Aleph output topology question

[mpmarino]

Hi folks and Mr. Pass,

I have a question regarding the aleph output section and the source/ccs source resistors in particular.

On a normal Aleph circuit, the source resistors are individually connected between each fet source and another common point. (the common point obviously depends on whether we are discussing the ccs or O.P.)

My question is this:
Would it not be advantageous to tie all sources together, not O.P. and ccs together, but all O.P. sources together.. and all ccs sources together. In the case of the O.P. you could then build a parrallel array of the correct value back to -v using a value/quality you may have on hand. You could up the bias a bit just by adding another one/two... in parallel. Right? or wrong

Also, it seems to me that if vgs should be matched, the source resistors should be either a high tolerance or be matched. Wouldn't my proposal negate the need for any decent level of tolerance and allow for better sharing, leaving vgs the only factor for concern?

Am I totally wacked out??

Thanks,
Marc

[BobEllis]

Individual source resistors are used to force current sharing. If you tie all of the output or CCS sources together (which I think is your proposal) you'll end up with the overwhelming majority of the current in one device.

Vertical MOSFETS have a negative temperature coefficient - Vgs at a given current drops with increasing temperature, or current increases with a fixed Vgs with increasing temperatrue. As it draws more current, it heats up more, until it "hogs" almost all of the current.

With a source resistor on each device, as it heats up and tries to draw more current the increasing voltage drop across the source resistor reduces Vgs, limiting the runaway. Of course, if the source resistoris too small, it won't be able to prevent runaway.

So, unless you like blowing up FETs, keep the sources separate.

[Nelson Pass]

True. With matching and good heat sinking you can keep this
effect down to a dull roar, but Source and Gate resistors are
almost always in good taste.

In Class A with a good sized bias, the temperature coefficient
gets less, and as the current increases it swings the other
way. This point occurs at a value somewhat higher than where
Class A amplifiers are usually biased, but you can use much
smaller Source resistance as the current goes up, not only
from less temperature coefficient, but also because the resistive
voltage drop of the resistor becomes a larger part of the bias
voltage and is linear.

[mpmarino]

Thank you gentlemen,

I was making the mistake that matched fets will act exactly the same way. I now see how it works. Even a slight mismatch would cause a huge imbalance.

That could have gotten expensive....

Hmm, ya learn a couple thousand new things everyday!