Matching jfets

[stringsthings]

Interesting thread.  I'll put my money on R.G.'s circuits since I've never had anything but good results
when it comes to his information.  I'm sure there's some validity all around, but in the long run, we're
talking about guitar pedal circuits.   Not rocket science, clean room stuff.

[R.G.]

yes, I've said it enough times now // and this will be one of the last

I sincerely doubt it. Unless your "correctness" contains the caveat that the hundredth from the end of a chain of a thousand is "one of the last".  Actually, I suppose you could talk

You'll be back with moaning about "correctness" in a few months or a year, whenever the topic bubbles up again. You've been doing that for years, maybe even an entire decade if I checked my files for how far back your axe grinding goes.

I do hope you give it up. But forgive me if I don't hold my breath. 

[LightSoundGeometry]

we had to to some basic Transconductance stuff in lab ..all I remember is nothing lol ..you can literally spend weeks in physics of it or try the quick and dirty

I just got a dod 201 in the mail today and ordered a bunch of fets to build me a p45 or a p90 that need tested so this thread caught my attention.

kicking myself right now..I wish I had my analog devices and theory textbook back ..its the only one I rented and had to turn back in ..I have two digital books here, not helping!  I did find a few free pdf books but they go nowhere in depth with the step by step math formulas like the one we had for class..I think is was published by Pearson.

the dod 201 kicks ass..its probably the best pedal on the market dollar for dollar

Yt videos are good but are nothing like the real book in front of you ..and guess what idiot threw away all his lecture notes, labs and test? oh yeah ..I thought I could find a lot of it on google/comp later on and didnt need hard copies..organization if materials is the second thing you learn in physics right after the scientific method. 

thanks rob, I figured there was a dl pdf someplace of the good stuff!

[Rob Strand]

Yt videos are good but are nothing like the real book in front of you

See this book, section 6, PDF page 215
Especially AN129, 6-39, PDF page 252
https://archive.org/details/NationalSemiconductorFetDatabook1977

Very readable articles, not too much math and doesn't break out into semi-conductor physics speak every second paragraph.

BTW, it is downloadable.

[Eb7+9]

two things:

(i) jFET matching (per se)
(ii) selecting jFET's for PHASOR applications

and let's add a third,

(iii) selecting jFET's for wide-range PHASOR appliactions

we know that proper matching of jFET's requires at least a pair of data points
this is because the two basic quanitites the math equations are based on have independent behavior

(this is all over the internet if you google ...)

indeed, there's a statistical trend that occurs between Vp and Idss values (ie., proportionality) ...
but within that there's lateral variance ... hence the need for two measurements if deterministic matching is to be claimed

in this 2005 thread Sebastian (STM) provides data to corroborate this
and also provides some relevant explanation as it pertains to this discussion ...

http://www.diystompboxes.com/smfforum/index.php?topic=29513.msg200858#msg200858

it's important to note that for phasor applications proper matching (per se) of devices is not required
we know what matters here most is control over Vp

because the closer we can get to it to higher the range of available resistance can be obtained from the jFET
and hence, the deeper an effect can be produced (see STM's comments in the thread)

that's on the Vp side of things ...

otoh, any slight and non-proportional variation in Idss leads to an altered Ron value ... defined by Vgs(off)/Idss
and this has a similar resulting effect on the response as does the variance present in the capacitors used in the same circuit

we know that as long as devices are chosen with close Vp values
their Idss values will lie nearby as well ... and so Ron variations can be assumed to be reasonably limited anyway

so, in reference to the whole "matching" thing and Keen's circuit

(a) it doesn't offer matching by nature
(b) it can be used to at least show that a device isn't dead, and that Vp might lie in a restricted but still somewhat broad circle /// so usable in a low-range phasor such as the P45/P90 w 20k/22k range limiters ... but, not any different than biasing a jFET using a source resistor and measuring Vgs ...
(c) main point, it cannot be used to select devices for high range phasor apps as that application requires much better knowledge of Vp

my objection is not mere pedantry, but more design based

firstly, generalized Phasor designs do not necessarily require matched devices, (w interpretation)
second, the method presented there does not offer matching (per se) or even accurate Vp matching ...

