MOSFET follower driver oscillation

[Sheldon]

I've been using a DN2540 follower as a buffer/driver for an 801A2 amp. Recently, I wanted to see if replacing the source resistor with a CCS would make a difference. Not sure if I hear a difference yet, but I have observed that if I leave the input to the amp open, the bias on the output tube starts to go up quickly. Doesn't happen with the amp connected to a normal source or with the input shorted. I suspect that the DN2540/10m45S combo is oscillating when given a chance. I have a 670 Ohm stopper resistor on the gate of the 10m45. Maybe a higher value, or should I look at something else? Maybe the high value for the gate resistor from the 2540 to the bias supply is an issue? Or could the protection zener be a problem? Suggestions welcomed.

Sheldon

[Fuling]

The zener should be connected "outside" the gate stopper resistor.

[Wavebourn]

Also, I would move output tube bias trimpot to a ground like it is done for your 1'st tube: noise in it when adjusting will cause peaks of current in the output tube.

[Sheldon]

Originally posted by Fuling
The zener should be connected "outside" the gate stopper resistor.

Thanks, that makes sense. I'll give it a try.

Originally posted by Wavebourn
Also, I would move output tube bias trimpot to a ground like it is done for your 1'st tube: noise in it when adjusting will cause peaks of current in the output tube.

Yes, that does seem the better way. Thanks.

Sheldon

[kevinkr]

One other thing that sometimes helps when all else fails is a small resistor right at the drain - 10 ohms ought to be enough. (De Q the surrounding wiring at vhf..)

[Sheldon]

Originally posted by kevinkr
One other thing that sometimes helps when all else fails is a small resistor right at the drain - 10 ohms ought to be enough. (De Q the surrounding wiring at vhf..)

Thanks Kevin, I assume you mean on the drain of the follower device, as the connection between follower and CCS is only a couple of cm. Though, I guess it couldn't really hurt there either.

Sheldon

[Sheldon]

I found the problem. Turns out that I had some metal debris that was causing some conduction from the drain of the 10m45 to ground. Cleaned that up and problem gone. I went ahead and made the circuit changes (drain stoppers, including gate stopper behind the safety zener, and moved the trim pot to the ground leg), just make the engineering more solid.

sheldon

[Sheldon]

Hmm, turns out the solder issue did not cure everything. These things seem to want to oscillate. I made changes as per the following and had problems when I tried to use a CCS on the driver tube. Thought I might have goosed something a little too hard, and cooked one of the IXYS devices. I had to replace one, so I just replaced both with DN2540, with the appropriate changes to get the correct current. All tests out fine on the bench. But when I power up the output tubes, things go south and the bias goes away. Here's the schematic with the IXYS shown. Any ideas, before I just go back to a resistor load on the source follower?

Sheldon

[gentlevoice]

Hi,

I previously encountered oscillation in many of my designs (very high bandwidth) but cured most all of them placing small ferrite beads (I guess they are normally used to filter out high frequency noise) on the transistor's pins.

In my opinion is also made the circuit sound better - somewhat quieter and less diffuse.

Regards,

Jesper

[Wavebourn]

Originally posted by Sheldon
Hmm, turns out the solder issue did not cure everything. These things seem to want to oscillate. I made changes as per the following and had problems when I tried to use a CCS on the driver tube. Thought I might have goosed something a little too hard, and cooked one of the IXYS devices. I had to replace one, so I just replaced both with DN2540, with the appropriate changes to get the correct current. All tests out fine on the bench. But when I power up the output tubes, things go south and the bias goes away. Here's the schematic with the IXYS shown. Any ideas, before I just go back to a resistor load on the source follower?

Sheldon

Ground loop may be?