Substitute 2SK134/2SJ49 with 2SK176/2SJ56

[dballes15]

I want to substitute 2SK134/2SJ49 with 2SK176/2SJ56 mainly because I have 48 of them. Its a common source output stage with 6+6 devices per channel. I noticed the Hafler DH-500 has a 680pf cap from gate to source on the first device of the 2SK176 output bank but not on the 2SJ56 side. Any recommendations?

[bear]

Essentially they are the same part, just selected for higher voltage...

You might want to say something about what you are going to do with them, like the circuit?

But you can substitute one set of Mosfets for the other set(lower voltage) type.

In most applications it is a good idea to static match them for Ids for a given Vg. Especially when the bias point is known to be a particular static Ids, match to that point. Curve tracing is better, but the static match usually will suffice in most instances.

_-_-bear

Edit: oh the small cap... that is usually added for stability and freedom from oscillation. Also the gate capacitance may not (and doesn't) match between the Pch and the Nch, so the small cap helps to make it symmetrical, if used that way... 680pf is a bit high, imho. Usually under 100pf is enough, but that will depend on the amp, and the layout and actual testing.

Edit II: a small cap PER device is better than one large one hanging on the end of the gate drive line...

[triode_al]

Hi dballes15,
Of course you have done the replacement, so please some comment on my thoughts:

- The substitution of 2SK134/2SJ49 with 2SK176/2SJ56 means installing a device with a larger difference in Ciss etc - so this would imply a slightly higher compensation than you had.
- The 2SK176/2SJ56 has a slightly different conductance, the threshold is lower so the unit will conduct a little bit more, but with a source resistor the difference is equalled, nevertheless the unit will get warmer .
- instead of 6 devices you can use 5, and leave the gate capacitance as-is in my opinion.

I am working on an F5 with 4 of these 2SK176/2SJ56 devices. The suggestion of a Ciss compensation in one 2SK176 is interesting. It will give both halves an equal bandwidth. It is something one can test in a bandwidth check.

Note: Nelson Pass plays the real capacitance significantly down from the theoretical capacitance.
So this remark of yours interests me.
alberti