MXR Phase 90 with switch, volume and depth

[ghiekorg]

Hi Guys,
i wanna build an MXR phase 90 phaser.
Online i could find 2 PCBs:
- Effectslayout: http://effectslayouts.blogspot.com/2015/04/mxr-script-phase-90.html
- Tonepad: http://tonepad.com/project.asp?id=7

Tonepad's layout seems easier to mod.

I would like to make the following mods (i marked them on the image below):
1 - Feedback switch. I guess easily doable by adding the ON/ON switch where the 22k resistor is (on the tonepad layout) leaving one lug empty.
     Can i just cut the trace and add it where the red dot is?
2 - Volume knob. As i read from tonepad page "The resistor to change the level is the 150k which connects between the collector and base of the PNP transistor: 120k to reduce gain, 180k to increase it".
     Can I just substitute this with a A250k knob where the green dot is (i guess that's the 150k resistor)?
     I am just not 100% sure how to solder it, just use lug 1 and 2 leaving the 3rd free?
3 - Depth knob. I read here https://www.diystompboxes.com/smfforum/index.php?topic=86588.0 changing the 1M resistor for a 1M pot should control the depth (purple dot on the image, lower left corner).
     Is it the right resistor?
     Must it be a linear pot?



I am not sure it would still fil in a 1590B with all the mods, shall I go for a 125B?
If you have any suggestions i am happy to listen to them

Thanks a lot 

[Mark Hammer]

A toggle for feedback resistance and wide/narrow toggle for sweep width are sufficient.  I tend to include a pot for "offset", since small variations can change the character of the effect more than small variations of the other two functions do.

There is a 1M resistor, coming off the trimpot wiper, limiting current fed to the FET gates.  It works in tandem with the 3M9 current-limiting resistor coming from the LFO.  Altering that total current changes the drain-source resistance of the FETs.  Twenty years ago, RG Keen posted a re-draw on GEOFEX of a tricked-out Phase 90, that replaced the 1M fixed resistor with a 1M pot.  I have found it more useful to replace the 1M resistor with 510k-560k and a 500k pot.  This will allow for moving the phasing from deep and "gurgly" to high and swirly, and tones in between.  As well, since the current coming from the LFO sums with that coming from the trimpot, use of the 510-560k fixed resistor prevents the total current from pushing the FETs into non-linearity (i.e., where they can't change resistance any more).

While we are on the topic, the width of the sweep - ignoring where in the spectrum the sweep occurs - is altered by the current-limiting resistance of the LFO.  Some drawings show a 3M9 resistor and others show 3M3.  The lower value will provide a somewhat "wider" sweep.  If a person uses 3M9 as the stock value, and a toggle to place 22M in parallel with it, you end up with 3M3, providing narrow/wider sweep.  Again, part of the consideration is that the total current, coming from the LFO and the bias trimmer, still allows all 4 FETs to move their drain-source resistance around.

Finally, RG shows the pair of 150k mixing resistors that combine the wet and dry signals at the base of the output transistor replaced with 47K each, and a 200k linear pot to mix the relative balance of each.  Given the stated values, I can't see this as completely removing the dry OR the wet signal, but that doesn't make it useless.  Personally, I skip any blend/mix pot and just go directly for a "dry-lift" to get a vibrato effect, but I can see where others would be interested in shades of wet/dry mixes.

Since you want to put all of this in a 1590B, it CAN be done, using the Tonepad layout, but your placement of any pots or toggles will need to be strategic, as well as your choice of electrolytic caps, how "high" the FET legs are and whether one uses 1/4 or /8W resistors for those mounted vertically.  For me, the most "spatially conservative" arrangement in a 1590B is likely to be:
-  speed pot
-  2-position mini-toggle for more/less feedback
-  2-position mini-toggle for wider/narrower sweep
-  3-position on-off-on SPDT mini-toggle for highest/stock/higher offset (1M default bias resistance, with 1M or 2M2 added in parallel to achieve 500k or 688k)

Use of a 125B would allow for use of an offset pot and a vibrato/phase switch, and comfortable spacing of all controls/switches.

[ghiekorg]

First of all, Thank you Mark for taking the time to write in such detail.

I have found it more useful to replace the 1M resistor with 510k-560k and a 500k pot.

So you mean substituting the 1M resistor with a 510-560k and add a 500k pot in series?

