SG3525 Half bridge battery charger- MOSFET blown issue.

sbdada09

New member
Hi everyone,
Currently I'm trying to make an half bridge converter with SG3525 and gate drive transformer. The converter is intended to charge a 60V lead acid battery pack.
Here is the schematic: View attachment SG3525-battery-charger.PDF

Charging control and protection circuit is not complete yet, so they are not mentioned in the schematic. Startup section is not also attached yet, currently the driver is running from a separate source. Feedback network is also open.

The final circuit will have IRFP460 MOSFET, but currently I'm checking with IRF740. Bellow are the gate signal. These were captured while the main power is off, only the driver chip was running. At first I made a basic gate driver circuit. Only a 10 Ohm gate resistor was connected in series after the GDT. But the signal was too slow, raise time was more than 600ns. Then I attached the diodes and transistors, and the signal improved. Don't know if it is sufficient though.




After assembling everything, I've powered up the converter and that was a complete fail.
  • First I've charged the primary bulk capacitor while driver was off, then unplugged the converter, and started the driver. It seems everything is ok, the converter run and tried to push some voltage to the secondary, unless the bulk capacitor discharged.
  • I did it several time to make sure that everything is working.
  • After that I powered up the converter and started the driver while power is on. Upper MOSFET shorted instantly.
  • After changing the FET, I did it again and that MOSFET shorted again.
  • After changing the MOSFET for the third time, I lowered the input voltage to 90VAC with a variac. Everything run perfectly. No MOSFET blown or shorted.
  • Then I've increased the input to 150VAC then 170VAC and tested every time. In 170VAC, I've loaded the converter with around 100W load, nothing went wrong. During the test, I was measuring the Vds of high side FET. it was nice and clean, no noise or spike was present there.

Finally I've moved to full 220VAC and it failed again. But this time I noticed some interesting things.
  • This time the converter didn't failed instantly, survived a couple of seconds.
  • A 100W lamp was connected in series with the main supply from the beginning of testing. An oscilloscope probe was also connected to drain and source of the high side FET to measure the Vds.
  • After starting the driver that series lamp started to oscillate roughly around 0 to 25% visual brightness. It didn't happen before. That Vds on the oscilloscope also indicated the oscillation. Didn't managed to measure though.
  • Within 2-3 seconds I heard a pop sound and everything stopped, series lamp indicated a full short circuit.
  • I measured everything again, these time all two MOSFET along with those two pnp transistor has shorted.

Can you please help me to solve this problem?
I've no more IRF740 or other high voltage FETs remain and have only two IRFP460 in another running converter. There is no way to get them now due to the current Corona virus lock-down situation. So I don't want to destroy them. But I've a lots of IRFP4321pbf (150V, 78A), so if you need any scope data while running on low voltage, I can provide that.

Here is my transformer details:
Main transformer:
  • Core: EE55
  • Primary: 9 Truns, 28SWG X 15
  • Secondary: 7 Turns, 28SWG X 15
  • Auxiliary: 1 Turn, 28SWG X 1
  • Airgap: none.

Gate Driver transformer:
  • Core: EE25
  • Primary: 20 Truns, 28SWG X 2
  • Secondaries: 20 Turns, 28SWG X 1 X 2 (Bifilar)
  • Airgap: none.

Thanks in advance :)
Regards,
Sukdeb Biswas.
 
Last edited:

Silvio

Well-known member
What I can see for the moment is that in both shots of your scope your running frequency is different one is 37 Khz and the other is around 50khz. According to the timing capacitor and resistor used the 50Khz frequency is rather the correct frequency.
I do not know the core material of your transformer but I wound one recently for 65Khz and the core material is PC40. The number of turns turned out to be 10 for the primary with a flux density of 1600 guass. My guess it could be the reason that your transformer lacks turns for the frequency used. and may it being at the point of saturation.
Your remedy for this problem may be that you need to increase the frequency to around 80 to 90 Khz so that you make up for the missing turns. I also suggest that you increase a bit the discharge resistor to widen a little the dead time. change the 47ohm resistor to 100 ohms. Check the dead time and see that is not less than 1uS. A lot of diyers usually use 12 turns primary for an EE55 trafo running at around 60Khz. It may also be the case that they do not know the material and use 12 turns to be on the safe side. One other option is to increase the primary winding turns.

For me its making sense that at a lower voltage everything seems to be working fine but as the voltage is increased every thing start to screw up. :)

Good luck and I hope you solve your problem.

Regards Silvio
 
Last edited:

sbdada09

New member
Thank you Mr. Silvio for your kind attention. That 37 hz or 50 hz is not the driver frequency. Those are showing because of the trigger, you can see that trigger courser (don't know the actual name) right corner of the screen resting on the bottom line of the signal. There are some low frequency ripple or noise on the bottom line and that trigger measuring that noise. The actual frequency is around 66.4Khz all the time. Dead time is around 500ns, I'll increase it.
I really don't know the core material, it's a cheap store bought core. So, increasing frequency or increasing turn, which one you suggest?
Also please see the signal raise time (on 2nd picture). Is that sufficient to drive IRFP460? If so, then I'm ready to sacrifice my last pair of MOSFETs :)
Regards,
Sukdeb Biswas.
 

