Very weird result using IR2110

[Neukyhm]

Hi there,

I have built an IR2110 operated ZVS Mazzilli driver. The idea is to make it work at 113kHz using a big 2kV 136nF capacitor and a ETD59 core instead of a CRT flyback.

I built the regular Mazzilli driver and it worked for 15 seconds until MOSFETs burnt. Then I thought about using the IR2110 instead of big and slow gate resistors, so I designed and built the attached circuit. Mosfets are both IRFP250N.

According to this blog the IR2110 can be used as dual low side driver. Today I have finally tested the circuit. I have to help it in order to begin oscillation, but it's working at >1MHz according to my oscilloscope. Do you guys have any idea why this is happening? It should be working at 113kHz; the other regular mazzilli driver worked at that frequency.

[ocset]

A load , as such, cannot be seen. Maybe the voltage on the drains of those fets is going above their maximums.?

[Neukyhm]

A load , as such, cannot be seen. Maybe the voltage on the drains of those fets is going above their maximums.?

Considering this is a transformer, the load is in the secondary, although I'm testing it without load, so the current should be minimum. This can also be seen in a PSpice simulation: without load the circuit will oscillate with very little current through MOSFETs

The voltage is not exceeding the 200v of an IRFP250N, I have measured it.

Maybe I should move to a real dual low side N-Ch MOSFET driver instead of using the IR2110, wich is a low & high side driver.

[Seekonk]

Did you power the two + pins of the driver with + supply?

[Neukyhm]

Did you power the two + pins of the driver with + supply?

Sorry I don't understand the question. I have two PSU. One is 12V only used to power the IR2110 and feed Mosfet gates. The other is 20-30V PSU for the LC circuit. Of course both PSU have common ground.

[Neukyhm]

In this application note page 21, the push pull circuit using IR2110 is very similar to my circuit except for the capacitor of a mazzilli driver. I don't know what went wrong 

Maybe I should try to use external PWM to drive the IR2110, although I like the way I designed it because the operation is like a mazzilli driver regarding auto-resonance.

[Neukyhm]

Today I have re- tested the circuit, this time without the IR2110 and using arduino instead. I have programmed arduino to send two train pulses A and B, being B opposite to A, then a TC4468 to lift voltage from 5v to 12v and then the mosfets.

Knowing the value of the capacitor and primary of the transformer, the resonant freq. is 113kHz aprox, so I programmed pulses A and B to be this freq.

The circuit worked, it sends a nice waveform even when going for voltages of almost 50V (causing drain voltages of more than 140V),
the mosfets endured well (gates have lots of ringing but it's understandable, the circuit is just a test with long paths).

Everything I have said above means that the IR2110 is failing someway, because the capacitor-inductor-mosfet circuit works fine.

[Neukyhm]

I have measured the voltage of the 12V PSU that is feeding the gates and IR2110 only, I found it to have much noise, see picture.

That voltage is from the PSU, it should be 12V DC exactly, so my PSU is really bad, or maybe the switching of the MOSFETs is causing transients that pull down the drains below 0v, so a voltage drop happens through the diodes. This voltage drop may be activating the UVLO so the IR2110 stops working. The circuit already has an electrolytic capacitor in parallel with a ceramic one just like the IR2110 datasheet recommends.

All the above is just a hypothesis. The fact is that there is someway a voltaje drop in the 12VDC PSU that, I think, is activating the UVLO.

[T3sl4co1l]

You can't substitute digital parts in an analog circuit and expect it to work.

The analog behavior of the original circuit is actually critical to its operation, even though it might not seem like it under steady state oscillation.

The improved digital equivalent, of the original analog circuit, is actually an OSC + PLL + driver.  Quite a bit more hardware, not to mention protective devices and such.

Gate bias resistors are easily solved with smaller resistors, better performance transistors (IRFP260 are ooooooold) or a modest analog buffer.  100kHz is not particularly taxing for this type of circuit.

Tim