... notice that the whole "jFET phasor" conversation is underlined by the fact that people are trying to copy what's already been done, pretty much everywhere ... there are, in fact, a couple of design assumptions that can be challenged here // esp. as it pertains to device selection ... I'm not putting down MXR's original efforts by saying this, just that with a little prudence we can bypass that whole issue altogether, and depart from it to obtain extended performance using a less restrictive set of conditions ... the solution is not that complicated

[R.G.]

Well, J.C., you made it three days on that "last time".  How long til the next "last time"?

[Eb7+9]

Pretty much every time this comes up, you pop in with a rant about it not being exactly right, and look for points about why it's not "RIGHT". Every since I called you out for publishing my artwork without attribution or permission in your "Rainbow of Sounds" book, you've been griped off.

just noticed this ...

pangs of guilt making you bring up my book finally ?! ... c'mon spill the rest of the story please, you're halfway there now ...

indeed, how about you tell the kind folk in this forum what happened after I sent you a complimentary copy of my book ... would it explain your strong emotional reactions and attacks towards my person, ... ? I'd like to be discussing electronics, but instead you keep dealing with adversity using insults, ... notice that's all you got in this thread so far (and others like it) ...

to me, at least, it's pretty obvious why ...

no Keen, I don't say your method isn't exactly right ... I just say that it has nothing to do with science, it's just bunk ...

that's all I'm saying ... no pedantry here bro

[R.G.]

OK, that makes one day since the last "last time". Your "last times" are getting more frequent, not ending. 

Yes, let's discuss your "book". Your book was a collection of effect schematics you collected from the internet, printed up and bound with a plastic comb binder. It included a substantial number of effect schematics that I researched and drew up, then posted for free on the net. Much of it was taken from the first effects schematic archive on the net, in which I was a collaborator. The schematics were unattributed, effectively presented as your work.

I still have that copy on my book shelf, waiting a moment just like this, when you'd bring it up. I have no guilt whatsoever about mentioning that set of events. I did not feel any gratitude at receiving a copy of my own work presented as someone else's. I can, in fact, prove my contentions about what you did.

Let's delve into it a bit. There are 57 effects listed in the book. Classifying who did what is a little foggy as you scribbled notes on a lot of them, so you'll likely claim "authorship" wherever you marked a page, but we can make a rough attempt. Of those, I personally drew up the schematics from scratch for 17. Jaime Heileman drew up about 17. Jaime and I together accounted for three. You drew up 7 from scratch, and modified or annotated my earlier work on three. You copied the manufacturer's schematics directly, sometimes with scribbled notes for another 15. You even did some original work.

Your book had a effect all right - it's one of the major reasons that I do not sell my PCB book - which is completely original, as I wrote every word and drew every bit of artwork myself - in soft copy. By doing that, your book of purloined schematics helped make it clear to me that intellectual theft would be the way the internet worked.

So it made it harder for everyone who is interested in that or anything I publish. I still regularly have to explain to people why they can't get soft copy.

I'm still a little amazed that you still feel the need to discredit a trivial circuit that is easy to put together, and has the cardinal advantage that it works. It's what the Brits call "cheap and cheerful", I believe.  You're off into explaining how it can't be right. And as I keep noting every time you pop back up with the same axe to grind- it's not perfect, but it works. You can explain how it's no good for as long as you like - and you've posting this same complaint since shortly after I posted that circuit - but you can't make it not work by explaining that it can't, much like the explanation of why bumblebees can't fly fails to explain why they do. Perhaps you're just mad at me for not agreeing with you, so you attack the poor circuit. 

But whatever your reasons, it's pretty clear that you will continue to net-bomb me with attacks about that, no matter what you say about being done with it. The real facts are that you are personally unable to stop doing this, no matter how often you say you will.

I'm not the one carrying the mental baggage here. I just don't like your intermittent attacks, and quit putting up with them.

Bro.   

[Mark Hammer]

This drives me nuts.  I've known R.G. since the alt.guitar days back in 1991, and I've had JC as a welcome guest in my home several times.  I love and respect them both, and it bugs the dickens out of me when my friends don't get along with each other.

But...be that as it may.  Most, though certainly not all, DIY builders of JFET-based phase shifters, are going to attempt to build some version of an MXR Phase 90 or Phase 45.  One of the essential facts a person needs to comprehend about both of those units is that they were designed to have a single user control, and reach enough of a compromise with all other possible control parameters, that the need for additional controls is sidestepped.  They aimed for a sweep width that was reasonably appropriate for both fast and slow speeds, and the P90 had a moderate amount of resonance that also worked for fast and slow sweeps.  Normally, users who have access to 3 controls, will set width narrow and resonance low for fast sweeps, and turn width and resonance up for long slow ones.  While not JFET-based, the Small Stone's "Color" switch did that adjusting by simultaneously lowering resonance, sweep width, and speed range in one switch setting, and doing the opposite for the other switch position.

Ultimately, as long as they are suitable for the role, and not entirely different from each other (e.g., a quartet made up of a J201, MPF102, 2N3819, and 2N5457), the importance of matching JFETs for phase-shifters depends on how wide a sweep one is aiming for.  Why?  Because the musicality of a phase-shifter is very much a function of the "turnaround"; that is how it reverses direction from the extreme top or bottom end of the sweep.

I'm sure many of you watched the Olympics this past summer, and you'll likely agree that 10M after they shoot out of the starting blocks, and 15M before the ribbon, there isn't a whole heckuva lot of difference between Usain Bolt, and all the other runners.  Some come out of the blocks faster, and have extra oomph as they reach the finish, but in "the middle", differences between runners are hard to spot.  Similarly, if one is going to maintain a modest sweep width that works well for both slow and fast sweeps, and not really "push" the JFETs to the limits of their range, a lot of people are going to find that almost any 4 randomly-selected 2N5952s (or 2N5457s, 2N5484s, MPF102s, etc.) from the same batch stand a pretty good chance of yielding phasing that sounds decent to them. 

Steve Daniels sent me sixteen 2N5457s to test out because he was curious about whether they would be suitable for P90s.  They were measured, and "ordered", from Vp of 1.69 to Vp of 1.4, Idss of 3.79ma to 2.91ma, and transconductance of 2.56 to 2.35.  Now, I do not possess the technical know-how to assess whether this is tightly or loosely matched.  But I can confirm that quartets made up of JFET #15 and 15, in conjunction with JFET #1 and #2, did not sound any worse than consecutive quartets made of, say, #5-#8 or #11-#14, simply because the moderate sweep range of the P90 clone (whose sockets I was plugging them into), did not make use of the limits of their sweep range.

Now, if one attempts to build a phaser that does have capability of achieving very wide sweep, with high resonance, I would imagine that selection and matching becomes more critical, since one does not want one or more of those JFETs to suddenly get stubborn about continuing to sweep for the top or bottom 20% of the sweep range.

One of the differences between issues of the P90 is sweep width.  You will see a 3M9 fixed resistor on the LFO output for some issues, and 3M3 on others.  3M3 yields a wider sweep.  You will also see 24k fixed resistors in parallel with the JFETs or 22k.  24k yields wider sweep than 22k (though I imagine not as much as the change from 3M9 to 3M3),

My hunch is that if one is attempting to clone/mod one that aims for wider sweep and higher resonance (i.e., better suited for slow sweeps), you will want JFET matching.  And if you aim for the narrower sweep, lower-resonance, format, matching won't be as big an issue.

HOW to match, when matching is needed, is something I have no expertise on.  I'm just saying that there are aspects intrinsic to the P90 and P45 that reduced MXR's need to engage in matching.  I'm sure they felt it was bad enough that they needed to fine tune the units with a trimmer.

[LightSoundGeometry]

This drives me nuts.  I've known R.G. since the alt.guitar days back in 1991, and I've had JC as a welcome guest in my home several times.  I love and respect them both, and it bugs the dickens out of me when my friends don't get along with each other.

But...be that as it may.  Most, though certainly not all, DIY builders of JFET-based phase shifters, are going to attempt to build some version of an MXR Phase 90 or Phase 45.  One of the essential facts a person needs to comprehend about both of those units is that they were designed to have a single user control, and reach enough of a compromise with all other possible control parameters, that the need for additional controls is sidestepped.  They aimed for a sweep width that was reasonably appropriate for both fast and slow speeds, and the P90 had a moderate amount of resonance that also worked for fast and slow sweeps.  Normally, users who have access to 3 controls, will set width narrow and resonance low for fast sweeps, and turn width and resonance up for long slow ones.  While not JFET-based, the Small Stone's "Color" switch did that adjusting by simultaneously lowering resonance, sweep width, and speed range in one switch setting, and doing the opposite for the other switch position.

Ultimately, as long as they are suitable for the role, and not entirely different from each other (e.g., a quartet made up of a J201, MPF102, 2N3819, and 2N5457), the importance of matching JFETs for phase-shifters depends on how wide a sweep one is aiming for.  Why?  Because the musicality of a phase-shifter is very much a function of the "turnaround"; that is how it reverses direction from the extreme top or bottom end of the sweep.

I'm sure many of you watched the Olympics this past summer, and you'll likely agree that 10M after they shoot out of the starting blocks, and 15M before the ribbon, there isn't a whole heckuva lot of difference between Usain Bolt, and all the other runners.  Some come out of the blocks faster, and have extra oomph as they reach the finish, but in "the middle", differences between runners are hard to spot.  Similarly, if one is going to maintain a modest sweep width that works well for both slow and fast sweeps, and not really "push" the JFETs to the limits of their range, a lot of people are going to find that almost any 4 randomly-selected 2N5952s (or 2N5457s, 2N5484s, MPF102s, etc.) from the same batch stand a pretty good chance of yielding phasing that sounds decent to them. 

Steve Daniels sent me sixteen 2N5457s to test out because he was curious about whether they would be suitable for P90s.  They were measured, and "ordered", from Vp of 1.69 to Vp of 1.4, Idss of 3.79ma to 2.91ma, and transconductance of 2.56 to 2.35.  Now, I do not possess the technical know-how to assess whether this is tightly or loosely matched.  But I can confirm that quartets made up of JFET #15 and 15, in conjunction with JFET #1 and #2, did not sound any worse than consecutive quartets made of, say, #5-#8 or #11-#14, simply because the moderate sweep range of the P90 clone (whose sockets I was plugging them into), did not make use of the limits of their sweep range.

Now, if one attempts to build a phaser that does have capability of achieving very wide sweep, with high resonance, I would imagine that selection and matching becomes more critical, since one does not want one or more of those JFETs to suddenly get stubborn about continuing to sweep for the top or bottom 20% of the sweep range.

One of the differences between issues of the P90 is sweep width.  You will see a 3M9 fixed resistor on the LFO output for some issues, and 3M3 on others.  3M3 yields a wider sweep.  You will also see 24k fixed resistors in parallel with the JFETs or 22k.  24k yields wider sweep than 22k (though I imagine not as much as the change from 3M9 to 3M3),

My hunch is that if one is attempting to clone/mod one that aims for wider sweep and higher resonance (i.e., better suited for slow sweeps), you will want JFET matching.  And if you aim for the narrower sweep, lower-resonance, format, matching won't be as big an issue.

HOW to match, when matching is needed, is something I have no expertise on.  I'm just saying that there are aspects intrinsic to the P90 and P45 that reduced MXR's need to engage in matching.  I'm sure they felt it was bad enough that they needed to fine tune the units with a trimmer.

so a website selling matched fets for a 90 either have no clue how a phase circuit works or selling snake oil 

I havent built mine yet but wont lie, was probably going to socket and sit there and swap fets in and out until I got something I like lol. I have 20-30 of them I got from SB

[Mark Hammer]

I don't think its snake oil at all.  If you want to make a stock P90, it may be a bit more than is truly necessary.  On the other hand, if you plan on maybe extending the sweep width a bit, it's probably a good idea to match.

I'm saying this from the perspective of someone who has bought a couple dozen, and clustered them into 4 or 5 workable sets, using simple Vgs matching, without too much trouble.  If somebody saves you the trouble of having to buy more than you really need, and does the matching for you for a few extra bucks, where's the harm?

My own experience is that current-limiting resistors on the LFO output (R18 below) can provide various sweep widths from around 2M2 or 2M4 up to 5M1.  One can fisah around for a pot with sufficient resistance, but the simpler thing is just to make R18 5M1 or 4M7, and use a 3-position toggle to bring in parallel resistors to get other sweep widths.  Fo instance, using a default value of 4M7, switching a 22M or 10M in parallel gets you 3M2 and 3M9, respectively, which are discernibly different sweep widths.

Combining this with variable offset, via R17, yields a much broader palette of tones.  Replace the 1M resistor with a 680k fixed resistor and 500k pot, and you'll be able to move the sweep range upwards and downwards, from gurgly to swirly.

