a) Finding the reason of heavy ringing at the load condition: 'falling positive half wave'. My first place to search would be the bootstrapping diode for the supply of the high side driver. A series connection of two shottky diodes (because of voltage rating) usually is working very good there.
b) Trying if distortions go lower if we lower the gate drive resistor.
Gate drive loops appear to be short enough, so that I would not expect the limitation from the geometrie. Which valuor the gate drive resistor is used in the moment? 27R? ==> Stepping towards approx. 10R would be my first trial.
c) Moderate change of output filter. I.e. stepping to 7uH and 1.5uF, but keeping an eye on the ripple current stress of the e-caps. ...popping metal cans and electrolyte in our eyes is no fun.
d) Snubbers: They seem to work pretty fine, but make pretty some losses in the resitors to charge the 1nF each. It might be worth to experiment with 220nF and 15 Ohms. Note geometry of the snubbers has massive influence on the behavior, so this is always an iterative optimization.
Edit: 220nF are of course complete nonsense. 220pF would be a reasonable goal for IRF540Z.
Thanks to lorylaci for pointing out.
Last edited:
a) for +-44V He would not even need series connection, there are lot of 100V shottkys, there is no contionous high current there, so BAT41 or BAT46 should do the job even.
b) he tried decreasing the 27r resistor, it reduced distortion, but increased drastically current uptake
c) I suggested him to redesing output filter, even calcualted the proper values, and gave the equations. Changed values increased power output with constant distortion.
d) the high snubber values you suggest would cause intensive idle power loss, he already experimented with it, I gave him some advices according ti IRF Class-D tutorial.
..I thought there would be still 1nF in place?
So 220pF should reduce the snubber losses by far.
I tried using no snubbers a few days ago..nothing exploded
but audible distorsions have increased alot...Also with 470pf distorsions are bigger and with 2 nF are lower but there is to much disipation.
So for me 1 nF and 4.7 ohm seems a good compromise.
I am little afraid of using a 100V fet because i ca see peaks up to 20-30V above rail,will a 100 volt fet work up to 150 volt peaks?irfp540z works but i do not thing it is safe for long term usage..
but audible distorsions have increased alot...Also with 470pf distorsions are bigger and with 2 nF are lower but there is to much disipation.So for me 1 nF and 4.7 ohm seems a good compromise
.I am little afraid of using a 100V fet because i ca see peaks up to 20-30V above rail,will a 100 volt fet work up to 150 volt peaks?irfp540z works but i do not thing it is safe for long term usage..
Last edited:
I fully agree and the body diode of the IRF540Z is one of the best you can find!
Increased idle losses with lower gate drive resistor:
During idle we will get increasing shoot through peaks when lowering the drive resistor for turning ON. But at heavy loads this will help to speed up current commutation and consequently switching losses.
Lowering the inductance of the output choke is likely to reduce idle losses in such an adjustment.
CPX:
Which supply voltage do you have without load?
Can you stay with 100V devices?
In general you will find much better switches and diodes for 100V rather than 150V.
For the boot strap diode of the high side supply I got good results with t
For first, I also think you can keep the snubbering as chosen.I tried using no snubbers a few days ago..nothing exploded
So for me 1 nF and 4.7 ohm seems a good compromise
Mechanism of influence on distortion is not fully clear to me, because your wave forms do not show catastrophic ringing. But missing explanation from my side does not change your findings.
You are little afraid. I would be very afraid.
Usually a 100V MosFet will start to go in avalanche below 150V (the colder the earlier, funny...)
Depending on type they can handle more or less repetitive avalanche energy, but I would not dare to make use of that without very detailed examination of the related energy in the observed voltage peaks.
Redesing to full bridge could make him less afraid, because then he would not have to worry about bus-pumping effect.
Bus pumping.. so far we were already afraid without that.
But finally CPX plans to go for a bridged config, which also does not suffer from bus pumping.
CPX
Another easy to drive MosFet with 150V would be the IRFB4615.
Up to 25A at 400kHz it was performing well on my bread board.
Above that I left its SOA, that's why I changed to IRFB4115.
If you stay with 350kHz + reduced frequency at high modulation (typical for self oscillating) + a nicely sagging rail, then IRFB4615 could be an option for you. If you want to try them, I could send some to you.
Switching losses of paralleled MosFets:
I cannot clearly answer this. Never examined this in depth.
Basically it offers the chance of double di/dt, but also has to handle the Qrr of both body diodes...
hi dzony pls can you post your revised schema on the relay part . because i build the schema but relay didnt work until i grounded emiter of bc337 next to r6 and connected R6 to base of 337 . the relay then switched, but did not delay on powering up . With the temp protection, the temp led lights but the relay does not switch off.