Silvio

Well-known member
Well ok about the switching frequency. What I suggest for you is that you put a series lamp (100W) in the input of the high voltage rail. This will prevent the stored energy in the capacitors destroying your fets in the event of something goes wrong. Even if you had to make a cut in the pcb for that. The Lamp will show you if you are good or not.

As far as I can suggest now that you tell me the switching frequency is 66khz well I guess you better increase the primary turns. As I said if the core material is not known than a 12 turn primary will be adequate. Well you have to adjust the secondary winding accordingly.

IRFP460 has a larger input capacitance and will need a little more charge during switching. You can monitor the the gate pulse with the scope and see how the rise time behaves. BE CAREFUL DO YOUR TESTS ON THE LOW SIDE ONLY OTHERWISE YOU MAY BLOW YOUR SCOPE. See that either your smps or the scope are isolated.

Regards, Silvio
 

sbdada09

New member
Hello Mr. Silvio, as you suggested, I've re-winded the transformer. Now the primary has 12 turns, and secondary has 10. Also changed the discharged resistor (Rd) to 150 Ohm and connected 2 series lamp. Now the basic diagram is like this:



You can see that I've removed the transistor and other component from the gate drive section and made it simple again. Because in that transistor configuration with the IRFP460 MOSFET, the gate and drain signal was a nightmare. With a 100 or 150 Ohm resistor on the Rd, both light bulb glows, which indicates a cross conduction. And with a 470 Ohm Rd, bulbs didn't glow, but the signal took around 500ns to raise and more than 1us to fall. That signal wasn't clear also, lots of ringing and bounces was present in both gate and drain.
Now with this simple resistor and diode on the gate, the signal looks pretty clean. It takes around 200ns (more or less) for both raise and fall (from 600mV to 10V or vice versa). Have a look:




These tests were performed at 220VAC and those two light bulb was connected all the time. So far, everything is good, nothing blown up yet.
Is these signals looks good or fast enough? Is there anything I can do to make it even faster?

Thank you again Mr. Silvio.
Regards,
Sukdeb Biswas.
 

Silvio

Well-known member
Dear Sukdeb, You can see that when using a pulse transformer the wave shape will not always be so pretty. This depends on how well this GDT is made. Noticing also the pulse arriving to the gate it seems that there is a small glitch during the rise time but I think it is quite OK and there should be no trouble. The actual wave at the drain source is not bad at all. Measure out the total switching time and see that your dead time is adequate so that it compensate for the rise and fall times, propagation delay etc. This would eliminate any shoot trough. The dead time should be just enough for proper switching.

If you like to experiment a little I usually put a 1uF capacitor (film type low ESR) in series between the totem pole and the GDT primary rather than direct like have in the original circuit. This will help in the event something goes wrong as it blocks the DC path. I would like you to try it as this may enhance your rise time. Try some other values like 2.2uf or maybe 4.7 uf etc. See what works best. (Give me some feedback on this)

One other thing that I noticed is that the base resistor feeding the totem poles to my opinion seems a bit low being 10 ohms. The more load the SG chip has the more the waveform will deform. I guess 47-100 ohms will be more adequate. SG3525 is only capable at 200mA max on its outputs.

Your next step is to load a little the smps and see for any flaws or deformation in the wave forms. If you find that it all ok then remove the current limiter lamps and load a bit more gradually while monitoring the wave forms at the same time. If things do not change than you are there.

Good luck Regards Silvio.
 

Silvio

Well-known member
Here are some test that I have been conducting. I made myself a GDT for a full bridge smps and I was trying to find the best way to get a good waveform.

The schematic below of my setup. You notice a snubber made up of 10nF and 47ohm resistor to tame down any spikes generated by the GDT itself.

I am sorry for the orientation of the pics. There is no way I can arrange them once they upload.

The first scope shot shows the waveform from the GDT with no load and the other shows the waveform with 600mA load



20200414_155625.jpg20200415_184654.jpg20200415_190118.jpg20200415_184732.jpg
 

stewin

Member
Here are some test that I have been conducting. I made myself a GDT for a full bridge smps and I was trying to find the best way to get a good waveform.

The schematic below of my setup. You notice a snubber made up of 10nF and 47ohm resistor to tame down any spikes generated by the GDT itself.

I am sorry for the orientation of the pics. There is no way I can arrange them once they upload.

The first scope shot shows the waveform from the GDT with no load and the other shows the waveform with 600mA load



View attachment 7232View attachment 7233View attachment 7234View attachment 7235

hi silvio ,
how many turns did you use and can ee16 be used in trifiliar winding with those type of insulated wire , as a halfbridge gdt ?
 

sbdada09

New member
Hello Mr. Silvio, Your signal looks very clean. Did you captured that without powering the main transformer? I found that if MOSFET is turned on with transformer, the signal deform. look at these two:




In my case, that snubber of GDT helps to flatten the top of the signal a little bit. It also decrease the raise time and increase the fall time by few nano seconds.
Didn't get the chance yet to test the other things you suggested. Hope I'll do it today and inform you.
Regards,
Sukdeb Biswas.
 

Silvio

Well-known member
Hi Steven, I do not think CAT 5 wire is siutable to fit in a ee16 trafo. Each wire has a dia of 1mm and the copper inside is 0.5mm. The reason I used this kind is to have more security regarding insulation
The volages in FB are in excess of 300v.
 
Top