Finally, R15 provides a fixed amount of feedback/esonance.  Replace it with a 12k resistor and 68nf cap in series, to limit bass at high resonance, and a 25k-50k pot to dial back resonance from max  Those three changes will get you a lot of tonal variation.  I added the variable offset to a Behringer P90 clone (couldn't add more because the SMD construction made adding the other mods difficults), and it ups the Behringer's game.

[LightSoundGeometry]

I don't think its snake oil at all.  If you want to make a stock P90, it may be a bit more than is truly necessary.  On the other hand, if you plan on maybe extending the sweep width a bit, it's probably a good idea to match.

I'm saying this from the perspective of someone who has bought a couple dozen, and clustered them into 4 or 5 workable sets, using simple Vgs matching, without too much trouble.  If somebody saves you the trouble of having to buy more than you really need, and does the matching for you for a few extra bucks, where's the harm?

My own experience is that current-limiting resistors on the LFO output (R18 below) can provide various sweep widths from around 2M2 or 2M4 up to 5M1.  One can fisah around for a pot with sufficient resistance, but the simpler thing is just to make R18 5M1 or 4M7, and use a 3-position toggle to bring in parallel resistors to get other sweep widths.  Fo instance, using a default value of 4M7, switching a 22M or 10M in parallel gets you 3M2 and 3M9, respectively, which are discernibly different sweep widths.

Combining this with variable offset, via R17, yields a much broader palette of tones.  Replace the 1M resistor with a 680k fixed resistor and 500k pot, and you'll be able to move the sweep range upwards and downwards, from gurgly to swirly.

Finally, R15 provides a fixed amount of feedback/esonance.  Replace it with a 12k resistor and 68nf cap in series, to limit bass at high resonance, and a 25k-50k pot to dial back resonance from max  Those three changes will get you a lot of tonal variation.  I added the variable offset to a Behringer P90 clone (couldn't add more because the SMD construction made adding the other mods difficults), and it ups the Behringer's game.

hey thank you mark for the info. The very first, random, fets I grabbed work perfect. Im going to run a switch like you said on the bb and experiment with the sweep although its fine like it is! im not going to much into phasers, I just wanted something to play gilmour and hendrix. Although i did order some lamps and ldrs goings to give a univibe circuit a shot next. still cant get the rog thunderbird working right either.

[LightSoundGeometry]

well I must of gotten lucky with the 45 because I cant get the 90 to work..not sure whats going on as I am still new to the phase, but the drains have 4.1 volts on them so I am thinking the Zener is not working right ..got a new bag of 5.1's in mail today.

and I thought you can measure R across in circuit as long as power is off ..im reading 180 ohms across the 22k source resistors ..I pulled one out and it measures 22k ..thought i had mixed up a bag again ..

hope I get this sucker working ..I sold my univibe to force myself to build me a good one and I am dying without it ,..the 2 stager only takes a vibe head like me so far in playing haha

[PRR]

> im reading 180 ohms across the 22k source resistors

With power off, no bias, the JFETs will go "full on" and act about like a 100r-1K resistor. 180r may be real typical.

If it was set in concrete, you could wire -9V to the JFET, force it "off" (>100K) and check again. But lifting one leg (either part, though lifting one end of resistor makes it clear what you are poking) is often easy enough.

[Mark Hammer]

well I must of gotten lucky with the 45 because I cant get the 90 to work..not sure whats going on as I am still new to the phase, but the drains have 4.1 volts on them so I am thinking the Zener is not working right ..got a new bag of 5.1's in mail today.

and I thought you can measure R across in circuit as long as power is off ..im reading 180 ohms across the 22k source resistors ..I pulled one out and it measures 22k ..thought i had mixed up a bag again ..

hope I get this sucker working ..I sold my univibe to force myself to build me a good one and I am dying without it ,..the 2 stager only takes a vibe head like me so far in playing haha

The zener is actually not all that critical.  It's really just there to provide a stable bias voltage to the JFETs.  Keep in mind that the bias is just a divided-down version of the supply voltage.  If it was 1978 and you were running the thing on a 9v battery, as the battery aged, you'd need to keep opening the pedal to re-adjust the bias.  The zener turns what might start out as 9.6VDC and gradually declines to, say 7.8VDC, to something that starts out as 5V1dc and stays 5V1dc for as long as there's just enough juice to power the pedal.  The benefit of the zener is that when the bias is set at the factory, that bias stays valid for the life of the pedal, and whatever power source you use.  But if your power source is stable (i.e., not a battery), and the trimpot has enough resolution for you to be able to dial in the appropriate bias, the zener is superfluous.

