Matching LTP FETs for Best Sound?

[Dave S]

I have read Nelson's paper about selecting the best FETs his Aleph power amps. Vgs selection is straightforward, however it is not at all clear (to me at least) what else we are supposed to measure for.

To quote Mr Pass:
"Once you have figured out what combination (or sets of combinations) of characteristis you want...."

I have searched this forum in detail and no one other than the "One and Only" seems to have figured out what combination (or sets of combinations) of characteristis we want.

Except for Vgs I can only figure that the input capacitance should be matched (to reduce hf distortion in the LTP?). Problem is I have tried to measure some difference in input capacitance (by driving the FET with a sq wave of known source impedance) and all my IRF9610s seem identical in this respect.

Someone, please enlighten me, before I just stick in a pair with similar Vgs.

 

[Alain Dupont]

Dave S,

>>Except for Vgs I can only figure that the input capacitance should be matched >>(to reduce hf distortion in the LTP?). Problem is I have tried to measure some >>difference in input capacitance (by driving the FET with a sq wave of known >>source impedance) and all my IRF9610s seem identical in this respect.

Its a good point, but as soon as they are from the same lot code there should be
very few variations on the input capacitance.

You have to mesure the Mosfet for the exact bias current they will be working with.

For my A75, I have checked them for Vgs only ; the one thing, is to have always the same current through the MosFet during the matching process, I had installed
a analog current metter on the drain and set the power supply to 15.00 volts
all I had to do was plug the MosFet wait 3 seconds, and read the Vgs.{The temperature of the room was set to 23 degrees one day before the matching}

I have done the mesures twice the same day and the errors between the two measures were less than 0.01 volt.

That worked fine for me, I have done that for the IRF110/9110, IRF 610/IRF9610
and IRF 232/9231 {with respective current for each type}

Alain.

 

[Nelson Pass]

The only thing that has worked reliably for me is to plug
the different parts into the actual circuit and see what you
get. Actual matching is not always what you want.

Of course its takes a lot of time.

This is something that virtually no manufacturer can
deal with, but the DIYer with a small budget for extra parts
can get results that the big boys can't.

 

[Dave S]

Thank You

Thanks Alain - Vgs matching is no problem, from one tube (25 pcs) of IRFP244s I easily got 4 sets of 3 matched to <10mV.
I have a tube of IRF9610s (50 pcs, all the same date code) so Vgs matching of 2 pairs will be no problem here.

Thanks Nelson for your reply (and of course copious thanks for letting us use your designs for DIY), I have also re-read the last para on page 10 of your "Practical Mosfet Testing for Audio" and it clearly states that swapping LTP Fets at random can give much less noise and THD (for the whole amp) - so that answers the question as to what I'm looking for. It also confirms that I need to get the amp up and running and then worry about selection of the LTP Fets.

In case anyone is interested, I'm building an Aleph 30-5 (front end is Aleph 30, output stage is Aleph 5) using my own pcb design. I will probably build one of the BLS variants as well, to compare to my scratch built Grounded Grid. I would expect that the BLS has the same issues with Fet matching.

Dave