Mosfet RF

Archived from groups: rec.audio.tech (More info?)

In article <cnt2gj$k7n$1@titan.btinternet.com>,
"Gareth Magennis" <sound.service@btconnect.com> wrote:

> Maybe Pooh Bear knows about this?
>
> I have a Mosfet amp I am repairing. Each side uses 6 2SJ56's and 6
> 2SK176's. I have found 4 dead ones, 2 of each per side which I have
> removed. One side has RF bursts on the positive going output signal, only
> under load. (4 ohm resistor network). They don't start until the output
> voltage hits just over 0.2 volts and don't get any bigger as the output
> increases, they are just added to the signal. You can see the bursts on all
> the earth points, even though the earthing and power supply systems seem
> very well implemented.
>
> The funny thing is I was getting exactly these symptoms on both channels.
> When I removed the dead Mosfets from one channel, the RF dissapeared from
> that side and it tests fine now. On doing the same to the other channel,
> they are still there. I have tried grounding the gates of individual
> mosfets but the amp doesn't like it and goes into protect.
>
> My question is, could this be due to another damaged Mosfet? There are no
> Zobel networks on this amp, just an inductor coil after the output relay,
> but it seems pretty well made and has 2 massive mains toroidal transformers.
> (Its a PIC Systems Ltd 20 30). Also, can I run the amp with 2 missing
> Mosfets per side and expect it to behave properly? I'm doing this because
> the devices cost 10 quid each and I need to know the repair is viable first!
>
> I know from past experience that Mosfet Amps can be rather weird with RF,
> hence this post.
>
>
> Thanks anyone,
>
> Gareth.

MOSFETs can get weird with RF because they're so damn fast. You can
solder capacitors across all three leads and some will oscillate in the
VHF spectrum using lead inductance.

It's possible that one of the MOSFETs has a damaged gate that's reducing
the bias voltage. It's typical for each MOSFET to have a series
resistor on the gate. It dampens RF oscillations and it makes it easier
to find blown gates. If any resistor has even the slightest amount of
voltage across it, the gate is fried. If they're all good and you're
sure the bias is correct, try increasing the series resistance to the
gates of the new MOSFETs.