JFETs for Phasers? (2020 Edition)

[trueno07]

Hey all

Looking to build a phaser or two. I have my eyes on the Maestro Phase Shifter, but my question isn't specific to the Maestro.

It seems like the 2N5485 recommended is getting harder to find and more expensive. I've read some that say the J201 or 2N5457 is OK and an equal number of people say it's not OK. Regardless, the 201 is harder to find now in the TO-92 format at a reasonable price. I've also read that the J112 is OK, but i'm not exactly sure if that's true or not.

So, if you were building a phaser in 2020 what JFET would y'all recommend? I'm 100% Ok with buying in bulk and matching myself. I'd rather do that and gain the experience than pay for a matched set.

[rockola]

It seems like the 2N5485 recommended is getting harder to find and more expensive. I've read some that say the J201 or 2N5457 is OK and an equal number of people say it's not OK. Regardless, the 201 is harder to find now in the TO-92 format at a reasonable price. I've also read that the J112 is OK, but i'm not exactly sure if that's true or not.

So, if you were building a phaser in 2020 what JFET would y'all recommend? I'm 100% Ok with buying in bulk and matching myself. I'd rather do that and gain the experience than pay for a matched set.

In 2020, it's time to embrace SOT-23. All of the above JFETs are readily available from e.g. Mouser as as MMBFJ201, MMBF5457, MMBF5485, and MMBFJ112. You'll have to get/design a PCB with SOT-23 pads, or use an adapter board. Hand soldering SOT-23 is not super difficult.

Matching JFETs is trickier with SOT-23, purely from a handling perspective. I recently bought a roll (1000 pcs) of both MMBFJ201 and MMBF5457, which should keep me (and my estate) in JFETs for life. In 1000pcs lots the unit price of all of the above JFETs is around the 10c mark. However, I have yet to build a phaser with either, because I haven't been able to figure out how to match a bunch of SOT-23 components without losing my sanity (and without dropping half of them on the floor). All the JFET matcher designs out there are made for through-hole components. It would be easy enough to add pads for SOT-23 but the actual process of getting the components out of the roll, into the matcher, securing them while the matcher runs, and keeping them sorted eludes me. If anybody has figured out a good workflow for this I'm all ears.

[Elijah-Baley]

I heard that the smd version of the JFET was more accurate, finer, so even easier to match them. Was it false?

[rockola]

I heard that the smd version of the JFET was more accurate, finer, so even easier to match them. Was it false?

I've heard that too. My beef is with the physical handling of the components, not the actual matching.

[rankot]

Regarding this, shall they all be the same, and which parameter is the most important to match? I_dss or V_gs(off)? What is the "ballpark" required - is it the same for all the phaser designs, or not? I wanted to build Ibanez Phaser, but couldn't find required values for JFETs, there's just remark on schematic that they need to be matched.

[merlinb]

Regarding this, shall they all be the same, and which parameter is the most important to match? I_dss or V_gs(off)? What is the "ballpark" required - is it the same for all the phaser designs, or not? I wanted to build Ibanez Phaser, but couldn't find required values for JFETs, there's just remark on schematic that they need to be matched.

In general, V_gs(off) takes priority. That's the thing you should match first. The bias trim can be adjusted to suit whatever V_gs(off) batch you have selected (within reason). With V_gs(off) matched, all the JFETs will be equally modulated. This gives smooth phasing, avoiding thumps. It's nice if there is also an LFO amplitude trimmer.

Of course, it's nice if you can match I_dss as well, but do that as a secondary consideration. Matching I_dss means all the JFETs will sweep down to the same minimum resistance, meaning you get the largest range of 'swoosh'.

Thumps are more annoying than losing a bit of swoosh that you never knew was possible anyway. That's why V_gs(off) takes priority.

[rankot]

Thanks, Merlin!

[Mark Hammer]

SK-30A JFETs have been used in a number of phasers.  I have no idea of their current availability or form factor.

[trueno07]

It seems like the 2N5485 recommended is getting harder to find and more expensive. I've read some that say the J201 or 2N5457 is OK and an equal number of people say it's not OK. Regardless, the 201 is harder to find now in the TO-92 format at a reasonable price. I've also read that the J112 is OK, but i'm not exactly sure if that's true or not.

So, if you were building a phaser in 2020 what JFET would y'all recommend? I'm 100% Ok with buying in bulk and matching myself. I'd rather do that and gain the experience than pay for a matched set.

In 2020, it's time to embrace SOT-23. All of the above JFETs are readily available from e.g. Mouser as as MMBFJ201, MMBF5457, MMBF5485, and MMBFJ112. You'll have to get/design a PCB with SOT-23 pads, or use an adapter board. Hand soldering SOT-23 is not super difficult.

