General Guitar Gadgets - Phase 90 Clone issue

[Rob Strand]

I thought of another possibility, maybe your trimpot is actually way off.   Try to find another completely different trimpot setting that sounds good.

[bettsaj]

As a matter of interest I've bought another set of Jfets from the same supplier... Likely from the same batch. I could measure those to get the vp measurements. They might be in the same ball park. Actually they have measurements on the bag.

Fairchild 2N5457 Batch 1922C

Vgs (10M) -0.496 V
Vgs (10k) -0.152 V
Idss 1.80 mA

Vgs (10M) -0.496 V
Vgs (10k) -0.151 V
Idss 1.80 mA

Vgs (10M) -0.496 V
Vgs (10k) -0.152 V
Idss 1.80 mA

Vgs (10M) -0.495 V
Vgs (10k) -0.150 V
Idss 1.77 mA


We can assume that the Jfets in the pedal are in the same ball park as they are from the same batch. Unfortunately I don't have the packaging the original one's came in.

[bettsaj]

I thought of another possibility, maybe your trimpot is actually way off.   Try to find another completely different trimpot setting that sounds good.

The trimmer that is in it has phasing at top dead centre... Either way of that there is no phasing, If i changed it to another value maybe?

[Rob Strand]

As a matter of interest I've bought another set of Jfets from the same supplier... Likely from the same batch. I could measure those to get the vp measurements. They might be in the same ball park. Actually they have measurements on the bag.

Awesome!

I don't know what JFET measurement jig they used (I suspect maybe Run-off-groove, ROG) but it doesn't matter much as the 10M reading should be very close to the Vgs_off for the JFET.

So that does confirm my suspicion that the Vgs_off values are low.   The 2N5952's are normally around 2.25V but your values are around 0.5V.  FYI: You might see values around 1.6V posted.  These are done on RG's (GeoFX) JFET matcher which gives lower readings.

he trimmer that is in it has phasing at top dead centre... Either way of that there is no phasing, If i changed it to another value maybe?

No don't change it.  It is fine as is.

To start, try changing R17 (1M) to 150k.    You might even have to go down to 47k!

[bettsaj]

The 2N5952's are normally around 2.25V but your values are around 0.5V.  FYI: You might see values around 1.6V posted.  These are done on RG's (GeoFX) JFET matcher which gives lower readings.

To start, try changing R17 (1M) to 150k.    You might even have to go down to 47k!

Mine are 2N5457... make any difference?

I'll change the 1M resistor as soon as I get home..... I know I have 150k.....

[bettsaj]

Hi rob,

Here's the latest... i changed R17 from 1M to 150k, it made a big difference, but still not perfect. So i swapped it for a 100k, and this was the result

Do you think this is about where it should be?

[Rob Strand]

Mine are 2N5457... make any difference?

Yes they are different.  The VP parameter is lower.   In your case the VP values are low even for 2N5457's.   There seems to be a pattern of low VP values these days.  One way to look at it is the circuit was designed for  2N5952's so you should use JFETs close to that.  The reality is these days JFETs are harder to get.  So what needs to be done is to tweak the part values in the circuit so it operates very close to the original.  That's pretty much what we did.   From a technical perspective higher VP JFETs keep distortion down.

Here's the latest... i changed R17 from 1M to 150k, it made a big difference, but still not perfect. So i swapped it for a 100k, and this was the result. 

That's a vast improvement.    The video is was really helpful.

Do you think this is about where it should be?

It's pretty darn close certainly in the range of variations in normal units.   

My feeling is the depth of the sweep could be a little more (but it could simply be the things you played biasing my opinion).  Maybe worth trying 120k?  Perhaps try adding a 22k in series with the 100k you have in there. You should tweak the trimpot in each case.  When I analyse the audio it seems that way too but I can't see things very accurately from that clip.  (BTW, the trimpot is harder to adjust accurately for the lower VP JFETs.  If I were building these things in production I'd probably makes some circuit tweaks around the trimpot to make life easier.)

There's still something wrong.  Your slowest sweep seems like about 1.5 seconds when it should be in the 7sec region.  Is your C7 value 15uF or 4.7uF?.

[Rob Strand]

Some more details.

A side effect of reducing R17 is C10 should be increase from 47nF in order to match the behaviour of the original.  At high sweep speeds the original reduces the sweep range and smooths the sweep.   The effect is noticeable for speeds faster than about 2Hz.   

With R17 at 100k you probably need to increase C10 to about 220nF.   With the cap in the sweep is reduced and you might find you like a larger R17 value.

The very early scripts used a 10nF cap for C10 instead of 50n and that tends to keep the sweep even over the speed range.   So perhaps you should increase C10 to at least 100nF.

