Phase 45 not working with 2n5952s

[Seljer]

So I found the Tonepad PCB of it I etched a couple of years ago and finally populated it yesterday evening. http://www.tonepad.com/project.asp?id=13

Using the specified 2n5952 JFETs I just couldn't get it to phase, it was passing audio and I could hear the LFO ticking and it did seem to have the "stuck phaser" sound so it seems that the JFETs were working. For the hell of it I tried some J201 FETS and later some 2N5457s and yay, it works. I can't however for the life of it get the sweep to sound smooth, espescially with slower settings. When the LFO comes around around it just gives a huge barrage of bass. Is this just a matter of getting a good pair of transistors and setting the trimmer to the sweet spot? The videos of this pedal on youtube sound much smoother.
Also, I've got like 50 of the 2n5952s with a couple of well matched sets as I ordered them purposely for phasers and I'm rather annoyed that I can't use them.

Using the JFET matcher from geofex, the Vgs voltage for the 2n5952 pair i used is around 1.7V, the J201s give about 200mV and the 2n5457s give around 600mV. I also have 2n5485s which all have around 2V and don't work in the circuit either

The 7k5 resistor is an 8k2, instead of the 250k trimpot I used a 100k because those where the only ones I had that fit the footprint on the PCB.

Voltages are as follows:

Supply voltage: 8.96V

Zener diode voltage (I used a 5.1V zener diode because it's what I had on hand) : 3.86V

Opamp 1 (TL072)
1 : 4.00V
2 : 3.86V
3 : 3.85V
4 : 0V
5 : 3.65V
6 : 3.75V
7 : 3.71V
8 : 8.96V

Opamp 2  (TL072)
1 : jumps between 1.2V and 8.3V
2 : travels from 3.6V to 5.6V
3 : jumps between 3.4V and 5.7V
4 :  0V
5 : 3.86V
6 : 3.86V
7 : 3.74V
8 : 8.96V

Transistor 1 (J201)
G : move around between 2.54V to 3.18V
S : 3.85V
D : 3.86V

Transistor 2 (J201)
G : goes from 2.58V to 3.86V
S : 3.86V
D :3.86V

The wierd issue is that the gate voltage on the second transistor goes half a volt higher than the other one! It does this even if there is no transistor installed in the socket.

I triple checked all the component values and for solder brigdes or bad connections and can't find anything so I'm stumped. Anyone get any ideas?

Also, for the ticking: is it worth increasing the size of the filter capacitors or will I have better luck just replacing the second opamp with a lower power device?

[Seljer]

It works! After 2 hours of progressively desoldering things while checking with the voltmeter it began to behave. Something was shorting out apparently (maybe the sockets I used) and pulling up the voltage up the middle lug of the trimmer, stopping the bias voltage from going low enough for the 2n5952s. I've soldered the transistors in directly and after a quick tweak of the trimmer I've got a nice swirly musical phaser, so...yay
Do as you wish with this thread.

[Paul Marossy]

Mine works with J201s, although I've been told that it's not supposed to.

[deadastronaut]

^ yep same here j201's work fine...

[Seljer]

Alright, I'm going to bump this thread back up because I've come to yet another phaser issue. Building the Phase 45 inspired me to dig up my Phase 90 that never worked too well. I found the main culprit rather quickly (one leg of the output buffer transistor had a cold solder joint).

Heres a link for the schametic http://www.tonepad.com/project.asp?id=7

The problem this time is that it's noisy, I've got a tornado of a whooshing phase sound when I'm not playing and its loud enough for me to deem it unacceptable. With an audio probe I'm getting very loud a chirping sound in time with LFO at pin 6 of IC3 and varying levels of the whooshy noise around all the other pins of IC2 and IC3 (though pins 2 and 3 of IC2 seem to be ok).

With a multimeter all the signal input/output pins of IC2 and IC3 measure 3.11 volts (same as the voltage on the zener diode), except for pins 2 and 6 of IC3 which measure 3.26 volts (and it moves around a bit with the LFO). I'm not sure if this is normal.

Changing opamps shifts the voltages around a bit (I'm currently using TLC272) but the issue is still there.

