jfet blues

[Brian Marshall]

Ok, so i designed a new pedal arround mpf102's.  I picked mpf102's because they sounded the best in the circuit.

It is a low gain overdrive, and ive sold a few, and people seem to like them so far. It uses 3 of them.  one as a buffer, and two as drain follwer gain stages.

Now i started out with a batch of 100 mpf102's, and have since used most of that batch up....  I started building a batch of 10 with my new batch of mpf102's.

the way the drain followers are set up..... Ill try to be brief as possible.

Gate biased to ground with a large resistor.
drain to V+ via 50k trimpot.
source to ground with 10k resistor, and a bypass cap.

Here's the problem (and i kind of thought this might be a problem eventually)

Too much transconductance on the new batch.  the original batch biased at 4.5v with about 22k to 25k resistance at the drain.  Some were a little higher actually.  

So i got my new batch of jfets, and they arent cooperating with me..... i have to turn the trimpots down to 12k or less to bias the drains at 4.5v.  this reduced the gain in the circuit a lot.

This is a big problem for me, and im pondering a few solutions...

First idea was to go through all the jfets and test them.... so i set up a 22k(drain), and 10k (source) resistor, and a 1meg resistor (gate to ground) on bread board, and went through all about 75 jfets all from the new batch of mpf102's..... and every single showed about 2.7 volts on my meter......   Ok so that idea sucked.... the whole batch is useless to me.

Then i thought of a new idea that might work a little better..... I was wondering if anyone had ever tried something like this.

use set reistors on the drain and source... say 25k and 10k, and bias the gate with a trimpot through a large resistor..... obiously the trim pot would need some other resistors as well to go along with it, but this seems to me like it would work, and be a much more reliable solution..... that way i could match transconductance, (at least i think it would)

Any comments woudl be much appreciated.

Brian

[Alpha579]

um, surely the Gate has no voltage on it, so it wouldnt affect the Jfet biasing...

[Brian Marshall]

um, surely the Gate has no voltage on it, so it wouldnt affect the Jfet biasing...

incorrect... the gate voltage effects the transconductance of the jfet
as the voltage at the gate goes up the transconductance of the jfet goes down

[Paul Perry (Frostwave)]

Maybe adjust the drain resistance? lower it & the gain will go down, I expect. Now you know why I hate using fets as anything but followers

[RDV]

Got any 2N5457? I think they'll work well as a sub for MPF102. You can get em cheap on eBay I think. http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=4666&item=3833479599&rd=1&ssPageName=WDVW About 22 cents a piece with shipping.
That guy is a good vendor.

RDV

[Jay Doyle]

That most definitely will work and in my opinion is the way to bias FETs.

Using that arrangement you can decide your gain by the ratio of resistors (within reason, you have to allow for proper current and also not exceed the total gain available from the FET, though those two normally go hand in hand) and then bias the gate to obtain your desired drain voltage. In your current case though it seems to me that you could drop the source resistor to 5k and get the same result; if you want to stick with your proven design.

Brian, the if gate voltage goes up in relation to the source than the transconductance goes down. Nitpicky but important.

Jay

[Brian Marshall]

Brian, the if gate voltage goes up in relation to the source than the transconductance goes down. Nitpicky but important.

Jay

that's actually what i meant.... i always get that confused, because it seems from the name the drain should be on bottom.... i was up late.... and kind of frustrated..

now that ive had a night to think about it though.... i dont understand why it is such a common practice to bias jfets on the drain.

[Jay Doyle]

[/quote]

that's actually what i meant.... i always get that confused, because it seems from the name the drain should be on bottom.... i was up late.... and kind of frustrated..

now that ive had a night to think about it though.... i dont understand why it is such a common practice to bias jfets on the drain.[/quote]

It is common because it is easy and requires less parts. It is a lot wasier to put a 1meg on the gate to ground and be done with it. I like to set the gain/headroom with the ratio of the source and drain resistors and then use an adjustable bias on the gate to set the drain voltage where I want it.

Works for me.

Here is a great app note of FET biasing that covers some of the advantages of what we are talking about:

http://www.vishay.com/docs/70595/70595.pdf

Take care!

Jay Doyle

[Paul Marossy]

I think the other part of the problem is that some JFETs are highly variable. Your first batch is obviously a lot different than the first.
Hey, did you read the JFET matching page at GEO? That's an interesting read...

[brett]

Hi.  Has anybody else noticed that the jfets with greater (more negative) Vgs also tend to have more variable Vgs and transconductance?  I suspect that it's a function of the manufacturing process.  

Some circuits that I couldn't get going with MPF102s (like the mu-doubler) because of biasing and transconductance issues work fine with J201s, which are at the other extreme of Vgs and are very high transconductance. Something like 2N5484s or 2N5485s might be a good compromises in your design.

cheers[/quote]

[R.G.]

Ok, so i designed a new pedal arround mpf102's. I picked mpf102's because they sounded the best in the circuit.

It is a low gain overdrive, and ive sold a few, and people seem to like them so far. It uses 3 of them. one as a buffer, and two as drain follwer gain stages.

Now i started out with a batch of 100 mpf102's, and have since used most of that batch up.... I started building a batch of 10 with my new batch of mpf102's.

Welcome to the real world of commercial electronics manufacturing.

You have just (re)discovered two things: (1) why consistency beats high performance in the commercial world and (2) why JFETs are not used much.

Your saving grace is that they do tend to run in batches. You can buy large batches, test out the really far out ones, and then modify a part or two to get correct operation with that batch. Fortunately, you're making few enough boxes that this is even an option. If you were having 5000 of these made at a time halfway around the world and no way to test them for sound until you get them here in shipping crates, you'd have a real (economic) problem on your hands.

No commercial venture would ever choose magic operation over consistency. That situation is economically unfit for survival, even if it was possible.

[Paul Marossy]

BTW, that guy on ebay that RDV linked to usually has some really good deals. I have bought several things from him now.

[Jered]

I can back up what RDV and Paul mentioned. Dan (the ebay guy) is one of the nicest, truly honest, and all around great guy. I live about a mile from him so I go to his house for my parts and over the last year he has become a true friend.Just a really down to earth guy with really great prices.
Jered

[David]

I can back up what RDV and Paul mentioned. Dan (the ebay guy) is one of the nicest, truly honest, and all around great guy. I live about a mile from him so I go to his house for my parts and over the last year he has become a true friend.Just a really down to earth guy with really great prices.
Jered

Whew!  That's a relief!  I bought a bunch of JFETs and MOSFETs from him yesterday.  ROG simulators, here I come!!   :twisted:

[Alpha579]

ah, so you mean the voltage on the gate in relation to the source?