[bettsaj]

My feeling is the depth of the sweep could be a little more (but it could simply be the things you played biasing my opinion).  Maybe worth trying 120k?  Perhaps try adding a 22k in series with the 100k you have in there. You should tweak the trimpot in each case.  When I analyse the audio it seems that way too but I can't see things very accurately from that clip.  (BTW, the trimpot is harder to adjust accurately for the lower VP JFETs.  If I were building these things in production I'd probably makes some circuit tweaks around the trimpot to make life easier.)

I'll swap it for a 120k if I have one.....

There's still something wrong.  Your slowest sweep seems like about 1.5 seconds when it should be in the 7sec region.  Is your C7 value 15uF or 4.7uF?.

My cap is 15uf

A side effect of reducing R17 is C10 should be increase from 47nF in order to match the behaviour of the original.  At high sweep speeds the original reduces the sweep range and smooths the sweep.   The effect is noticeable for speeds faster than about 2Hz.

Ok, I'll look at increasing that too

With R17 at 100k you probably need to increase C10 to about 220nF.   With the cap in the sweep is reduced and you might find you like a larger R17 value.

Good call..... I'll change it

The very early scripts used a 10nF cap for C10 instead of 50n and that tends to keep the sweep even over the speed range.   So perhaps you should increase C10 to at least 100nF.

As a matter of interest, i've just started to build another phase 90 clone, this time from Fuzzdog pedal Parts in the UK. He uses the same FET's as me as they're readily available so I've started to compare the GGG schematic and the Fuzzdog schematic and unless i'm seeing things I think the values on some of the components are different, likely to compensate for the FET's having a low vp. You'll notice that the component numbers are different between the two schematics, however it's not hard to see that on the GGG schematic C10 is a polarised cap, and on the fuzzdog schematic C10 is C7 and is a non polarised cap at 47n

I've included the Fuzzdog schematic and BoM below


R1 10K
R2 470K
R3 150K
R4 150K
R5 56K
R6 150K
R7 10K
R8 10K
R9 22K
R10 10K
R11 10K
R12 22K
R13 10K
R14 10K
R15 22K
R16 10K
R17 10K
R18 22K
R19 22K*
R20 150K
R21 3M9
R22 150K
R23 4K7
R24 470K
R25 150K
R26 47R
R27 10K
R28 1M
R29 CLR**
R30 1M
R31 1K
R32 390R
C1 10n
C2 47n
C3 47n
C4 47n
C5 47n
C6 47n
C7 47n
C8 10n
C9 15u elec
C10 100u elec
C11 100n
C12 22u
C13 10u
Q1 2N5087***
Q2-4 Matched FETs
D1 5.1v zener
D2 1N4001
D3 LED
IC1 TL072
IC2 TL074
SPEED 500KC
T2‡ 20KB/22K TRIM
T1 470K TRIM

[Rob Strand]

Quote

    There's still something wrong.  Your slowest sweep seems like about 1.5 seconds when it should be in the 7sec region.  Is your C7 value 15uF or 4.7uF?.

My cap is 15uf

Hmmm,  1.5sec is a long way off from 7sec.  There's something wrong around the LFO.   Unless there's something wrong with that cap!     In your vid there was something shorted for while maybe that was the cause.  You should be able to check the slowest speed with your watch.   Another possibility is the pot value 500k, yes?  Maybe measure it with you meter.

I'll swap it for a 120k if I have one.....
Ok, I'll look at increasing that too
Good call..... I'll change it

I think it's worth tinkering around a bit.

As a matter of interest, i've just started to build another phase 90 clone, this time from Fuzzdog pedal Parts in the UK.

I went over the schematic and I could only see changes related to the feedback (known as Script vs Block differences).   I also downloaded their build doc.  No changes related to the JFETs on the Fuzzdog.   The ggg schematic shows the feedback for the Block but the built unit is for the Script, as it leaves out the feedback resistor.   What could be the case is the 2N5457's Fuzzdog are using have higher VPs; they are usually 1V or 1.2V.    Your VP value of 0.5V is very low even for 2N5457, in fact it's on the edge of the spec, something you never really see on "real" 2N5457's.

[bettsaj]

Hmmm,  1.5sec is a long way off from 7sec.  There's something wrong around the LFO.   Unless there's something wrong with that cap!     In your vid there was something shorted for while maybe that was the cause.  You should be able to check the slowest speed with your watch.   Another possibility is the pot value 500k, yes?  Maybe measure it with you meter.

I'm using a 100k linear pot..... I did have a 500k rev log in it at the beginning but swapped it as it failed and was shorting out internally. The sweep was intermittent and when I isolated the pot, discovered it was the pot that was failing.

I'm also going to be cheeky and email Lee Callaghan at Fuzzdog to find out what VP values his FETs are. His FETs are all tested by him on a working phaser test rig so he knows when he supplies them they will work on his circuit. Unfortunately he doesn't sell the FETs separately, only with a full kit.

