PH-1R Schematic Questions- Advices About JFETs & Other Improvements (Data added)

[abakuzam]

Hello , hope you all doing well. I have been wanting to build this pedal for quite a long time. But i have some questions about it, i need some clarification. I read everything i can find about it in this forum.

1) As far as i understood Q1, R1, R2, R3, C1, C2 makes an input buffer? I'm not sure i want to keep it, but i wanted to make sure which components to remove to get rid of it.

2) Besides all the mambo jambo at the right bottom corner, to remove the bypass sistem i should remove Q6, D4, R55, R30 and C27?

3) D3 is 5.1V Zener for bias, can i just use a resistor to have same voltage dividing?

4) I have quite a lot fake JFETs ( J201 & 2N5457 ), i'm planning to match them and use them. Looking at the datasheets, 2N5457 looks closer to 2SK30A-GR when it comes to Idss & Vgs(off), 2N5457 has higher Vgs(off) 6v max compared to 5v max on 2SK30A-GR, to cope with that , and make sure JFETs will turn off, should i change the D3 (higher? ) or playing around with trimmer VR4 is just enough? Will i need to adjust 100K - 10nF networks according to Rds values? Is it that important?

I had another topic about fake JFETs i have, maybe useful for other people : https://www.diystompboxes.com/smfforum/index.php?topic=121553.40

5) Addind LFO led ( maybe 2 for ramp up - down effect? ) , i haven't done my research on this one, i'm not sure it would effect the lfo ticking, maybe i can try transistor based one to not disturb the lfo much. This is not my priority.

6) My goals are:

- Redraw the schematic without bypass system and add improvements to the schematic. With matching component names to the original schematic.
- Use TL074 and TL072 for 6 Op-Amps in total instead of 3 Dual Op-Amp ICs. I don't think it would be a difference in sound much, i'm looking at it layout point of view.
- Adding more filter on main power supply
- Changing all electrolytic capacitors =< 1uF to film ones ( i think R.G did this on this schematic.)
- Decoupling the TL022 from power supply . ( Maybe other IC's too? i don't think it would hurt)
- Finally putting it on a single sided layout ( fingers crossed ) and of course share it on here 


6) Is there any improvements i can make on this schematic? Or other suggestions?

Here is the schematic and the service manual link : https://elektrotanya.com/boss_ph-1r_phaser_pedal.pdf/download.html

[Rob Strand]

1) As far as i understood Q1, R1, R2, R3, C1, C2 makes an input buffer?

Yes.
But you will need some form of buffer because the following circuit will load down the guitar pickup.

2) Q6, D4, R55, R30 and C27?

Yes.  Once Q6 is removed anything before D4 has no effect.

3) D3 is 5.1V Zener for bias, can i just use a resistor to have same voltage dividing?

No.  You will find it will go out of adjustment when the supply voltage changes.

should i change the D3 (higher? )

No leave.

R43 can be adjusted so the sweep matches the VP values of the JFETs.

Watchout using typical and maximums in the datasheet.  IIRC the 2SK30A-GR ends up with VP's around 3V typical.  Whereas the 2N5457's are around 1.5V to 2V.  A lot of JFETs on the web have VP's that don't line up with datasheets or JFETs from 10 years ago.  There's also a trend for VP to be low nowadays.

Will i need to adjust 100K - 10nF networks according to Rds values? Is it that important?

Yfs0 is the important parameter, or rds0 = 1/Yfs0.   IIRC the 2SK30A-GR's have rds0's higher than  2N5457's so that means the 100k is higher and the 10nF is lower.  Common values are 22k to 33k  for the resistors and 50nF for the caps.    If it's important to keep the old sound then you would need to measure everything before you change it then tweak the values to match it after.

Addind LFO led ( maybe 2 for ramp up - down effect? )

I posted one a few months back using a MOSFET but for the LFO on the Boss unit you could easily use a transistor.  (I'll try to find the thread(s).)   Straight off the opamp with some diode drops might also work.

Is there any improvements i can make on this schematic?

Replace R8 with a short and remove R9.  Then increase R32 from 22k to 47k.
That will help decrease the noise but increases the risk of overload.


For all your mods, you have to ask yourself what performance changes are you trying to achieve?

