J113 for phasers

[Aentons]

I put this set of 2n5952 from small bear in it and it is slightly less distorted but it's still there, still noticeably more than the VFE prod build. The bias trimmer is now fully maxed out to get these into the correct range. It was just lower higher than mid with the j113s

[Aentons]

Here is a quick video of the difference

Both of these have 2n5952
[youtube width=480]https://youtu.be/M5apzyOt2tQ[/youtube]

[Aentons]

And here is a video with J113 vs 2n5952
[youtube width=480]https://youtu.be/XC9xG65Zk4s[/youtube]

[jimilee]

I noticed the volume is considerably louder when you activate the tractor beam. What happens when you turn the volume down? Also, when you wail on a phaser, it will sound like that. The phase 90 does the same thing. If you, or any guitar player, is going to be wailing on it like that, usually distortion is involved anyway. I like the 5952s better than the J113s. Around 5:27, I think you'll hear the same distortion coming in, no distortion pedal, this is straight. It's just the nature of the beast.

[Aentons]

I noticed the volume is considerably louder when you activate the tractor beam. What happens when you turn the volume down? Also, when you wail on a phaser, it will sound like that. The phase 90 does the same thing. If you, or any guitar player, is going to be wailing on it like that, usually distortion is involved anyway. I like the 5952s better than the J113s. Around 5:27, I think you'll hear the same distortion coming in, no distortion pedal, this is straight. It's just the nature of the beast.

That's just the output level trimpot not set quite right. Turn it up, turn it down, distortion sounds the same. I forgot to demonstrate that turning the guitar volume down does clean it up.

Sorry for the wailing... just trying to show what the extreme input did to the signal. Im ok with the distortion being there, that's part of why I like the phase 90 so much, but I think it is noticeably more extreme with the J113.

Im not quite sure I understand why the distortion is better/worse with different JFets as long as they fit the spec and are biased properly. Some people don't have an issue with j113s in a phase 90 and some people have trouble with 2n5952's in them. It seems to be hit and miss. I'm just looking to understand the possible differences and potential fixes.

I'm going to try some different chips to rule out any trouble with those...

[jimilee]

I noticed the volume is considerably louder when you activate the tractor beam. What happens when you turn the volume down? Also, when you wail on a phaser, it will sound like that. The phase 90 does the same thing. If you, or any guitar player, is going to be wailing on it like that, usually distortion is involved anyway. I like the 5952s better than the J113s. Around 5:27, I think you'll hear the same distortion coming in, no distortion pedal, this is straight. It's just the nature of the beast.

That's just the output level trimpot not set quite right. Turn it up, turn it down, distortion sounds the same. I forgot to demonstrate that turning the guitar volume down does clean it up.

Sorry for the wailing... just trying to show what the extreme input did to the signal. Im ok with the distortion being there, that's part of why I like the phase 90 so much, but I think it is noticeably more extreme with the J113.

Im not quite sure I understand why the distortion is better/worse with different JFets as long as they fit the spec and are biased properly. Some people don't have an issue with j113s in a phase 90 and some people have trouble with 2n5952's in them. It seems to be hit and miss. I'm just looking to understand the possible differences and potential fixes.

I'm going to try some different chips to rule out any trouble with those...

Hey, I get it. I guess it just comes down to parts and part quality. You have a significant amount of gain in the input section I believe, that's what it sounds like to me.


Sent from my iPhone using Tapatalk

[Aentons]

It occurred to me that one of the differences between mine and the prod build is that I have the suggested noise reduction cap across the diode. I used a 22uF. So I checked the voltage on the zener diode and this is what I got:

VFE prod = -3.32 V
mine = -2.52 V

I clipped one leg of the cap from the diode but I still get -2.52.
That voltage doesn't move when I tweak the trimmer.
That value seems low.

I also measure 2.06v across trim pot which seems low as well


Edit: I posted a question specifically about the zener over here. Turns out, that's not the issue.
https://www.madbeanpedals.com/forum/index.php?topic=32271.msg310154#msg310154

[Aentons]

I did some searching an found a great answer about jfet distortion from RG here:

https://www.diystompboxes.com/smfforum/index.php?topic=39904.0

"The resistance of a FET channel also varies with the applied Vds. This variation of channel resistance with Vds is smaller than with the Vgs, but it does exist. This variation of channel resistance with both Vgs and Vds leads to intermodulation of the signal with the LFO in phasers. The distortion is unrelated to the notches and peaks. The signal through each phase stage gets distorted by the intermodulation. Small enough signals are not distorted by objectionable amounts. The exact details of this intermodulation varies from FET to FET."

That makes a lot a sense.

I'm still trying to understand why my diy and prod TB builds have noticeably different amounts of distortion when played side by side on the same rig.

At this point, I can only only assume that the JFets in the prod build have more carefully chosen(higher?) Vgs(off) and possibly other characteristics.

I found a chart of some mitigation techniques that keep the input and jfet voltage interaction to a minimum. #1 is like the Phase90. The P45/401 technique is not shown here but seems similar to #2


TLDR:
The extra parts around the jfets in the Phase45/DOD401 are there to lessen the interaction of the guitar's input signal on the JFet's variable resistance which (if not present) would cause intermodulation distortion in the output signal. So for the Phase90 you have to choose your JFets more carefully and not just match vgsoff

[jimilee]

Right, as RG graphed out on this page:

http://www.geofex.com/article_folders/fetmatch/fetmatch.htm

2.5 is a good starting range. That's why was surprised  and mentioned the low VGs jfets you matched in one of my previous responses. I knew I'd seen what you were saying about higher VGs before. Nice research.


Sent from my iPhone using Tapatalk

[Aentons]

Right, as RG graphed out on this page:

http://www.geofex.com/article_folders/fetmatch/fetmatch.htm

2.5 is a good starting range. That's why was surprised  and mentioned the low VGs jfets you matched in one of my previous responses. I knew I'd seen what you were saying about higher VGs before. Nice research.


Sent from my iPhone using Tapatalk

Hmm.. maybe I'm reading it wrong but I thought that article is just talking about matching vgsoff numbers and the 2.5 range just happened to be where that particular set of 5485's had good matched groups. It makes no mention of specific ideal ranges or whether you should look for matches near the minimum or maximum rating.

I think I remember reading somewhere RG did mention an ideal range for a specific jfet. It was maybe something like -1.4 - 1.8v.  but ( I think ) any ideal range would depend on which type of jfet you are talking about and (for instance) that's why you can use really low vgs  j201s. Anyone please feel free to correct me. I'm still confused about how to identify a good range for a specific type jfet (like the J113) that would minimise the intermodulation distortion.  RG's comment... "Small enough signals are not distorted by objectionable amounts" makes it sound like the LFO bias voltage should be as low as you can get it, but I'm not sure. I think I read that the lfo voltage wiggles within half a volt. So, if that is the case, then (it seems like) to minimize IM distortion you would want a vgsoff of -0.5v (or just above) in all cases.

I'll see if I can find the article.


Edit: Here is part of RG's comment from this post:
https://www.diystompboxes.com/smfforum/index.php?topic=79931.0

"I would advise using 2N5485 if you can find it. It's in current production. But if you simply must use a BF245 (temporarily, while you find the best parts, maybe), use the BF245A. Match the JFETs for Vgsoff. I like most JFETs at about 1.4 to 1.8V in this circuit, so get a set where they all measure the same, and within that range."

[StevieF]

<edited - I meant to say 5-6x more sensitive not 50x lol>
I just built myself a phase 90 clone using j113. I was a unhappy with some of the tone (and clipping distortion) I was getting so I used Tina sim to simulate the difference between the "ideal" (ha) versions of each jfet. I swept the notch up and down and discovered that that (according to the sim) the j113 is about 5-6x more sensitive across the parallel 22k than the 2n5952. the LFO voltage swing reaching the fet gate is about 250-300mv peak to peak, but I can make the whole j113 sweep in about 50mv. So to compensate I changed each of the parallel resistors for 4.7k and the 47nf caps on the non inverting input to the filter sweep op amps were changed to 220nf (to keep the filter frequency in the same place).
I'm much happier with this sweep tone, although its still clipping a bit. I think its due to the PNP collector voltage being too high, so its not swinging in the center of its range. I'll drop the collector resistance further (at 10k compared to original 56k atm) and let you know how I get on.
Also the 10k in series with the zener isn't enough to properly reverse bias the zener, so I used a 1k to give it enough current. I'm using a psu not a battery so I'm not bothered. It still only draws 6mA ish anyway which is hardly a lot.
Steve

[Aentons]

I just built myself a phase 90 clone using j113. I was a unhappy with some of the tone (and clipping distortion) I was getting so I used Tina sim to simulate the difference between the "idea" (ha) versions of each jfet. I swept the notch up and down and discovered that that (according to the sim) the j113 is about 50x more sensitive across the parallel 22k than the 2n5952. the LFO voltage swing reaching the fet gate is about 250-300mv peak to peak, but I can make the whole j113 sweep in about 50mv. So to compensate I changed each of the parallel resistors for 4.7k and the 47nf caps on the non inverting input to the filter sweep op amps were changed to 220nf (to keep the filter frequency in the same place).
I'm much happier with this sweep tone, although its still clipping a bit. I think its due to the PNP collector voltage being too high, so its not swinging in the center of its range. I'll drop the collector resistance further (at 10k compared to original 56k atm) and let you know how I get on.
Also the 10k in series with the zener isn't enough to properly reverse bias the zener, so I used a 1k to give it enough current. I'm using a psu not a battery so I'm not bothered. It still only draws 6mA ish anyway which is hardly a lot.
Steve

Wow thanks, that's some awesome detective work there. Out of curiosity, did you happen to note a 2n5952 mV sweep range for comparison to the 50ma in the j113(or is that the 50x sensitivity you refered to)? Also, do you have a vgsoff measurement on the jfets you are actually using in the circuit?

[StevieF]

I just edited my original post. I meant 5x not 50x. I was thinking of 50mv. lol
There was a model for the ideal 2N5952 in Tina so I used that. I found the voltage at its base 130ish Hz notch and adjusted the gate until the notch began to move and then used that as a starting point. I then gradually changed the voltage until the notch was up at 2khz and that was about 250-300mv, which is the swing voltage from the LFO that reaches the gates. I then just replaced the jfet with a j113 in the sim and did the same. It was about 50mv of difference on the gate to produce the same effect. So I reduced the R and C impedances to a 1/5 of what they were in the sim and it was pretty close. then I just translated that to the components on the board.

I don't have the vgsoff for the j113s I'm using no, but I used the op amp jfet matching circuit that is available, and they are all about 0.95v gs to produce a 10k resistance.