Author Topic: Most imortant characteristic to match for equivalent JFETs?  (Read 915 times)

Lizardmn

If I am trying to find an alternative model JFET to use in a discrete op-amp config like in a blues driver, which are the most important numbers on the data sheet that I should be looking to match? Vgs, Idss, etc? The package doesn't need to be the same btw.
Or, if I can't find a good match, is it possible to tweak some resistor values to compensate and get the same result fairly easily?
Thanks

antonis

« Last Edit: July 28, 2022, 07:42:32 AM by antonis »
"I'm getting older while being taught all the time" Solon the Athenian..
"I don't mind  being taught all the time but I do mind a lot getting old" Antonis the Thessalonian..

Lizardmn

Re: Most imortant characteristic to match for equivalent JFETs?
« Reply #2 on: July 28, 2022, 07:57:51 AM »
For differential pairs I think. I'm looking to find an alternative for 3x 2SK184's, though one of those is for the unity gain input buffer where, to my understanding at least, the FET choice isn't nearly as crucial as in the gain stage.
These resources look helpful, thanks.

antonis

Re: Most imortant characteristic to match for equivalent JFETs?
« Reply #3 on: July 28, 2022, 09:28:52 AM »
Don't know about your particular circuit but this: https://www.analogisnotdead.com/article25/circuit-analysis-the-boss-bd2 might help you with regard to items configuration/values..

P.S.
Forgot to mention R.G.'s article about JFETs matching.. :icon_redface:
http://geofex.com/Article_Folders/fetmatch/fetmatch.htm
« Last Edit: July 28, 2022, 09:41:33 AM by antonis »
"I'm getting older while being taught all the time" Solon the Athenian..
"I don't mind  being taught all the time but I do mind a lot getting old" Antonis the Thessalonian..

Vivek

  • Awesome!
  • ****
  • Posts: 1440
  • Total likes: 454
  • Do good for others, good will happen to you !
Re: Most imortant characteristic to match for equivalent JFETs?
« Reply #4 on: July 28, 2022, 10:33:47 AM »
Please post pertinent part of your schematic.

For buffers and switches, FET can be quite forgiving

I understand self stabilising FET amps are also quite stable


But for compressor, distortion stage, matched pairs etc, one needs more care.

Mark Hammer

Re: Most imortant characteristic to match for equivalent JFETs?
« Reply #5 on: July 28, 2022, 12:24:46 PM »
Don't know about your particular circuit but this: https://www.analogisnotdead.com/article25/circuit-analysis-the-boss-bd2 might help you with regard to items configuration/values..

P.S.
Forgot to mention R.G.'s article about JFETs matching.. :icon_redface:
http://geofex.com/Article_Folders/fetmatch/fetmatch.htm
Matching for gain in differential pairs may be different than matching for purposes of usable drain-source resistance range in phasers.  RG's article is directed at the second one.

Kevin Mitchell

Re: Most imortant characteristic to match for equivalent JFETs?
« Reply #6 on: July 28, 2022, 04:11:42 PM »
Don't know about your particular circuit but this: https://www.analogisnotdead.com/article25/circuit-analysis-the-boss-bd2 might help you with regard to items configuration/values..

P.S.
Forgot to mention R.G.'s article about JFETs matching.. :icon_redface:
http://geofex.com/Article_Folders/fetmatch/fetmatch.htm
Matching for gain in differential pairs may be different than matching for purposes of usable drain-source resistance range in phasers.  RG's article is directed at the second one.
Yup. Matched VBE is what you're going to want for differential pairs - rather than specs for variable resistance.
Some folks even match for HFE on top of VBE but I'm convinced that is not necessary.
  • SUPPORTER
I'm archiving all of my DIYLC work for the home-etchers - Project Layouts & Templates

antonis

Re: Most imortant characteristic to match for equivalent JFETs?
« Reply #7 on: July 28, 2022, 04:25:43 PM »
@Kevin: JFETs don't have neither VBE nor hFE..
(VGS and gm, perhaps..??)
"I'm getting older while being taught all the time" Solon the Athenian..
"I don't mind  being taught all the time but I do mind a lot getting old" Antonis the Thessalonian..