If a person uses 3M9 as the stock value, and a toggle to place 22M in parallel with it, you end up with 3M3

So instead of a knob you would use a switch adding/removing the 22M resistor in order to have 3.9M or 3.3M. All clear.
As, as you said, they work together (this mod and the one before), is it necessary to have both mods or one is enough?

Finally, RG shows the pair of 150k mixing resistors that combine the wet and dry signals at the base of the output transistor replaced with 47K each, and a 200k linear pot to mix the relative balance of each.  Given the stated values, I can't see this as completely removing the dry OR the wet signal, but that doesn't make it useless.  Personally, I skip any blend/mix pot and just go directly for a "dry-lift" to get a vibrato effect, but I can see where others would be interested in shades of wet/dry mixes.

Here i am a bit ocnfused. Are we talking about the same 150k? Because i read it controls the volume, not the wet/dry signal. Once i have control over "depth" and speed i don't need a wet/dry i guess.
I just need a general volume to keep everything at unity (i saw generally the phase 90 has a volume boost) but then i guess it could be enough to remove the 150k resistor just before the OUT, use a 180-200k resistor between the collector and base to boost the signal and add a volume pot before OUT. Am i wrong?

-  2-position mini-toggle for more/less feedback
-  2-position mini-toggle for wider/narrower sweep
-  3-position on-off-on SPDT mini-toggle for highest/stock/higher offset (1M default bias resistance, with 1M or 2M2 added in parallel to achieve 500k or 688k)

I am getting a bit lost with all these mods to which mods are they refering?
the first ot the 22k, the second to the 3.3/3.9 M and the third to the 1M resistors?

About the enclusure i will check when home if the 125B would fit in my super tight pedalboard

Thank you

[Mark Hammer]

I'll go in order of your questions.

1) Yes, if you want variable offset, solder a 510-560k resistor to one of the lugs of a 500k pot, and run wires from the free end of the added resistor and one of the free lugs of the 500k pot to the pads where the 1M resistor used to be.  You now have a variable resistance that ranges from 510k (or 560k) to 1010 or 1060k.

2) Yes, adding in 22M with a switch gets you a wider sweep when the combined parallel resistance is dropped from 3M9 down to 3M3.  Script and Block issues seemed to be optimized for different speeds.  If the goal is a slower speed, then a wider sweep and more resonance/feedback is both tolerable and preferable.  If the goal is generally slower throbbing or even bubbly speeds, then narrower sweep and less resonance/feedback is preferred.

3) Admittedly, yes there are a LOT of 150k resistors in the circuit!  IN this case, a wet/dry mix/blend control would involve the 150k resistors shown in bright green, here.  The one on the left carries dry signal to the mixing node and the one on the right brings wet signal to the same point.  RG's suggested wet/dry blend would have each of those 150k replaced with 47k, with a "free" end left unconnected to that blue bar between them.  Their free ends go to the outside lugs of a 200k linear pot, and the wiper of that pot then goes to one of the pads on that blue bar.


4) I can understand the desire for some form of level balance between effect and bypass.  Personally, I haven't found much of a difference, but the simplest thing is to just stick a 50k pot (linear or log) on the output, with a 100k fixed resistor from the ground lug to the ground connection, instead of the stock 150k resistance.  It's mot like there is SO much differences in level that you'd want to attenuate much at all.

5) The described mods, in order:

• two-position feedback, use 27k as the default feedback resistor, and use a toggle to switch a 100-120k resistor in parallel with it for a higher-resonance tone
• two position sweep width, use a 3M9 "default" resistor on the board, and use a toggle to add 22M in parallel to achieve 3M3 for a wider sweep (note, resistances from the high 2megs to the low 4 megs will "work", but these values 3M3/3M9 are found in different issues of the P90)
• phasing-range/offset; yes the suggested 3-position toggle bumps the frequency range of the phasing upwards from the stock low-and-throaty sounds; lower resistances move the range upwards

Clearer?

[ghiekorg]

Yes now it's much clearer sorry sometimes some stuff gets too complicated for me and i can't follow properly yet
so...
1) Clear. Approved.
2) Also clear. Just thinking... what if i place a 3M resistor and add in series a 1M pot? that would give me all the positions between 3M and 4M. Would it work too? I just like pots more than switched
3/4) OK then i guess we were talking about 2 different things (as i understood from tonepad i was talking about the 150k resistor on top of the transistor, not below) i think i am fine, no need for a wet/dry pot. Then in case there is a big difference in volume i will just add a general  volume pot, but i guess i will see it when the pedal is finished.
5) Sorry, why using a switch to change between 27k and 22k instead of getting rid of it like suggested on tonepad?