Matching JFETs is trickier with SOT-23, purely from a handling perspective. I recently bought a roll (1000 pcs) of both MMBFJ201 and MMBF5457, which should keep me (and my estate) in JFETs for life. In 1000pcs lots the unit price of all of the above JFETs is around the 10c mark. However, I have yet to build a phaser with either, because I haven't been able to figure out how to match a bunch of SOT-23 components without losing my sanity (and without dropping half of them on the floor). All the JFET matcher designs out there are made for through-hole components. It would be easy enough to add pads for SOT-23 but the actual process of getting the components out of the roll, into the matcher, securing them while the matcher runs, and keeping them sorted eludes me. If anybody has figured out a good workflow for this I'm all ears.

Well, seems like it's 2020 and it's time to come up with a way to match SMD JFETs.

I knew they were selling the 201 in a JFET package, didn't know that the others were being sold as well. Thanks!

[rankot]

I have ordered few PCBs for ROG JFET tester which support SMD, so if anyone's interested, they'll be available soon for cheap. I can also share gerber files if one wants to make their own for non-commercial use.

[StephenGiles]

I would use a 4009 CMOS invertor chip, which contains 5 MOSFET transistors inside. By using a single integrated circuit for all transistors means that no matching of devices is required as would be the case if we were using discrete components.

This of course was used in the second version of the EH Badstone and more recently by Oakley and for tired eyes is still available in DIL!!!!!!!!!!!!!!!!!!!!

[Mark Hammer]

That's 5 MOSFet transistors for each of the 6 buffer/convertor sections it contains.  It is not like a 3046 or 3086 transistor array with 5 discrete transistors.

But yes, they can be used as if they were voltage-dependent resistors.  They are, however, a little less immune to clipping than JFETs, which means you have to be careful about input signal levels.

[StephenGiles]

That's 5 MOSFet transistors for each of the 6 buffer/convertor sections it contains.  It is not like a 3046 or 3086 transistor array with 5 discrete transistors.

But yes, they can be used as if they were voltage-dependent resistors.  They are, however, a little less immune to clipping than JFETs, which means you have to be careful about input signal levels.

Ah I forgot about the clipping problem. Is there a trick to clamp input to a safe level - simply, and then unclamp it before the output, but not a compander? I'm sure I read something about that.

[Mark Hammer]

Years back, Mike Irwin had discussed this very problem with me regarding the ETI phaser (shown below), that used a 4049 as the control element for 6 stages.  The challenge is that one can't easily incorporate the sort of network one sees in the Phase 45 and similar, around the FET to provide distortion immunity in the face of hotter inputs.  All one can do is keep the input level modest - perhaps with some padding - and then boost the mixing stage. 
You can see here that the input to the first phase-shift stage is attenuated by the 10k/1k divider, and then at the mixing stage, the wet signal uses a 5k6 mixing resistor, compared to the 56k resistor the dry signal employs, in order to achieve a 50/50 balance when mixed.  And I assume the transistor stage provides some overall gain to the mixed signal.
That's certainly not a recipe for "disaster", but it imposes some constraints on design, and some necessary compensations that can undermine things like S/N ratio.

[StephenGiles]

It would be interesting to see the schematic of the Oakley MOS Phaser http://www.oakleysound.com/MOSPhaser1-BG.pdf

[nickbungus]

J112s or j113s work well in powders. The vgs in j113s is a tad lower

[Yazoo]

I went through the process of matching smd Jfets for a Jet Phaser build. I didn't do anything fancy. I just etched a small footprint board and wired it up to the fet matcher circuit on a breadboard. I then just clamped each transistor, remembering not to sneeze! Probably not the best way but it did work.

[POTL]

Hello everyone.
Maybe I'm wrong, but it seems to me that in 2020 it makes sense to build phasers on optics or ota.
Optics:
1) We have no problems with clipping, as in phase 90
2) We do not need to match components
3) We do not need to look for obsolete parts
4) Optics easily accessible
OTA:
1) You still have no problems with clipping
2) You still do not have to match cosponents
3) Yes they are overbearing, but still available everywhere
Yes, they sound different, but I'm sure the circuits can be adapted for better phase 90 sound.

[Mark Hammer]

It has almost always made more sense to go with LDRs or OTAs as control elements, for all the reasons you note.  If the manufacturer has access to a decent-enough supply, I suspect that FETs were often preferred because they didn't have to take up as much room - especially now that they are available in tiny surface-mount form - and could also be compensated to be a little more immune to clipping.

But that said, I think LDRs are often where it's at.  One of the advantages they provide is that they can often accomplish a sort of automatic adjustment of sweep width at higher modulation speeds.  The advantage that OTAs provide is that they can be easily reconfigured to filter modes other than allpass with only a few minor changes.