Kevin Mitchell

Re: Most imortant characteristic to match for equivalent JFETs?
« Reply #8 on: July 28, 2022, 04:47:01 PM »
@Kevin: JFETs don't have neither VBE nor hFE..
(VGS and gm, perhaps..??)
You're right. I suppose I meant VGS which R.G's circuit actual measures for.
I've yet to use jfets in differential pairs. My head is in BJT world.
  • SUPPORTER
I'm archiving all of my DIYLC work for the home-etchers - Project Layouts & Templates

antonis

Re: Most imortant characteristic to match for equivalent JFETs?
« Reply #9 on: July 28, 2022, 05:13:36 PM »
My head is in BJT world.

Your head is in the right side of the world.. :icon_wink:
"I'm getting older while being taught all the time" Solon the Athenian..
"I don't mind  being taught all the time but I do mind a lot getting old" Antonis the Thessalonian..

Rob Strand

Re: Most imortant characteristic to match for equivalent JFETs?
« Reply #10 on: July 28, 2022, 08:10:48 PM »
If I am trying to find an alternative model JFET to use in a discrete op-amp config like in a blues driver, which are the most important numbers on the data sheet that I should be looking to match? Vgs, Idss, etc? The package doesn't need to be the same btw.
Or, if I can't find a good match, is it possible to tweak some resistor values to compensate and get the same result fairly easily?
Thanks

The JFETS used in the Boss discrete opamps have very specific characteristics in order that the discrete opamp has enough open loop gain.    Some years back a forum member used more common JFETs and wasn't happy with the gain.   I checked via a spice simulation it confirmed that result.

If you look at the datasheet for 2SK184GR you will get:

   min   max   typ
IDSS   2.6   6.5   ?               [mA]
VP   0.2   1.5   ?               [V]
Yfs   4000   ?   15000        [uS]

You might estimate typical values for IDSS at 4.2mA to 4.6mA and VP=0.55V to 0.61V.
However the plots of IDSS and VP in the datasheet imply Yfs could be higher say 19000uS
and that could mean VP is a little lower than 0.55V to 0.61V.

The key JFET features is a low VP and a large Yfs.    Those specs certainly limit the choices of JFETs.

The high Yfs more or less determines the gain but you don't want VP to be too far off because that will upset the JFET bias currents.   It is possible to tweak the JFET bias current by adjusting the source resistor on the differential pair.   If VP is way off the diff amp could start to behave differently.  The BD2 operates at 8V so you have a bit more slack than some of the other Boss pedals which have 5V supplies on the discrete opamp.

You could also move to SMD devices.  I believe the BD2 SMD versions uses 2SK880GR devices.


FWIW, the issue is *choosing* a JFET with suitable specs so the device works.   We aren't really matching a group of JFETs.    If we wanted to hand select JFETS which were close to the Boss JFETs then we might need to *measure* the JFET parameters.    (Remember RG's *matching* jig measures Vgs values not VP,  the measured Vgs values are lower than VP by some factor like 1.2 or 1.7.)   The problem is the JFETs used in Boss discrete opamp aren't like common JFETs, esepcially the high Yfs.   You might be able to use switching JFETs (if you can get low VP).



« Last Edit: July 28, 2022, 08:42:05 PM by Rob Strand »
Send:     . .- .-. - .... / - --- / --. --- .-. -

Rob Strand

Re: Most imortant characteristic to match for equivalent JFETs?
« Reply #11 on: July 29, 2022, 01:50:57 AM »
FWIW, I had a closer look at picking the Boss diff-pair JFETs on the discrete opamp without losing gain.  I didn't look-up my old notes.

In broad terms if you can choose the JFETs such that,

         beta > 7000u  (u=1e-6)

         where,
               beta = IDSS / VP^2
               IDSS and VP are the common JFET parameters.

then the maximum gain of an unmodified stage will be within 1dB of the maximum gain of the typical case.  The 1dB matching constraint is chosen fairly nominally but isn't unreasonable.   FYI the nominal beta is about 19000u.

