Purpose of Capacitors Around Output MOSFET's??

[Ron AKA]

Could anyone explain what the purpose is of the two capacitors used around the MOSFET output stage on the Hafler power amps? Circuit drawings can be found at these links:

DH-200 (Page 13-14)

DH-220 (Page 14-15)

On the DH-200, it shows a C13 (390 pF) on the positive side going between the MOSFET Source(s) back to the Base feed. This is not shown on the negative side of the channel, but this may have been omitted for simplicity, as I recall I may have one there too. On the DH-220 (newer version) a capacitor is in the same location (C401) but increased to 680 pF.

What is the purpose of this capacitor?

Also on the DH-200 it shows a C18 (680 pF) on the positive side on the Drain. Again not shown on the negative side, but I'm sure it is there. On the DH-220 this cap is there as well, shown on both sides (C402, C408) and is a 0.1 uF film. They all go from Drain to chassis ground.

These caps are all hard wired right on the MOSFET's on the heat sink, not on the board. Interestingly the DH-220 has another 0.1 uF film (C21, C22) right on the board going from positive and negative rail to common. It basically seems to be doing the same thing as the film hard wired on the Drain.

What is the purpose of this capacitor? I'm assuming it is to kill noise on the voltage supply, but why two in close proximity?

Slowhand talks about needing these for stability in this post.

"The output transistors have power bypassing on the DH-220, while those in the earlier DH200 do not. These are 0.1mf, 100v mylar caps to chassis ground. For stability, you should add these if they are not present."

I don't think Slowhand has been around since 2006, so can you give your opinion?

And I know it is asking a lot, but of these capacitors which would you leave as is? Delete? or Add? I'm thinking of leaving the ones from Source to Base as is, as I don't have a clue what they are for. And I would replace the 680 pF mica power supply bypass with a 0.47 uF metalized polycarbonate as they are the same cost and size as a 0.1 uF MP, and will be using a number of them elsewhere.

If you can take the time to answer, all thoughts and suggestions will be appreciated.

[dimitri]

p- and n-channel MOSFETs have different input capacitance, this is an attempt to make them equal

[Ron AKA]

Originally Posted by dimitri

p- and n-channel MOSFETs have different input capacitance, this is an attempt to make them equal

You are right, I checked a photo of my amp and they are only on one side. The DH-200 and DH-220 use the same MOSFET's, so I wonder why they changed the value of this capacitance from 390 pF to 680 pF? Just a better estimate of how much is needed to equalize them?

Also, am I right in assuming that the power supply bypass caps should be on both sides? Again they only appear to put it on one side for the DH-200, but both sides on the DH-220.

[dimitri]

I wonder why they changed the value of this capacitance from 390 pF to 680 pF?

Gate resistors values are also changed

am I right in assuming that the power supply bypass caps should be on both sides?

Correct, but C18=680pF (DH-200) is not the power supply bypass, it is also for symmetry between p- and n-channel MOSFETs

[Ron AKA]

OK, thanks again. I'm not doing too good in my assumptions so far, but I'll try one more. Is the probable reason they went with same 0.1 uF on both n and p sides in the DH-220, that the 680 pF is so small it becomes insignificant compared to the +/- tolerance of the 0.1 uF power supply bypass?

[Ron AKA]

Thanks for the very helpful responses. Can anyone shed some light on how a 0.1 uF film cap located right next to the power supply to the MOSFET could prevent instability?

[MagicBox]

That's nothing but standard supply decoupling, typical wima 100nFs are ideal for that. It's recommended when working with opamp packages as standard practice. With mosfets it's no difference. The caps essentially provide a small reservoir for instant current supply. That plays no role for normal frequencies but it does for the high frequency loop regulation.

Edit: normally you don't need them with mosfet output stages, unless you manage to create a fast output stage that is able to follow input up to a few MHz.

[Ron AKA]

Originally Posted by MagicBox

That's nothing but standard supply decoupling, typical wima 100nFs are ideal for that. It's recommended when working with opamp packages as standard practice. With mosfets it's no difference. The caps essentially provide a small reservoir for instant current supply. That plays no role for normal frequencies but it does for the high frequency loop regulation.

Edit: normally you don't need them with mosfet output stages, unless you manage to create a fast output stage that is able to follow input up to a few MHz.

The amp is said to be -3 dB at 100 kHz. Obviously this is not an audio issue, but I suppose it is not good for the MOSFET's if they become unstable when faced with a very high frequency. Thanks for the help.

[MagicBox]

Yeah, the amp is most likely limited by a LP at the input, however the control loop operates at much higher frequencies and if designed correctly, the amp can stabilize transients at speeds of several MHz. If any stage is not sufficiently fast, or starts to phase shift while a previous stage is faster, it will turn up as oscillation. Those bypass caps sort of make the mosfets faster as they are able to draw that instant current. If they can't, previous stages want to start overcompensating and throw it into oscillation. One will have to either slow down the previous stage or somehow speed up the next stage.