Servo Offset Biasing for JFETs

coldcraft · Post by **coldcraft** » 15 Jan 2014, 15:48

Here's an idea I am working on, to use an opamp servo to set a FET drain bias to exactly 4.5V or whatever desired voltage by offset biasing the Gate (Voltage through R5).

Overkill? Absolutely. However, with two or more of these stages used in a design, the set point (Vr) can be varied to adjust the bias of multiple stages simultaneously. Has anyone seen anything like this before? I'm having a difficult time finding info on basic servo circuits.

RnFR · Post by **RnFR** » 16 Jan 2014, 02:53

Hey I'm back! Looks cool, maybe I'm missing something, but doesn't the differentiation between JFETS make adjusting more than one stage kind of pointless? It's been a while...

coldcraft · Post by **coldcraft** » 16 Jan 2014, 04:07

RnFR wrote:Hey I'm back! Looks cool, maybe I'm missing something, but doesn't the differentiation between JFETS make adjusting more than one stage kind of pointless? It's been a while...

each stage would have its own opamp providing the bias, so by adjusting R2/R3, each opamp would do what it had to do to set each JFET to the same drain voltage. at least that was my thought. told you it was overkill.

commathe · Post by **commathe** » 16 Jan 2014, 07:44

Not necessarily overkill. Could actually make a large production run JFET pedal more feasible and easy, especially since you don't have to do any biasing by hand. For a DIY project though, probably more hassle than just using a trim-pot.

MoonWatcher · Post by **MoonWatcher** » 16 Jan 2014, 12:58

commathe wrote:Not necessarily overkill. Could actually make a large production run JFET pedal more feasible and easy, especially since you don't have to do any biasing by hand. For a DIY project though, probably more hassle than just using a trim-pot.

I think that even for DIY it's not really extreme overkill. The nice thing about op amps is depending on the number of JFET stages you are running, you can resort to a single, dual, or quad op amp (I don't recall seeing many designs that use more than 4 JFET's especially).

I think that once you get used to designing layouts that would encompass the servo components that it might even prove to be less of a hassle than adjusting trimpots. And if you socket the transistors, it makes testing a plug and play thing as you don't need to rebias every time you swap a transistor. In essence, you might be able to come up with a better selected group of transistors in a fraction of the time.

So that is a benefit that a company doing a large production run wouldn't be able to exploit. I think it would also make more clear just how much of an audible difference there might be between any two JFET's.

And for testing purposes, you could build a jig with toggles to let you switch out two different JFET's in real time.

I think this has a lot more substance to the application than we may realize at first.

DrNomis · Post by **DrNomis** » 16 Jan 2014, 13:17

I think this Servo-Biasing idea has alot of merit.....

coldcraft · Post by **coldcraft** » 16 Jan 2014, 14:19

certainly if someone starts talking about price and component number, this is overkill. I don't think it will necessarily bias correctly for every device, but its meant to be another step up from "offset biasing", which is applying a positive voltage to the base, without using a trimmer on the drain/source. So if you want to guarantee 4.5V bias for symmetrical headroom, you can, but you can also pull it away from that by changing the reference voltage that the opamp is comparing the bias to.

DrNomis · Post by **DrNomis** » 16 Jan 2014, 15:20

It's actually a pretty cool concept, I don't think I've ever come across Servo-Biasing of Fets before, I've pretty much treated them as the solid-state equivalent of Valves when it comes to biasing, I wonder if Servo-Biasing has any effect on the mu, or voltage-gain of a Fet, I know that you can use a second Fet as an active-load on the first Fet's Drain and you do get an increase in voltage-gain......

lead2203 · Post by **lead2203** » 17 Jan 2014, 19:04

DrNomis wrote:It's actually a pretty cool concept, I don't think I've ever come across Servo-Biasing of Fets before, I've pretty much treated them as the solid-state equivalent of Valves when it comes to biasing, I wonder if Servo-Biasing has any effect on the mu, or voltage-gain of a Fet, I know that you can use a second Fet as an active-load on the first Fet's Drain and you do get an increase in voltage-gain......

Servo-Biasing is also used in tube stuff too. Check the Crate Blue Voodoo bias circuit for one example.

coldcraft · Post by **coldcraft** » 17 Jan 2014, 19:17

lead2203 wrote:
Servo-Biasing is also used in tube stuff too. Check the Crate Blue Voodoo bias circuit for one example.

Which Model? I've seen the BV120 service manual and it doesn't have anything out of the ordinary for the biasing.

aquataur · Post by **aquataur** » 17 Jan 2014, 21:38

coldcraft wrote:certainly if someone starts talking about price and component number, this is overkill.

Component price his is not an argument for the d.i.y.-er. Manufacturers may debate a possibly redundant capacitor and play hazard for the sake of saving a few cent in lieu of a big production run.

I think I have seen something similar from the few things that came from Russia with love

before they started to discover the buck. However, this is justified.
I don´t recall where, because these web pages are instable and frequently disappear, letting alone readabilty.

BTW, did you notice, that hardly anybody outside the countries, where English is the official language, cares to keep their pages in English? Hardly any Spanish, French, Russian pages, although there is a big community. Must have to to with national pride. Or is there little English taught?

have fun,

-helmut

mmolteratx · Post by **mmolteratx** » 19 Jan 2014, 01:51

I like it. I've thought about something similar before, but never got around to actually implementing it.

mmolteratx · Post by **mmolteratx** » 19 Jan 2014, 02:51

Just breadboarded this, and it works pretty damn well. Bang on 4.5V with a 2 dozen FETs I've rolled through. Could use a few improvements just to make my EE profs happy though. Will work on it some more when I've got more time. You could eliminate C6, downsize C5 to 100n (or even smaller, but physically, 100n is a good size), and of course use a different op amp and moderately different values for everything. My stock is pretty low right now, so I used a bunch of discarded parts. Might also be a good idea to bump C7 up to 10-22µ or bigger as well. I only had one 47µ cap and used it for the FET circuit instead.

coldcraft · Post by **coldcraft** » 19 Jan 2014, 03:27

Great! Thanks for taking the time to breadboard that. Did you run audio through it? Hows it sound?

mmolteratx · Post by **mmolteratx** » 19 Jan 2014, 05:05

coldcraft wrote:Great! Thanks for taking the time to breadboard that. Did you run audio through it? Hows it sound?

Just like a normal JFET stage really. The boost stage is based on a Baby Pink Booster and it sounds pretty much identical, save for a few adjustments I had to make to use it with 5457s instead of 5952s. Gonna lay out a board for a few cascaded stages and see how that works. I suck too hard at breadboarding and have had a few too many beers to attempt that tonight.

EDIT: There are some issues with the bias drifting though, and that's the issue I'd try to fix. It's not huge (varies by ~ +/-.5V), but it is noticeable. I've got a few ideas I'll try tomorrow after a run to get some parts.

coldcraft · Post by **coldcraft** » 19 Jan 2014, 15:37

mmolteratx wrote:EDIT: There are some issues with the bias drifting though, and that's the issue I'd try to fix. It's not huge (varies by ~ +/-.5V), but it is noticeable. I've got a few ideas I'll try tomorrow after a run to get some parts.

interesting. I think that could probably be fixed by adjusting the servo components, but I don't know enough about them yet.

MoonWatcher · Post by **MoonWatcher** » 19 Jan 2014, 18:20

coldcraft wrote:I think that could probably be fixed by adjusting the servo components, but I don't know enough about them yet.

I don't either, but would have to believe the same thing. Probably just a little finessing, and determining how much voltage variation would be negligible.

I wonder if anything could be borrowed from how Boss set up Q15 in the OD-3? How do they manage to mass produce that thing without the need to sort the FET's?

I don't mean for the OD-3 analysis to carry things away from servo biasing, especially now that we have confirmation that it really should work (well) to a certain degree.

coldcraft · Post by **coldcraft** » 19 Jan 2014, 18:49

the OD-3 uses whats called auto-biasing, where the Gate goes through a resistor to ground, and you cross your fingers that the JFET characteristics let it bias correctly. Sending the gate to a voltage other than ground is called offset biasing, and that's what the servo is doing.

The od-3 also has feedback clippers, and I would guess that they limit the signal before the bias has a chance to effect the clipping.

http://www.freeinfosociety.com/electron ... ossod3.gif

gus · Post by **gus** » 19 Jan 2014, 19:35

The first schematic has low current drive 9VDC and 100K drain and source resistors
The 2nd has a lot of parts
There are other ways to Design a circuit to drop in JFETS. Fender did it in a solid state Harvard preamp.
Look at the other place there are few threads.
Servo control of gain stages is old, power amps have used it. I have built a power mosfet overdrive with a textbook servo many years ago It is built in a coffee can.
Have you seen the Vishy app notes?
Also what is the time constant of the servo and how does it repond to a signal with DC offset

coldcraft · Post by **coldcraft** » 19 Jan 2014, 21:23

I tried it out with a MPF102 and a J201. I'm not sure its working the way I intended. If the drain is at 4.5v, then the opamp will output 0v (without negative feedback). So if the JFET is capable of biasing up with the gate at 0v,then it will work. the J201 I had wouldn't work.

In reality, we want the opamp to output whatever it needs in order to make the drain 4.5v, which is not the same thing. for that I need to so some more thinking. I am not 100% sure it can be done with 1 opamp.