Gain isn't the only thing which affects the sound.   I can imagine if we kept VP < 1V as well then the circuit would perform more or less the same as a VP of 0.5V.  Going to say VP = 1.5V might affect things, it would need a sheet-load more investigation.

===>   So the big problem with common JFETs is beta ends up around 1000u to 3000u, perhaps stretching to 5000u.  Much lower than the desired lower limit.   In other words common JFETs are likely to produce a noticeably lower maximum gain.

FYI:  If yfs0 (=2*IDSS/VP) is available in the datasheet you can eyeball beta from beta = (1/2)Yfs0/VP. If VP is 0.5V to 1.0V then beta = Yfs0 to Yfs0/2 depending on VP.   For high beta we need a high Yfs0, in the order of 7000uS to 14000uS *minimum*.  Most common JFETs have Yfs0 < 6000us.   This is just another way to eyeball the fact normal JFETs don't cut it.

As far as tweaking the circuit goes.  I didn't look at in detail other than general intuition and a few simulations to confirm my suspicions.

When the correct JFETs are chosen it might come as a surprise that the two JFETs don't run at the same bias current.   Why's that?   It helps push the gain up (even with the right JFETs the maximum gain is less than an opamp for the same set of gain resistors.)    The gain is increased by choosing the 2k2 drain resistor relative high and that causes a skew in the bias currents.   You wouldn't want to go any higher.

In order to get more gain we can decrease the source 4.7k resistor *but* this is undesirable as it makes the bias current skew on the JFET even worse.   If we try to reduce the 2k2 to compensate then there's no gain improvement or even a gain loss.

The only clean way I could see to get more gain is to use a high gain (hFE) transistor for the PNP.   That has only has a small effect.  The reason it works is it decreases the AC load on the first JFET which increases the gain.  It does this by increasing rpi (the input resistance of the PNP transistor.)

All in all there's no simple way to get more gain with giving-up something.   In this case it would mean a lot of testing to make sure you haven't stuff-up the sound of the discrete opamp.

So the conclusion is it is best to use JFETs close to the original in order not to lose gain.  That's the same conclusion I came to last time I looked at this issue.   It's also fairly obvious as-is the Boss design has already been pushed quite hard to get it to work at least something like a normal opamp when the gain control is set high.

« Last Edit: July 29, 2022, 07:45:39 PM by Rob Strand »
Send:     . .- .-. - .... / - --- / --. --- .-. -

Clint Eastwood

Re: Most imortant characteristic to match for equivalent JFETs?
« Reply #12 on: July 29, 2022, 03:12:23 AM »

===>   So the big problem with common JFETs is beta ends up around 1000u to 3000u, perhaps stretching to 5000u.  Much lower than the desired lower limit.   In other words common JFETs are likely to produce a noticeably lower maximum gain.


The readily available J113 has a relatively high beta of  around 9000u. But Vp is greater, so to make it work the biasing will have to be changed.

Rob Strand

Re: Most imortant characteristic to match for equivalent JFETs?
« Reply #13 on: July 29, 2022, 07:04:46 PM »

===>   So the big problem with common JFETs is beta ends up around 1000u to 3000u, perhaps stretching to 5000u.  Much lower than the desired lower limit.   In other words common JFETs are likely to produce a noticeably lower maximum gain.


The readily available J113 has a relatively high beta of  around 9000u. But Vp is greater, so to make it work the biasing will have to be changed.
That one should work.  The typical VP should be OK but extreme devices with high VP might need help.

I didn't check the J113 because I remember there was a problem with the datasheet.  It turns out it was the pinout.

https://www.diystompboxes.com/smfforum/index.php?topic=126803.msg1214064#msg1214064
Send:     . .- .-. - .... / - --- / --. --- .-. -

Rob Strand

Re: Most imortant characteristic to match for equivalent JFETs?
« Reply #14 on: July 30, 2022, 12:16:12 AM »
Here's a thread for a BJT version which tries to loosely match the JFET version,
https://www.diystompboxes.com/smfforum/index.php?topic=129424.0
Send:     . .- .-. - .... / - --- / --. --- .-. -