Thanks a lot

[savethewhales]

This is nice! I am designing a pedal based on the P90 and I actually considered having volume, but after all I didn't took it far.

I have a feedback pot, a depth pot, speed pot and a mix/ammount pot.

Bear in mind that, before any questions, the LFO circuit I designed is different than the MXR one.

Any information you wanna know we can talk more!

[Mark Hammer]

Yes now it's much clearer sorry sometimes some stuff gets too complicated for me and i can't follow properly yet
so...
1) Clear. Approved.
2) Also clear. Just thinking... what if i place a 3M resistor and add in series a 1M pot? that would give me all the positions between 3M and 4M. Would it work too? I just like pots more than switched

It would "work", but you'll find yourself using the pot in the 7:00 and 5:00 positions, so why take up the panel space?

3/4) OK then i guess we were talking about 2 different things (as i understood from tonepad i was talking about the 150k resistor on top of the transistor, not below) i think i am fine, no need for a wet/dry pot. Then in case there is a big difference in volume i will just add a general  volume pot, but i guess i will see it when the pedal is finished.

There isn't much volume difference between effect and bypass.  I suspect you'll find the suggested 50k more than enough; plus, no accidentally turning the effect off completely and frantically thinking "Why isn't it working?!!!!".

5) Sorry, why using a switch to change between 27k and 22k instead of getting rid of it like suggested on tonepad?

There is a DC path from the output of the first op-amp, through to the output of the last phase-shift stage.  Though each phase-shift stage is nominally unity gain, unless all those 10k resistors are EXACTLY 10k - no more and no less - there is the risk of some modest gain over the 4 stages.  Switching "between" things often runs a risk of switch-popping as DC finds its path briefly interrupted by switching.  Creating different settings via parallel components, runs less of that risk.

Thanks a lot

My pleasure.  Let us know how you like the end result.

[ghiekorg]

It would "work", but you'll find yourself using the pot in the 7:00 and 5:00 positions, so why take up the panel space?

Switches are nice to see but they are so tall compared to the pots... i always have problems with them. the part inside the enclosure is 9mm for a pot and 13mm for the switch. I will think about that

There isn't much volume difference between effect and bypass.  I suspect you'll find the suggested 50k more than enough; plus, no accidentally turning the effect off completely and frantically thinking "Why isn't it working?!!!!".

hahaha I made a second version of this possible mod. I don't have to change the traces so i won't do it. IF then there is a difference in volume i will add it

Creating different settings via parallel components, runs less of that risk.

As the feedback mod is with/without the 22k resistor, it's just weird to me that a 27k resistor sound like a broken trace.  Or are we talking about 2 different things?

I just post here the modified layout to double check i understood everything properly.
I will also order the components. I couldn't find a 250k trimmer so i guess i will go for a 200k. Or better 500k?

@savethewhales: thanks how did you add a pot for the feedback?

[savethewhales]

@savethewhales: thanks how did you add a pot for the feedback?

feedback on the later versions of P90 is a 24kohm resistor (or 22, can't remember), so I added in series a pot of 500k that when it is on maximum resistance, the total resistance is 524k, and feedback is 0, when it is on minimum resistance, total resistance is 24 k, and the feedback is maximum.

So the value of the fixed resistance is your choice (mine was same value of the resistance which sets the filter frequency on the all pass = 24k ohm).

[Mark Hammer]

Remember that the feedback resistance is relative to the input resistance to the inverting pin, which is 10k.  I've experimented with lower-than-22k values of feedback resistance and the results below 18k are NOT pretty.  Personally, I'd be disinclined to go below 20k.  Once you go above 30-33k, the feedback signal is so small, compared to the other input signal to that stage, the feedback becomes largely inaudible.  So, my suggested values of 27k and 22k are for a mild feedback, and more of a stock feedback.  As for a 500k pot, I think you'd find that anything above the first 40k of resistance doesn't change very much at all.

Note that older 6x741 issue of the P90 placed 22k resistors in parallel with the JFETs, and used a 22k feedback resistor, while the 3xTL062 issue used a 24k feedback resistor, and 24k in parallel with the JFETs.  Technically, the one with 24k should sweep a little lower, though whether one can hear that is hard to say.  The larger feedback resistor, however, will dim the feedback sound a bit.