[abakuzam]

R43 can be adjusted so the sweep matches the VP values of the JFETs.

Watchout using typical and maximums in the datasheet.  IIRC the 2SK30A-GR ends up with VP's around 3V typical.  Whereas the 2N5457's are around 1.5V to 2V.  A lot of JFETs on the web have VP's that don't line up with datasheets or JFETs from 10 years ago.  There's also a trend for VP to be low nowadays.

Yfs0 is the important parameter, or rds0 = 1/Yfs0.   IIRC the 2SK30A-GR's have rds0's higher than  2N5457's so that means the 100k is higher and the 10nF is lower.  Common values are 22k to 33k  for the resistors and 50nF for the caps.    If it's important to keep the old sound then you would need to measure everything before you change it then tweak the values to match it after.

My main goal is catching the old sound with the jfets i have, ( fake j201s, 5457s, original j201s and 5457s, a few dozen 4393s but i'm not sure worth using them ) and match the circuit accordinly, so i will measure and calculate all my jfets. Hopefully use them on this circuit.

Replace R8 with a short and remove R9.  Then increase R32 from 22k to 47k.
That will help decrease the noise but increases the risk of overload.

After i make the layout, i can try this. I'm not sure how it would decrease the noise, less gain on opamps? 

For all your mods, you have to ask yourself what performance changes are you trying to achieve?

Main focus for changes are just noise concerns and maybe an indication for phasing effect.

Thank you. I will keep this thread updated.

[Rob Strand]

My main goal is catching the old sound with the jfets i have, ( fake j201s, 5457s, original j201s and 5457s, a few dozen 4393s but i'm not sure worth using them ) and match the circuit accordinly, so i will measure and calculate all my jfets. Hopefully use them on this circuit.

If you just built to circuit as is it would probably work but IMHO it's probably not faithful to the original in performance.

The common JFETs had rds0 = 200 to 250 ohms; here rsd0 = 1/yfs0 and yfs0 = 2*Idss / VP.  Whereas the 2SK30A-GR were more like 350 ohms or so.   So to help coax everything into to right zone you probably want to scale the resistances down by a factor of 1.5 and the caps up by 1.5.   So the 10nF caps become 15nF and the 100k's become 68k.

It's debatable whether the other parts around the JFETs should be scaled.  I'd probably leave those as is.

When you measure your JFETs you will be able to work out rds0 for your JFETs and choose a better scaling factor.    When you measure VP you need to measure at low currents.   Some of the matching circuits produce a number for "VP" which is fine for matching but the measure voltage isn't really VP, it's also always somewhat smaller in magnitude.

After i make the layout, i can try this. I'm not sure how it would decrease the noise, less gain on opamps?

The two resistors form a divider which reduces the signal level to the opamps so the opamp noise is now louder relative to the signal.   If you pull the divider arguably to match the existing circuit you should leave the top resistor in but use half the value.

Main focus for changes are just noise concerns and maybe an indication for phasing effect.

Removing the divide will help reduce noise for sure.   Beyond that you can use pre-emphasis and de-emphasis.  The same trick used on chorus and flanger units.

Often the noise reduction methods work without any overload issues on guitar.  For keyboards sometimes you can get distortion.  However in this case signal is strong and the signal to noise is inherently better.   The issue is if you are Boss and have to make a pedal for guitar *and* keyboard you have to pick a scheme which finds a delicate balance between overload and noise.

[abakuzam]

I added the redrawn schematic, didn't have much time to check everything, but it should be okay, i did some changes mentioned in the topic. Component names matches with the original layout, other components ( like R100, C100 etc. added by me)

If you just built to circuit as is it would probably work but IMHO it's probably not faithful to the original in performance.

The common JFETs had rds0 = 200 to 250 ohms; here rsd0 = 1/yfs0 and yfs0 = 2*Idss / VP.  Whereas the 2SK30A-GR were more like 350 ohms or so.   So to help coax everything into to right zone you probably want to scale the resistances down by a factor of 1.5 and the caps up by 1.5.   So the 10nF caps become 15nF and the 100k's become 68k.

It's debatable whether the other parts around the JFETs should be scaled.  I'd probably leave those as is.

