Looking for a FET...can't find it anywhere!

[Flatpicker]

[quote author="edanderson"]going back to the original question, i'm guessing that one reason mr. dorsey, whose oktava mic mods were published in recording magazine, specified the 2sk170 because it was available pre-sorted for Idss...[/quote]Initially it was because the '170 is a little quieter. Then he had to re-bias to match the '170's IDSS.

In all fairness, I probably spoke out of line earlier when I said the original sounded better because I didn't take time to re-bias the '170. That's one of those things I was going to do, but have not gotten around to yet. :sad:

 

[edanderson]

yes, of course the 2sk170 is quieter, but think about it. you're writing an article where you want to be able to say "okay, swap these components for new components with these new values and it WILL work." with just about any other fet, because the spread of Idss values from part to part is so wide, you would have to pick and choose new biasing resistors OR select the fets. with the 2sk170BL, the Idss is known within a range, and the given bias resistor values should always work. this simplifies the fet and biasing resistor part of the mod quite a bit; plug and play, basically. you don't even have to own a meter.

so, i'd say starting with the 2sk170BL is a good starting point. ideal? perhaps not, but any other fet will take some fiddling. that doesn't mean it isn't worth it tho. especially if absolute lowest noise floor isn't your #1 criteria for success.

ed

 

[Flatpicker]

[quote author="edanderson"]...with the 2sk170BL, the Idss is known within a range, and the given bias resistor values should always work...[/quote]A very good point!

 

[Gus]

No the bias will still need to be adjusted for almost any fet used. For a better hint at why I have a "problem" with the 170 read the PDF at the neumann site "Microphones". Oh you might have to do some simple math.

My ears "tell" me the cap drain to gate "hurts" the sound (do the km84 test I posted before for a hint). I don't care what the noise specs are because it is a microphone circuit and seem to have litttle to do with the noise at the gate to capsule part.

Also look at how most pads are setup in microphones and think about gate to source cap.

The 170 specs are good for noise but it must be a bigger die in a 170 or masked different because of the added cap.

I think the fet "upgrade/replacement" might not be an improvement but might hurt the sound.

I don't think the new gefells or neumanns use a 170.


What do I know about microphones anyway?

 

[edanderson]

when i emailed scott about the Icss issue, this was his reply:

[quote author="scott dorsey"]Partly because of noise concerns, and partly leakage. You want the lowest noise AND the lowest input capacitance AND the lowest leakage. And you have to balance all three of these.

The Chinese guys use smoke alarm FETs with really low leakage, but high gate capacitance and high noise.

You can get better input capacitance than the 2SK170, but you get higher 1/f noise. This gives you less hiss and more rumble. It's all a tradeoff.[/quote]

so it sounds like, at least in most version of the oktava mics, the FETs being replaced are high capacitance to begin with.

[quote author="Gus"]What do I know about microphones anyway?[/quote]

Gus, i'm not trying to disagree with you at all. you have thought a lot about mic circuits and most importantly done tests to prove your ideas to yourself. i'm just trying to illustrate why, for a very beginner-oriented magazine article, certain parts decisions make sense. i wouldn't say that just because they were published, they are the definitive word on the best component choice possible. only that, in a certain context, they were the "easiest" choice to produce a predictable outcome.

ed

 

[Gus]

The Scott Dorsey statement

"The Chinese guys use smoke alarm FETs with really low leakage, but high gate capacitance and high noise. "

A 170 has high cap!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

This makes me wonder what he is talking/writing about.

What microphones is he talking about? Most/all of the china stuff I looked at have 2sk170s or j305s I think I might have saw a j105 in one.

As best I can tell there are not many FET dies made. Most numbers on fets are for the case style or pinout or IDSS or pinchoff or etc...............................because fets made from the same die are all over the place if what I have read is true about fets and from the bag of 1000 fets that I have that are all over the place.

If you look at the better spec sheets you will see the fets are sourced from a process number ----- So a 2n3918 is the same die as a 2n5484(5,6 IDSS sort) but one has a very wide spec range and one is more selected after the die run. I think the j305 is the same die a the 3918 I think I got the part numbers right.

Think like. CPUs are speced for speed, FETS are speced for IDSS etc.

The schoeps schematic on the web shows a fet like a NS process 50 fet

So that smoke alarm fet might be the same die as the orignal part speced for all we know.

I guess I am a little pissy because I just shared some information that I have not seen anyone else share or even KNOW about that was backed up with some testing. Some of the hints are very important for a good design.

A few people have tried microphones I have built and/or worked on that have 2n3819s and they liked the "sound" and did not note any noise.

The more I learn about pro audio the more it seems like the HiFi crap all over again. There are some very good designers but you often get what you pay for.

Sometimes I think there is a VERY small group that realy knows how to DESIGN a microphone or even understand some of what is going on.

 

[Flatpicker]

[quote author="Gus"]No the bias will still need to be adjusted for almost any fet used...[/quote]Absolutely, but I think Ed's just saying that for SD's purposes, (get paid for a DIY article) the '170 was the next logical step up the ladder from the "smoke alarm" FET.

Of course what you are saying is that this is absolutely no indication that the '170 is the best FET for the job, and that point is very well taken. (For those just tuning in, Gus knows a TON about condenser mic circuits!)

In fact, thinking of this I'm going to have to revisit some mic mods and try different FETs with a lower G to D cap. Crap - there goes the rest of the summer! :sad: :wink: