matching JFETS ?

[plexi12000]

might build a phaser...i know you need "matched" JFETS.  Curious...what exactly is being "matched"?

[nickbungus]

Great question, it was only a few months ago I was in the same boat.

After determining what JFETS I needed, I bought 100 of them, although you can probably get away with about 25-50.

I then built a Greatly Improved JFet Matcher II.

I then got a piece of paper and then drew 100 boxes and numbered them.  After that I measured each of my JFets using my matcher circuit, stuck them in a box on my paper and wrote down both the Vp reading and Idss reading in a spreadsheet.

I chose to match on Vp (I believe) because I'd read RG Keens article on JFet matching, and that's what he used. (I hope I got that right).  That article will probably answer your question better than my response!  I just thought I'd share my technique.

By using a spreadsheet, I was able to order them and find the closest matches.

[nickbungus]

Actually if J112's will do for you, I now have 94 spare!  If you PM me your address I'll send a batch of matched ones out!  Save you a job.

[MrBinns]

you bought 100!

I bought 4. I guess I got extremely lucky

[Elijah-Baley]

On my list project there is a Phase 45, and I need JFET matched, I have some kind.
I'l try several combinations and I'll choose the better to my hear.

You could try, too, and get lucky, maybe.

(With Phase 45 somebody gets a less subtle effect with J201s ).

Waiting your news.

[nickbungus]

you bought 100!

It was one of those things, when you bought one it was about £1 but when you got a 100 it was 9p each, so at £9 for a 100 I couldn't go wrong.

But I'm happy to give them away if anyone does need some I'll happily post.

[R.G.]

might build a phaser...i know you need "matched" JFETS.  Curious...what exactly is being "matched"?

Good question. "Matched" means that multiple devices have been selected to be alike in some way. That phrase "in some way" is not a casual toss-off to end the sentence. There are many characteristics of a semiconductor that can and are measured. Usually only one or a few characteristics are important to be similar. The more characteristics that have to be "matched", the more difficult it is to find matched devices.

It's worse for JFETs. JFETs are wildly variable, much more so than bipolar transistors.

In a phaser, you want all of the JFETs to be at about the same resistance from drain to source when their gates are at the same voltage compared to the source voltage. Matching JFETs for phasers refers to testing them for having very close drain-source resistance at a given voltage from gate to source. My JFET matching setup and the derivatives do specifically this. They do not match other characteristics. But Rds@Vgs is generally enough for a phaser.

[bool]

... or a "dynamics" circuit (comp/gate etc.) ...

[amptramp]

Just for added information, look up the Paul Nelson phaser which uses quad CMOS analog switches which are switched on and off above audio frequencies for varying duty cycles to achieve the phaser effect.  The advantage of this is there is no need for matching since the devices are either open-circuited or shorted with a negligible resistance of about 100 ohms.  It is a little more complicated, but you order only the parts you need and every one you build sounds like every other one you build.

[LightSoundGeometry]

might build a phaser...i know you need "matched" JFETS.  Curious...what exactly is being "matched"?

Good question. "Matched" means that multiple devices have been selected to be alike in some way. That phrase "in some way" is not a casual toss-off to end the sentence. There are many characteristics of a semiconductor that can and are measured. Usually only one or a few characteristics are important to be similar. The more characteristics that have to be "matched", the more difficult it is to find matched devices.

It's worse for JFETs. JFETs are wildly variable, much more so than bipolar transistors.

In a phaser, you want all of the JFETs to be at about the same resistance from drain to source when their gates are at the same voltage compared to the source voltage. Matching JFETs for phasers refers to testing them for having very close drain-source resistance at a given voltage from gate to source. My JFET matching setup and the derivatives do specifically this. They do not match other characteristics. But Rds@Vgs is generally enough for a phaser.

are you the real RG ?  I think its really cool that you post here   I own some visual sound products and very happy with them ..the one spot is a great gadget

[armdnrdy]

are you the real RG ? 

Do you think that a fine DIY site such as this would host a fake R.G.? 

In the flesh...well...virtual flesh!

And we are glad to have him!

[LightSoundGeometry]

are you the real RG ? 

Do you think that a fine DIY site such as this would host a fake R.G.? 

In the flesh...well...virtual flesh!

And we are glad to have him!

