, after about 15 seconds of operation at 125VAC, with a pulse width of 150us and 270 BPS, the coil started to discharge with less intensity, as if it was losing power automatically (the coil has been working under those parameters correctly with a 378A OCD, 350A should not have been a problem). There was a drop in performance. Then, I disconnected everything and reconnected it again and the strange phenomenon was happening again.

The third time in a row that I did this, while raising the voltage on the variac again, the coil stopped working. I thought it might have been some fuse or something, but everything was fine. I checked the diode bridge and it was also fine. The IGBTs were also all fine.

Since I didn't detect any apparent faults and since the iGBTs I checked and they were good, I thought maybe some kind of degradation had occurred inside the FGHL75T65MQDTs. Therefore, I built another full bridge, this time with 8 FGH60N60SFD (I had them on hand from the workshop). This way, it would be more robust as it would be a parallel connection and I could even take it to 400A OCD (I attach pictures of the bridge).

Today, I have connected the new bridge with the 8 IGBTs and nothing happened... until... After removing the voltage from the variac (because the coil was not working again), I disconnected the 1.3b driver, reconnected it quickly and the coil released a short discharge. I repeated the sequence. I turn on the interrupter, then the 1.3b driver, then the variac, and nothing happens. I disconnect the variac, disconnect the 1.3, reconnect the 1.3 and the coil quickly discharges and turns off. I thought it might be due to the FGH60N60SFD's being fake, but I checked them by testing and they seemed to be fine.

Since the same thing happened with the FGHL75T65MQDT bridge and with the new 8 FGHL60N60SFD, I started to think that maybe the 1.3b driver could have a bug. Even so, I have checked the output of the GDT with a test feedback signal and the waveforms looked correct. I'm desperate because I don't know what could be happening. I think I'll even solder a new 1.3b board with all the new components (hopefully we can fix the problem here :'()

In summary:

1º) Coil works well at 350A OCD, but suddenly it starts to lose some power until it doesn't work anymore on subsequent starts
2º) I check the full bridge of FGHL75T65MDQT, fuses and diode bridge and they are fine. Regardless, I make a new bridge out of FGH60N60SFD (I had them on hand from the shop). I check first that they are not fakes...
3º) I connect the new bridge of 8 FGH60N60SFD and the coil still doesn't work. When I unplug the variac, and quickly turn on the 1.3b driver, the coil fires a brief discharge (this doesn't always, just sometimes).
4º) I go crazy and I think it's because of some fault in circuit 1.3b.... But I can't detect where. I have checked the feedback diodes and they are fine

I don't know what could be happening. Hopefully we can fix it. I attach photos of the coil discharging before it stops working, the new full bridge and the complete set. Thanks a lot!

[davekni]

Does sound like odd symptoms.  Perhaps something in interrupter optical input or OCD circuitry.  Perhaps the OCD resistor failed (ie. cracked from thermal stress).  Or it could be an open connection (ie. cold solder joint) within UD1.3.  When a CMOS input pin is open (not driven due to bad connection), it may be floating to one state immediately after power up and then drifting to another shortly after.

I presume that SMPS in the back right is for UD1.3 power.  It might be getting confused by 120kHz fields and shutting down.

Do you have a scope?  Or just DMM for testing.

[Alex_1996]

Does sound like odd symptoms.  Perhaps something in interrupter optical input or OCD circuitry.  Perhaps the OCD resistor failed (ie. cracked from thermal stress).  Or it could be an open connection (ie. cold solder joint) within UD1.3.  When a CMOS input pin is open (not driven due to bad connection), it may be floating to one state immediately after power up and then drifting to another shortly after.

I presume that SMPS in the back right is for UD1.3 power.  It might be getting confused by 120kHz fields and shutting down.

Do you have a scope?  Or just DMM for testing.

Hi! I have a very bad oscilloscope from Aliexpress that barely measures well, but at least it lets me intuit the output waveforms of the GDT with a test signal (120kHz). Most of the tests I do with the digital multimeter (to check components)  :'(

What is so strange to me is that the coil was working fine (for the first one I build) and then suddenly lost power on the discharges until it just stopped working. When I do the tests, the driver lights are on. The OCD has now been set to 100A for testing and still no coil discharges. I don't understand how a driver can work and then stop working. I don't know which component of the 1.3b is the one that is failing (assuming that is what it is) 

The 24VDC power supply is grounded right out of the box and the coil worked fine. I don't know if some integrated has deteriorated or something? It is surprising that everything worked fine until I set the OCD to 350A. Now, even if I set it to 100A, it still doesn't work. And super strange that when I turn off the variac and quickly turn on and off the 1.3b a small beam comes out of the coil 

I attach a couple more pictures about the 1.3b driver inside its metal case and the complete set of connections (the TC wires are not well stranded, but it's a picture from the early days, that's why it has also the old FGHL75T65MQDT jumper)

Thank you very much for all your help!

[Rafft]

Why build 1.3 when there is 2.7 w/c is still possible with thru-hole parts. I even built mine on a protoboard.

And try to get a decent scope. It is needed in debuging stuff

[Alex_1996]

Why build 1.3 when there is 2.7 w/c is still possible with thru-hole parts. I even built mine on a protoboard.

And try to get a decent scope. It is needed in debuging stuff

Hello!

Basically because it's the first time I build a machine like this and I wanted to learn from scratch

When this coil is working again I will try to raise money to buy a good oscilloscope, but for the moment my financial means are limited  :'(

Still, I am happy to have achieved 70cm of discharge on my first DRSSTC with almost no lab equipment and no explosions   

Thank you very much for your advice