When you measure your JFETs you will be able to work out rds0 for your JFETs and choose a better scaling factor.    When you measure VP you need to measure at low currents.   Some of the matching circuits produce a number for "VP" which is fine for matching but the measure voltage isn't really VP, it's also always somewhat smaller in magnitude.

I've added my all measurements - couldn't add excel files - all my measurements and general average just for the info for other people. My BF245A & C looks legit according to datasheets.

It seems i will need that scaling factor for 100k's and 10nF's as you mentioned.

Allthough my multimeter has 10M input empedance, should i re-measure Vp for Rds(10M)? I did it with 10k

Overall according to my measurements , i can get some transistor matches ( within %2-3 differance ) to use in this circuit.  Progress will be updated.

Removing the divide will help reduce noise for sure.   Beyond that you can use pre-emphasis and de-emphasis.  The same trick used on chorus and flanger units.

Often the noise reduction methods work without any overload issues on guitar.  For keyboards sometimes you can get distortion.  However in this case signal is strong and the signal to noise is inherently better.   The issue is if you are Boss and have to make a pedal for guitar *and* keyboard you have to pick a scheme which finds a delicate balance between overload and noise.

I'm close to layout, i'll route as the original schematic. then play around with those values.

I need to research that emphasis topic and come back with more questions haha

Thank you for really informative replies !

[Rob Strand]

Allthough my multimeter has 10M input empedance, should i re-measure Vp for Rds(10M)? I did it with 10k

Overall according to my measurements , i can get some transistor matches ( within %2-3 differance ) to use in this circuit.  Progress will be updated.

For matching you will only get very minor improvement by changing to 10MEG.

However, the Vgs(10k) values you get when you use 10k does not give a measurement of VP.
The magnitude of Vgs(10k) is always less than Vgs(10MEG) because the test current is higher with 10k.
So when you calculate rds0 using Vgs(10k) your rds0 value will be low.

What test circuit are you using?

For RG's circuit with 10k, you can calculated VP from the Vgs(10k)  measurement,

VP = Vgs(10k)  / (1- sqrt(ID/IDSS))

ID = 4.5V / 10k  = 0.45mA  ; test current
IDSS must be measured.

So take your first J201,
Vgs(10k) = 0.728V
Idss = 1.714mA

VP = 0.728 / (1-sqrt(0.45/1.714)) = 1.49V

However to me that looks a bit high so I'd probably remeasure a few parts using 10M and see if that method of estimating VP from Vgs(10k) is any good.

On the other side of the coin when you calculate rds0 with the estimated VP is doesn't look too bad.

[abakuzam]

However, the Vgs(10k) values you get when you use 10k does not give a measurement of VP.

I thought Vgs(off) = Vp, I guess I was wrong.
That's what I used for yfs0 calculation thus rds0.
I need to read and learn more

What test circuit are you using?

I used rullywow's layout and schematic(added) for Vgs(off).

Measured the voltage drop across 100Ohm resistor then calculated the current for Idss

I should try it with 10M resistor also , to get better results.

Thank you.

[Rob Strand]

I thought Vgs(off) = Vp, I guess I was wrong.

In some testers, the measured Vgs voltage is just a number to help matching.

In that tester, the measured Vgs voltage aren't really Vgs(off) or Vp.  The meaning of the
number is Vgs at the test current.  The test current is 4.5V/10k = 450uA.  At that current
the Vgs measurement cannot be interpreted as Vgs(off) or Vp.
In fact the measurement produced is *way* off Vp.

If you change the resistor to 10M  then the test current is 4.5V / 10M = 0.45uA.  With that
low current the Vgs measurement is a reasonable approximation to Vp.   Off hand it's like 1%
off or less, some small error anyway.

That's what I used for yfs0 calculation thus rds0.
I need to read and learn more

So those numbers will be off by a large factor.

I used rullywow's layout and schematic(added) for Vgs(off).

OK thanks.  That's pretty much RG's tester.

FYI the formula I gave tries to estimate the real VP based on the measured IDSS and measured Vgs(10k).
It's definitely a better number than Vgs(10k), especially for calculating rds0.   The number produced
is often different to Vgs measurement using 10M because of measurement errors and assumptions about the JFET following a square law.