I read two of his articles this morning ! very good stuff for a newbie like me. The article are so in depth I have to book mark them and reread a few times lol

[armdnrdy]

I keep three folders containing R.G.s material on my PC.

The first is titled: Writings by R.G. that I understand.
The second is titled: Writings by R.G. that I do not understand.
and the third folder is titled: Writings by R.G. that I will never understand. 

[plexi12000]

I keep three folders containing R.G.s material on my PC.

The first is titled: Writings by R.G. that I understand.
The second is titled: Writings by R.G. that I do not understand.
and the third folder is titled: Writings by R.G. that I will never understand. 

HAHAHAHAHA!!!  -- aint that the truth, armdnrdy!!!   

[plexi12000]

Thank you guys for answering- i appreciate it!   Maybe its my beer talking but.....RG, I love you ,man!!!  hahahaha!!!!

So i guess, i'm not able to use my meter to match, huh?  (i can check HFE's on it though.)

oh- wait a sec....i guess i'll start with RGs article linked above---  thanks for that too!!

[tubegeek]

are you the real RG ?

There is no real R.G. He didn't exist, so we created him.

[italianguy63]

Yes.  RG is a spambot.  LOL   

Actually, a very helpful and very real dude.

A JFET tester will become a "tool of the trade"  you will probably need to build.  Along with an "Audio Probe" and "Signal Injector" eventually, if you don't already have them.

Here is my JFET matcher--



EDIT-- note the "free" harbor freight DMM!  It is a great use for one.

MC

[italianguy63]

And-- off subject.  But realated to my last post...

If you are nice to him, Henry Rollins will hold your probe for you:



If you are not nice.. he will pound you into oblivion.

Carry on...........

(EDIT NOTE:  The red probe was made to discharge caps in amps.. so they don't shock the crap out of you, the black one is the audio proble.)

MC

[Mark Hammer]

WHY do they need to be reasonably matched?

Because a phase-shifter relies on shifting the phase.  That is, the effect comes from notches being moved around.  What that implies is that the JFETs - or whatever kind of control element the design uses...because there are several kinds - have to always respond to  the LFO controlling them.  Any given JFET will have a range of applied gate-voltage/current, outside of which it will not change drain-source resistance.  This is why the P90, and any other JFET-based phase shifter always has a bias trimpot.  The bias moves the JFETs into the range where they can be made to keep "moving"/responding to the LFO.

When the JFETs are not matched, then finding a bias range where all of them are responding to the LFO, at all points in the sweep cycle, becomes much much harder.  The result is that, for example, all 4 JFETs are changing drain-source resistance for part of the sweep, but at a certain point one or more just stops changing for the remainder of that direction of the sweep, until the sweep again returns to that point where those JFETs are susceptible to change.

All the JFETs have to have their oars in the water for the phaser to move in the desired direction.  When they are not matched, some may have one or both oars "out" some of the time.  I will point out that their respective drain-source resistance does NOT have to be matched, only their ability to change in response to the LFO.

Because the risk of inability to respond occurs as one moves to the extremes of the sweep, mismatched JFETs run the risk of having less musical-sounding "turnarounds", whether at the top or the bottom of the sweep.  Naturally, this will be most likely to occur when using a wider sweep.  You will note that some posted schematics use a 3M3 resistor on the LFO output, while some will use a 3M9.  The larger value yields a more limited sweep, which is a little better suited to faster speeds.  I don't know if mismatchng can be entirely offset by adjusting speed width, but it might help a little, at the potential forfeiture of wider slow sweeps that you might like to have.

One of the reasons why some folks can have such an affection for photocell-based, PWM-based, or OTA-based, phase shifters is that matching is a non-issue.  The photocells will always respond - even if non-identically - to the light shining on them.  The CMOS switches (used for PWM) will all respond identically, and the OTAs (e.g., in Small Stone, Moog, or similar) will also respond similarly.  It is only in the case of JFETs that there can be a point where there is a risk of them performing non-identically.

[Rob Strand]

Curious...what exactly is being "matched"

What needs to be matched is the Vgs_off, the gate source voltage where the current is zero.
In reality this converts to matching the Vgs voltage when the drain current is low, say 10uA.
Most matching circuits match Vgs at low current.

When this is done, and assuming the all the JFETs are the same part, the JFET resistance will tend to match well enough over the whole range of Vgs voltages.   

If they don't match the phasing effect will sound weak.