MOSFET heating on my amp...

[thefish]

I finished the second revision of my class D amplifier (schematic) but there is a lot of MOSFET heating. If I remove the 0.1uF caps from the power supply there is less heating but more overshoot.

[IMGDEAD]http://www.siliconlite.com/images/AMP/Board-A.jpg[/IMGDEAD]

High and Low side gate signals with power supply caps:
[IMGDEAD]http://www.siliconlite.com/images/AMP/gates_w_cap.jpg[/IMGDEAD]

High and Low side gate signals without power supply caps:
[IMGDEAD]http://www.siliconlite.com/images/AMP/gates_wo_cap.jpg[/IMGDEAD]

There is also a good deal of hissing when the input is muted (this was not in revision 1).

 

[thefish]

Anyone have any ideas why this would be happening? Should I use a larger cap on the power supply (maybe 1uF) connection to the FET's? I put a FM radio right next to the amp (withen 6 inches) and get no interference. When I actually rest the FM radio on top of the amp it interfears a lot (even makes the digital dial jump around??).

When the power supply caps are removed it seems that just one of the low side mosfets gets hot fast, the rest are ok.

 

[Pafi]

Do you really use IRF510 as power switch?

Should I use a larger cap on the power supply (maybe 1uF) connection to the FET's?

Yes, definitely, as close to FETs as possible, but probably this won't solve this problem.

 

[thefish]

Do you really use IRF510 as power switch?



Yes, definitely, as close to FETs as possible, but probably this won't solve this problem.

The schematic is in error, they are really IRF540's as the power switch. Also for testing the amp is only running at 13.8V right now, not 50V as designed.

 

[Pafi]

Then you can try decreasing Rg.

 

[thefish]

I swapped the 0.1uF caps with 1uF caps and the traces look much nicer now but it still had the heating.

[IMGDEAD]http://www.siliconlite.com/images/AMP/gates_w1uF_cap.jpg[/IMGDEAD]

Searching through my resistors I dont have any smaller ones for the gate I am thinking maybe 10ohm should do it.

 

[fredos]

Simply more dead time...You did'nt see the overlap on your scope???

 

[lumanauw]

Hi, Thefish,

I see in the picture, your core is green/blue, in magnetics catalog it is type -52, usually for low frequencies (like type -26 yellow/white).
Have you try type -2 core (red/clear) ?

 

[thefish]

Simply more dead time...You did'nt see the overlap on your scope???

I do see the overlap, but on my other amp (revision 1) with the same dead time I get 0 heating with the same load and same input.

 

[thefish]

Hi, Thefish,

I see in the picture, your core is green/blue, in magnetics catalog it is type -52, usually for low frequencies (like type -26 yellow/white).
Have you try type -2 core (red/clear) ?

I need to order some new cores, I do have some type 12 cores...

 

[theAnonymous1]

Wouldn't type 12 be at the other end of the spectrum, i.e. too high frequency(20-200mhz)? They would also require almost twice as many turns as type 2.

 

[thefish]

Wouldn't type 12 be at the other end of the spectrum, i.e. too high frequency(20-200mhz)? They would also require almost twice as many turns as type 2.

I think that is right, that is why I did not use them, I really should \order the correct cores.

 

[Pafi]

Searching through my resistors I dont have any smaller ones for the gate

What about 0 ohm? Or a diode parallelled with 30 ohm. You can try decreasing the voltage of gate driver also.

 

[Gyula]

12 Ohm paralleled with 1n5819, driven by IR2110 would be correct, even without dead time. Is the deadtime 1.2 us now? I suggest you the antiparallel diode in order to avoid switch-on through the Miller-capacitance too.

 

[thefish]

I found some old 10ohm 1/4 watt axial resistors and made them fit into the surface mount spots now there is zero heating. I upped the power supply to 19V with a 6ohm load and it is staying cool to the touch.

 

[thefish]

Well I have let it run for 6 hours with a 19V supply at 6 ohms and it is cool (only the inductors are a bit warm). It sounds nice as a sub amp, for mids and highs it is ok, no real hissing. Does not seem to cause any FM interference unless radio is within 2 inches (and parallel to the doughnut face) of the inductors. The output is not very nice looking on the scope 2V p-p noise on the signal.


Scope image at 5V/div with a 100hz test signal input into amp:
[IMGDEAD]http://www.siliconlite.com/images/AMP/Output_5V-Div.jpg[/IMGDEAD]

I think changing the output inductors should clean this up. Thanks for the help everyone.

 

[Pafi]

Output waveform is quite good, considering that there is no feedback! Residual is normal, you can leave it, but you have to change inductors in order of less heating (of inductor). Try Arnold MPP cores, or Micrometal -2, -14!

You're welcome!