It quietens down quite a bit if I remove the feedback resistor. Is this just something I have to live with if I have a phaser with feedback? That it takes whatever noise may be present on my guitar cable and emphasises it noticeably? The circuit is dead silent if I ground the input.


edit: also, for now this is running hooked up with alligator clips on top of my desk so thats probably not helping the noise floor

[kaycee]

I've been mucking about a lot with P45/90 circuits of late, building on vero although I do have a couple of madbean Smoothie boards laying around for when I finish modding.

Yes, you will get some swooshing noise when your not playing, and the more (noisey) gain you have going on the more you'll hear it. For clean it will only be the normal amp buzz that you'll hear phasing away. And yes naked the circuit tends to tick, but with short wires and boxed its quiet.

The feedback/resonance is taking the phased signal and feeding it back in to the phasing stages, so if you've got noise in the first place, you will of course get more noise via that. When its working properly the feedback/resonance mod just makes phasing deeper and more pronounced, not noisier.

I think that you will find that you have a cold joint or solder bridge somewhere, a good check and reflowing your joints may well solve it, good luck!

[Paul Marossy]

Just thought I'd throw this out there:

I bought some 2N5952s the other day so I could get my Phase 45 circuit "the way it's supposed to be". Popped them in there and NO phasing at all. Put the J201s back in it, and it works.

This is weird. It's supposed to be the other way around from what I've been told. 

[deadastronaut]

^ yep i had exactly the same thing, the only fets that worked for me were the j201's...i tried quite a few out.

the j201's weren't fussy on matching either..

[Paul Marossy]

^ yep i had exactly the same thing, the only fets that worked for me were the j201's...i tried quite a few out.

the j201's weren't fussy on matching either..

That has exactly been my experience too.

[R.G.]

J201s are odd, even for JFETs. They have  very low Vgsoff, from as large as -2V to as little as -0.1V according to the datasheet. They also have a small Idss, which is the biggest current that can flow through the channel.

In a JFET used as a switch, you have to have enough Vgs reverse bias so the gate is reverse biased with all signals. For the J201, Vgsoff is so small that you're tempted to only use a volt or two to turn it off. But the source voltage moves with the signal in most cases, so Vgs has to be big enough that the sum of signal peak voltage plus Vgsoff has to still leave the device turned off.

In JFETs used as variable resistances, the Vgs bias is set between -0V andVgsoff. The problem with this is that the signal voltage on the source also adds/subtracts from the Vgs. So the signal modulates the effective Vgs and so self modulates the channel resistance. This can produce a form of distortion in signal resistance situations, envelope modulation in compressors and attenuators, and phase shift distortion in phasers, as well as out and out signal distortion when the signal gets much bigger.

It's not that J201s can't work in a P45 or P90. It's that they're not the best choice unless one uses small signals and/or likes the oddities it introduces. I've posted the idea that the J201 is not the best choice for this application before, and that's probably the origin of the gossip-chain idea that they won't work.

I suspect the 2N5292 issues are from not having quite the right bias conditions for them as compared to the setup for J201. The bias settings are different, as are the sizes of the LFO voltage needed to sweep them.

[Paul Marossy]

In JFETs used as variable resistances, the Vgs bias is set between -0V andVgsoff. The problem with this is that the signal voltage on the source also adds/subtracts from the Vgs. So the signal modulates the effective Vgs and so self modulates the channel resistance. This can produce a form of distortion in signal resistance situations, envelope modulation in compressors and attenuators, and phase shift distortion in phasers, as well as out and out signal distortion when the signal gets much bigger.

I can hear a little bit of distortion in my Phase 45 clone, that's probably why. But it's not an unpleasant thing. It adds to the "complexity" in a nice way.

It's not that J201s can't work in a P45 or P90. It's that they're not the best choice unless one uses small signals and/or likes the oddities it introduces. I've posted the idea that the J201 is not the best choice for this application before, and that's probably the origin of the gossip-chain idea that they won't work.

The funny thing is that I've tried like half a dozen other types, and none of them worked. Although I didn't also monkey with the zener diode when I was experimenting. I would like to build a Phase 90 sometime, but I am not so sure what the outcome would be now.  

[R.G.]