[Rob Strand]

'm using a 100k linear pot..... I did have a 500k rev log in it at the beginning but swapped it as it failed and was shorting out internally. The sweep was intermittent and when I isolated the pot, discovered it was the pot that was failing.

Ah, that makes sense now.

I'm also going to be cheeky and email Lee Callaghan at Fuzzdog to find out what VP values his FETs are. His FETs are all tested by him on a working phaser test rig so he knows when he supplies them they will work on his circuit. Unfortunately he doesn't sell the FETs separately, only with a full kit.

It can't hurt to see what JFETs he is getting.

Here's a stack of 2N5457 measurements.   Note these are *real* ones, before JFETs started to go bung.
The measurements are at low currents so they are real VP values.
https://viva-analog.com/characterizing-and-matching-2n5457-jfet-transistors/

The 2n5457s I used to get here were mostly around 1.2V to 1.5V.  (I think I said 1V to 1.2V before, sorry).

[bettsaj]

It can't hurt to see what JFETs he is getting

So Lee has got back to me and said basically what you've said.... He normally sorts them and matches them to within 1.2 to 1.6. Anything lower normally goes back in the bag. He's also said that he can't remember ever measuring anything that low.

I've also now emailed the supplier so see if he has any that measure higher than 0.5 or 1.0v

[Rob Strand]

So Lee has got back to me and said basically what you've said.... He normally sorts them and matches them to within 1.2 to 1.6. Anything lower normally goes back in the bag. He's also said that he can't remember ever measuring anything that low.

Those values look more normal for 2N5457s to me.

I've also now emailed the supplier so see if he has any that measure higher than 0.5 or 1.0v

Actually, he might have a record of the VP values for the JFETs in your unit.   You can only ask.  Your pic shows 2N5457 with batch code BC27.

[bettsaj]

So Lee has got back to me and said basically what you've said.... He normally sorts them and matches them to within 1.2 to 1.6. Anything lower normally goes back in the bag. He's also said that he can't remember ever measuring anything that low.

Those values look more normal for 2N5457s to me.

I've also now emailed the supplier so see if he has any that measure higher than 0.5 or 1.0v

Actually, he might have a record of the VP values for the JFETs in your unit.   You can only ask.  Your pic shows 2N5457 with batch code BC27.

I've just checked his website, and it looks like he uses this rig to test Jfets...... He even sells kits of this rig http://stompville.co.uk/?p=601

Here's the page where he sells them http://www.stompville.co.uk/shop/2-2n5457-matched-quad.html

[bettsaj]

Just also ordered a set of 4 matched jfets from GGG

[bettsaj]

The latest...

R17 is still at 100k

I've changed C10 to 220n and this is the end result

[Rob Strand]

I've just checked his website, and it looks like he uses this rig to test Jfets...... He even sells kits of this rig http://stompville.co.uk/?p=601

OK so it's sup'd-up version of RG's tester with the option of using a 10M resistor (using the multimeter impedance).

Here's the page where he sells them http://www.stompville.co.uk/shop/2-2n5457-matched-quad.html

OK.  He must update the data for each set he sells or something since the JFETs on offer now have a different batch code to the JFETs used in your phaser.   Yours have batch code Fairchild BC27.

The latest...

R17 is still at 100k

I've changed C10 to 220n and this is the end result

You might have misunderstood the motivation for changing C10.   The primary reason is to make the sweep on the modded unit (with R17 around 100k) behave like the original unit (with R17 1M).   C10 makes the sweep smoother at higher speeds (and avoids the risk of weird pulsations at high speeds as well).   It can't change the sweep speed.

If you really wanted to you could tweak the sweep circuit to get a 7 second sweep with your 100k pot but that would be a separate set of mods which addresses only the sweep speed.

[bettsaj]

So as I said in the video... changing C10 has smoothed the phase out, and achieved it's intention.

How would I go about changing the sweep length? I intend to change the pot back to a 500k rev log as soon as I can as I need the 100k for something else.

[Rob Strand]

So as I said in the video... changing C10 has smoothed the phase out, and achieved it's intention.

Something you said at the end made me think you expected to fix it.  No worries, as long as you know.

How would I go about changing the sweep length? I intend to change the pot back to a 500k rev log as soon as I can as I need the 100k for something else.

Basically you have reduced the resistance by 1/5 and that means you need to increase the 15uF cap by a factor of 5, so 5 x 15uF = 75uF.   So to do that you will need 47uF in parallel with 33uF (or 25uF).   The other thing is if we make the cap bigger we need to make the 4.7k resistor (R20) five times smaller, so about 1k; that ensures the maximum speed is correct.     It's a bit of a pain but OK for a temporary fix.

You could also do a "nearly there" fix by changing the 15uF cap to 47uF and R20 to 1.5k.   That will give you upto about 4.5 seconds, which is a lot better than 1.5 seconds.