[ghiekorg]

Thank you guys.
I slept over it and i decided i wanna go crazy 
I will do a 6 knobs version of it: wet/dry, rate, gain, offset, feedback and sweep.
I updated the layout and i will now make a list of the components i need. Feel free to have a quick look at it and tell me if something is wrong.
Here is what i did:
- WET/DRY: removed the 150k resistors and placed 47k's. Removed the connection between them and Placed a 250K B pot with the wiper going to the transistor
- GAIN: removed the 150k above the tranistor and placed a 100k B pot (better A?) in series with a 100k resistor to give me a 100-200k range
- OFFSET: removed the 1M reistor and placed a 500k B pot in series with a 560k resistor to have 560-1060k range
- FEEDBACK: removed the 22k feedback resistor and added a 25K B pot in series with a 15k resistor to give me a 15-40k range (i won't use the whole range but i wanna hear what happens )
- SWEEP: removed the 3.9M resistor and placed a 1M B pot in series with a 3M resistor to have a range of 3-4M



Thank you 

[11-90-an]

sorry for OT, but just wanna make sure to prevent tons of debugging, crying, swearing ...

are the JFETs you intend to use matched? just wanna make sure... 

[ghiekorg]

sorry for OT, but just wanna make sure to prevent tons of debugging, crying, swearing ...

are the JFETs you intend to use matched? just wanna make sure... 

I actually wanted to ask about that... to make them matched is it enough to be the same thing (for example 4x 2N5952) or they have to be "specially" matching?

[11-90-an]

sorry for OT, but just wanna make sure to prevent tons of debugging, crying, swearing ...

are the JFETs you intend to use matched? just wanna make sure... 

I actually wanted to ask about that... to make them matched is it enough to be the same thing (for example 4x 2N5952) or they have to be "specially" matching?

there needs to be a "specific" matching for them... apparently each JFET is quite different, even with the same part names... the goal is to find 4 of the similar-est JFETs for your phaser.

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

https://www.diystompboxes.com/smfforum/index.php?topic=72583.0

[ghiekorg]

Oh. That's annoying. :/
i can find a quartett of 2N5457, not of 5952, but i am then not sure all the other components will work as expected... 

[11-90-an]

Oh. That's annoying. :/
i can find a quartett of 2N5457, not of 5952, but i am then not sure all the other components will work as expected... 

Try doing that JFET testing jig that RG made in the link I sent you. You just might get lucky... 

[ghiekorg]

i think i would rather go for the quartett it costs 8.5$ but i should work. i think it's too risky to buy lots of 5952 (they go for 1,25$ each) to just match 4. What will i do then with the rest of them?

[savethewhales]

i think i would rather go for the quartett it costs 8.5$ but i should work. i think it's too risky to buy lots of 5952 (they go for 1,25$ each) to just match 4. What will i do then with the rest of them?

Good decision! I did the exact same (I bought from Musikding) and it saved me a whole lot of stress.

Make sure to make not only one test (search for the J C Mallet JFET matching test too, it was the closest to reality if I'm not mistaken). Take notice that, if bought from musikding, they come with the Vpinch off written. Surprisingly or not, the Vpinch off written misses by around 0.15 V (in my case negatively).

Now, if you choose to not do a matched set, the phaser should work honestly, but it's a pain to make it work properly. The rds equivalent resistance of each JFET will be "maybe very" different and that will mess up your calculations all the way. You'd have to do it auditively or seeing in the osciloscope.

Oh and I think it's a good idea to do a Offset trimpot, that (which if I understood), will make the sweep happen more lower on the spectrum on higher depending on the current value of the pot, right? You actually gave me a good idea but I already have mine soldered AND have a box for it with holes in it.

Any questions, feel free to do!

[11-90-an]

i think it's too risky to buy lots of 5952 (they go for 1,25$ each) to just match 4. What will i do then with the rest of them?

hundred more phasers... 

[ghiekorg]

@savethewhales:
Yes i bought them there too, even though i don't like their service... i just buy there whatever i can't find on UKelectrics. I also bought several diodes and a bigger trimmpots, just in case...
I hope the mods i made are correct now and i ordered the right components
I also made a modified version of the printable layout, in case someone wants it (based on the image above)

@11-90-an:
hahahha yes haha It's so weird. I never even liked phasers before some weeks ago. Atm my less favourite modulation effect is flanger. Maybe i will end up doing one after this but i should also start to save some money i guess. I mean, compared to the commercial units i saved hundreds of $ still