In the P90 and P45, the zener sets the drain and source voltage (among other things), and the trimmer sets the Vgs value to put the gate-source voltage in the middle of the full-on to full-off range. The LFO voltage then wiggles the actual Vgs around that middle-ish setting, moving the resistance of the JFET channel up and down and causing the phase shifting action.

Each JFET will have its own unique Vgsoff, as well as its own curve of drain-source resistance (rds)  per Vgs values. The idea of matching is to get several JFETs close enough so that the same Vgs that they all get has them doing similar rds values at the same time. It's less pleasing as a phaser if one or more of them either is way different from the others, or bumps into minimum or maximum rds before the others.

It's also better if the LFO voltage is neither so small that the rds value doesn't move much, nor so big that a lot of the LFO swing is wasted because the JFETs are already outside their minimum or maximum rds range. The trimmer in the P90 and P45 let you set the center of the LFO swing, but not the size of the LFO waveform. Getting either one too far from the "good" range of JFET rds will make for a less useful sounding phaser.

The P90 and P45 designs were set up for an LFO size that complements the 2N5292 and it's 3-to-5V Vgsoff. The 2SK30 and 2N5485 have very similar ranges. The J201 is significantly smaller. But the thing is, if you like the way a J201 sounds, great. Use what sounds good.

But I'm not surprised that subbing in a 2N5292 in a circuit where you like the sound of a J201 would be less than pleasing. If the 201 sounds good, the setup is going to be non-optimal for a 5292 and others.

[Paul Marossy]

What I am wondering about is why is it that some people can use J201s not 2N5952s, and vice versa. I understand about the Vgs and the relationship to the JFETs working properly or not, but it seems like there is some other factor involved. Does the opamp used make a difference?

[R.G.]

What I am wondering about is why is it that some people can use J201s not 2N5952s, and vice versa. I understand about the Vgs and the relationship to the JFETs working properly or not, but it seems like there is some other factor involved. Does the opamp used make a difference?

It's the pesky details of the circuit, I suspect, not the people. I personally have used 2N5292, 2N5485, and 2SK30. You have used J201. So we know all of those work. As for exactly what makes them work or not, it's in the circuit. For instance, we'd have to see exactly what the bias point was and the extent of the LFO swing was for all cases to see where they phased most pleasingly.

It's probably not the opamp. In all cases the cap and JFET are connected to the (+) input of the opamp, and to a first order, that is too high an impedance to make the opamp have an effect.

My best guess is the circuit setup: the zener voltage, the trimmer/bias voltage for the gates, and the size of the LFO voltage, all of which will vary. It should be possible to tune the basic circuit to work for any of the JFETs. My guess is that there are different tuning points for them.

I guess that is the same as saying the "other factors" are the details of the circuit and setup. 

[Paul Marossy]

It's probably not the opamp. In all cases the cap and JFET are connected to the (+) input of the opamp, and to a first order, that is too high an impedance to make the opamp have an effect.

My best guess is the circuit setup: the zener voltage, the trimmer/bias voltage for the gates, and the size of the LFO voltage, all of which will vary. It should be possible to tune the basic circuit to work for any of the JFETs. My guess is that there are different tuning points for them.

I guess that is the same as saying the "other factors" are the details of the circuit and setup. 

Hmm... so maybe some of it has to do with the variability of the parts (like tolerances, etc). Of course, the FETs we know can be all over the place. And in the originals, those low voltage ceramic caps they used must be all over the place tolerance-wise. I wonder if they are like the old wahs where some sound really good and others just don't sound very good.

[R.G.]

The devil is always in the details. 

[Paul Marossy]

The devil is always in the details. 

Always! 

[Jaicen_solo]

To be honest, I found in my limited fumbling with the 45/90 circuit that pretty much all the FET's I have can be made to work with the right combination of Zener diode and trimmer setting.

[R.G.]

That's what I'd expect.

[kaycee]

Hmmmm..? So I've found that I've built a couple of these that either wouldn't phase, or phased weakly whatever jfet I put in them. I did try a higher value trim, but didn't alter the zener diode, if I'd of changed that for a different voltage maybe they'd of worked?

Whats the consensus on the value for the zener? I've seen both 4k7 and 5k1 quoted, anyone got a favoured value or tip on that front?