[LightSoundGeometry]

well I must of gotten lucky with the 45 because I cant get the 90 to work..not sure whats going on as I am still new to the phase, but the drains have 4.1 volts on them so I am thinking the Zener is not working right ..got a new bag of 5.1's in mail today.

and I thought you can measure R across in circuit as long as power is off ..im reading 180 ohms across the 22k source resistors ..I pulled one out and it measures 22k ..thought i had mixed up a bag again ..

hope I get this sucker working ..I sold my univibe to force myself to build me a good one and I am dying without it ,..the 2 stager only takes a vibe head like me so far in playing haha

The zener is actually not all that critical.  It's really just there to provide a stable bias voltage to the JFETs.  Keep in mind that the bias is just a divided-down version of the supply voltage.  If it was 1978 and you were running the thing on a 9v battery, as the battery aged, you'd need to keep opening the pedal to re-adjust the bias.  The zener turns what might start out as 9.6VDC and gradually declines to, say 7.8VDC, to something that starts out as 5V1dc and stays 5V1dc for as long as there's just enough juice to power the pedal.  The benefit of the zener is that when the bias is set at the factory, that bias stays valid for the life of the pedal, and whatever power source you use.  But if your power source is stable (i.e., not a battery), and the trimpot has enough resolution for you to be able to dial in the appropriate bias, the zener is superfluous.

now that makes perfect sense

i got my lamps and ldrs in the mail , might just skip over the 90 and try a vibe, the layouts/parts count are about the same size ..i was going to buy a madbean board but he is oos until 2017 looks like

[strassercaster]

do the savvass layout for the harbinger one. I have done a couple it is killer . i love phase and vibes.

Are there any good jfets left? I have ordered lots of 20 from 6 different sources. None test as high on the voltage as some i pulled from my mxr phase 90. they are very low like.48 when using the rg testing jig. I even bought 3 different matched sets. One from ebay and 2 from general guitar gadgets. none sound as good as my mxr evh or phase 90 script. I thought getting good germaniums was hard

[LightSoundGeometry]

do the savvass layout for the harbinger one. I have done a couple it is killer . i love phase and vibes.

Are there any good jfets left? I have ordered lots of 20 from 6 different sources. None test as high on the voltage as some i pulled from my mxr phase 90. they are very low like.48 when using the rg testing jig. I even bought 3 different matched sets. One from ebay and 2 from general guitar gadgets. none sound as good as my mxr evh or phase 90 script. I thought getting good germaniums was hard

going to try it ..got me a madbeans harby1, naughty fish and a dirtbaby PCB board coming in mail.. stil trying to gather all the darn parts ;;hell mang lol

[LightSoundGeometry]

do the savvass layout for the harbinger one. I have done a couple it is killer . i love phase and vibes.

Are there any good jfets left? I have ordered lots of 20 from 6 different sources. None test as high on the voltage as some i pulled from my mxr phase 90. they are very low like.48 when using the rg testing jig. I even bought 3 different matched sets. One from ebay and 2 from general guitar gadgets. none sound as good as my mxr evh or phase 90 script. I thought getting good germaniums was hard

well Billy, just got the madbeans and small bear stuff in mail today ..going to the do madbeans layout, alex's layout and the russian dudes layout..probably wont get to it until sunday. I have a steel wedge I m housing a tonebender in and I have to do that project first ..its a mFer drilling into steel but these wedges are freaking sweet ..get a hold of me if you want me to come over to saint louis this week and check out your lab! I sent you text too.

on the testing ..I found some stuff from Bob Pease who shows how to test just about any device with almost nothing , like 3 parts lol..the dude is awesome.. something pease would say " you dont need that 50,000 analyser  ..all you need is a 146768UE and two 21K resistors and we can look at blah blah blah " hahaha .."why 21? well is all i had at the time and all we need!"  hahah

one good side effect of building a TB is listening to Led Zep I & II and some old Beck for 3 straight days