[abakuzam]

If you change the resistor to 10M  then the test current is 4.5V / 10M = 0.45uA.  With that
low current the Vgs measurement is a reasonable approximation to Vp.   Off hand it's like 1%
off or less, some small error anyway.

I think i can live with %1 error in my situation, i'll try to measure all tomorrow with 10M and update my results.

FYI the formula I gave tries to estimate the real VP based on the measured IDSS and measured Vgs(10k).
It's definitely a better number than Vgs(10k), especially for calculating rds0.   The number produced
is often different to Vgs measurement using 10M because of measurement errors and assumptions about the JFET following a square law.

So that means, with better accuracy than Vgs(10k) we would have better results on rds0 , thus changing 100k's and 10nf's would be more realistic to get the actualy sound.

Not mainly related about the topic but i have found a local source for to92 2SK30A-GR's today, they didn't have it at the time, i think they will bring more (500pcs), i might also pick some and test them too, i'm not sure they are fake or not. They also had to92 2SK117-GR's but they looked a bit sketchy to me because of the lettering, i had original 117GR's and SK30AGR's before i was sure about. I found a pic for 117GR's i saw today. And i didn't want to pay 1.5$ each

[Rob Strand]

I think i can live with %1 error in my situation, i'll try to measure all tomorrow with 10M and update my results.

Yes, it's not enough to worry about.  Even a 1M test resistor only has a small error.

So that means, with better accuracy than Vgs(10k) we would have better results on rds0 , .

Yes, those Vgs(10k) are not a good indicator.  You can add another column to your spreadsheet which calculates VP from Vgs(10k) and IDSS (and ID = 450uA) using the formula above, then change your yfs0 and rds0 calculations to use that new VP.

thus changing 100k's and 10nf's would be more realistic to get the actualy sound.

If you want to use different JFETs then, yes, scaling the design gives you the best chances of getting closer to the original.    You can go to far though.  For example if you use JFETs with VP=0.5V then you will be able to get it to work but you might see a little more distortion.   If you use VP's above 1V or 1.5V then it would be close enough. 

I've found some info specifically for 2SK30A-GR's.   As you can see the three sets of numbers have things in common and things that are different.    The general trend is the rds0 is a little higher than more common JFETs (except the J201's which have high rds0).

Also note there is variation across 2SK30A-GR's so the best you can do is target the middle.   Beyond that might be overthinking things a bit.   Once you in the general zone, the bias trimpot helps tweak any small differences.   You just don't want to rely on the bias trimpot to try to make completely different JFETs act the same.

1) Datasheet
Idss = 4.3 mA
VP= 2.15 V
Yfs0 = 4000 uS
rds0 = 250 ohm

2) Measurements (averages)
Idss = 4.4 mA
VP = 3.03 V    ;estimated from other params
yfs0 = 2900 uS 
rds0 = 345 ohm

3) Measurements (averages)
IDSS = 3.77 mA
VP = 2.14 V     ; from curve fit
yfs0 = 3514 uS
rds0  = 284 ohm 

Not mainly related about the topic but i have found a local source for to92 2SK30A-GR's today, they didn't have it at the time, i think they will bring more (500pcs), i might also pick some and test them too, i'm not sure they are fake or not. They also had to92 2SK117-GR's but they looked a bit sketchy to me because of the lettering, i had original 117GR's and SK30AGR's before i was sure about. I found a pic for 117GR's i saw today. And i didn't want to pay 1.5$ each

I guess it's upto you if yo want to buy more.  From your tables of data you probably have enough JFETs to choose something in the right ball-park and select some matched JFETs as well.

[PRR]

> In that tester, the measured Vgs voltage aren't really Vgs(off) or Vp.
> The test current is 4.5V/10k = 450uA. 
> If you change the resistor to 10M  then the test current is 4.5V / 10M = 0.45uA.

Half a mA (450u) isn't "cut off" for a few-mA JFET.

FWIW, the specsheet for J201 specifies cutoff as 10nA.

10nA is awful hard to measure (wants a 200Meg measurement).

Last week I asked SPICE to plot its data for an old JFET. It happens the internal notion of Vp is 3.0V. I plotted log so I could see current all the way down.


The 450uA point is way off of 3.0V.

The 0.45uA point is like 2.97V, pretty close to the "ideal" answer.

The 10nA point is insignificantly different from the 0.45uA point.