BOSS CE-3 Signal Allowance HEPL ME!)

[Rob Strand]

I asked one owner, he said that he, too, has an interrupt signal ... Maybe this is a mistake in the early pedals?

Interesting someone else has the problem. 

I don't know if there is a problem with the board.  The Boss schematic and the Boss documents I have for CE3 are for the "First Version".   I do not know if there exists Boss documents for a "Second Version" and if those later documents have a fix/mod for the issues.

Sometimes with difficult problems like this the schematic is correct but the PCB does not match the schematic.   I have looked at the PCB, not everything, and it does seem to follow the schematic.  I think for now we have to assume the board is OK but there is a fault.

I thought, suddenly they had a series of defective parts. And so I tried to solder instead of Q4 2SK117, and it gave nothing. (Of course I took into account its polarity)
Ok, but I still don't understand how to connect with what.
I did as you advised, everything worked, but ... The problem remained !!!!

Even though the problem still remains the result is good!   That means the original parts and circuit on the PCB for Q4, R61, C31, D3 are probably working correctly.   It also means the problem is probably not around those parts.

OK so replace Q4 and remove all the added parts so the unit is back to normal.

I'll try to summarise what we know:
- The problem is:  when switching from Bypass to Effect in Mode II the (delayed) signal is muted for 0.5 second.
- The problem is only in Mode II.    Mode I works correctly.
- Changing the RC time constant for the JFET gates does not change the problem (reply #30)
- Changing the gate voltages to the JFET does not change the problem (reply #30 and #32).
- The delay was removed when the bias trimpot was set to an extreme setting; but the delay signal is lost.
- Replacing the JFET switch Q4 and the RC network did not change the problem. 

So the main points here are:
-  Mode I behaves different to Mode II.
- Q4 and the switching circuit appears to be working.

So what is different between Mode I and Mode II?   One difference is Q3.  So maybe there is a problem with Q3 or IC1b (pins 5, 6, 7).   

So the next step would be,
- measure the voltages on IC1b (pins 5, 6, 7) and IC1a (pins 1,2 3) in both Mode I and Mode II
- then try removing Q3 and see if the 0.5 sec mute in Mode II is affected . (you will get no sound in bypass)
- then try putting a wire across Q3 drain and source and see if the 0.5 sec mute in Mode II is affected.

[midiez1997]

I measured voltages, and they turned out to be the same in both modes.
The most amazing thing is that after removing K3, the delay disappeared!))))
With the wire the same result !!!
We almost won this problem!)

[midiez1997]

I tried to solder 2sk117 ... the problem remained
I forgot to say that, in addition to the delay, cotton is perceived as if it were heard. it's very quiet and barely noticeable, but it sounds like a sound True-bypass.

[Rob Strand]

I measured voltages, and they turned out to be the same in both modes.
The most amazing thing is that after removing K3, the delay disappeared!))))
With the wire the same result !!!
We almost won this problem!)

Yes, we are very close.   The problem occurs when Q3 *changes* between off and on, which only happens in Mode II.  However, it is not obvious what the problem is.   

It would be useful if you could report the voltages on all pins of IC1.  Maybe report the DC voltage on the source of Q1, where it connects to R3.

I think there is a DC problem around the Q3 switch.  Notice Q3 is DC coupled to IC1 pin7 (ie. no cap).   So that's why it would be useful to report the voltages.  When Q3 changes state the circuit must find a new DC equilibrium.  Somehow the signal gets blocked when this is happening.

Normally all the opamps and JFETs all sit at the bias voltage.  The bias voltage is the voltage across C20. 

If DC output of IC1 pin 7 not the same as that at IC1 pin 1 the JFET Q3 will have DC across it and that might be creating problems with the switch.   There are many ways DC problems can arise: 

- IC1 faulty.   That causes DC offsets.  It can also cause DC currents to flow in or out of the opamp inputs.

- Bad solder joint on IC1.  If IC1 has any open input pins the circuit's DC state can wander. 
  Try resoldering the pins of IC1

- C3 leaking.  Initially I thought C3 could be leaking.   The high bias voltage setting could make the voltage across C3 go negative.  That puts a small amount of DC current into IC1.  However, since your measurements show no DC voltage change on IC1 pin 7 so I no longer believe that is the cause.

So where to go next?

Reporting DC voltage measurements would help me see a DC issue.

However, I can suggest a possible fix.  However be warned this fix could hide the fault causing the DC problem.    The idea is:
- Connect a capacitor between IC1 7 pin and R15.  Maybe a 1uF non-polar capacitor.
- If that doesn't work 100% try further adding a 1M resistor.  One side of the 1M connects to the point where R15 and the added 1uF capacitor joint together.  The other side of the 1M connects to the positive side of C20 (anywhere on that track is OK, like IC1 pin 3).

[midiez1997]

Ic1a 1(5,27v) 2(5,27v) 3(5,24 or 5,25v)
Ic1b 5(5,20v) 6(5,27v) 7(5,27v)
I solder contact on IC1!
I'll try with a capacitor today)

[Rob Strand]

Ic1a 1(5,27v) 2(5,27v) 3(5,24 or 5,25v)
Ic1b 5(5,20v) 6(5,27v) 7(5,27v)

When I allow for the 1M input impedance of your multimeter those voltages look correct. 

I solder contact on IC1!
I'll try with a capacitor today)

Yes try the capacitor.   

This is a very hard problem to solve.  I've been on this forum for nearly 20 years and this problem is one of the most difficult to solve.   So many measurements look correct but we know it doesn't work!

[midiez1997]

I'm a little confused ...
I first soldered the capacitor, it seems to be better. Then i soldered the resistor, and nothing seemed to change. Then I otpayal everything, and I realized that nothing seemed to have changed ... There's no subsidence. But! Over the ears beats the sound of the pedal. Ok Like a click, but he is very deaf.

[Rob Strand]

I'm a little confused ...
I first soldered the capacitor, it seems to be better. Then i soldered the resistor, and nothing seemed to change. Then I otpayal everything, and I realized that nothing seemed to have changed ... There's no subsidence.

The easy answer is that might not fix the problem!   

But! Over the ears beats the sound of the pedal. Ok Like a click, but he is very deaf.

I don't quite understand you here.  Are you saying
1) after you do the mod the pedal stops working? and you can only hear a very quiet chorus sound?
or,
(2) After you removed all the changes the pedal *now* doesn't work?

Either way something is wrong.

[midiez1997]

If you remove the resistor and the capacitor circuit works. She also works with them.
But about the delay, I myself was confused.
I'd rather write it down and drop the link to the file, and we'll see graphically what is happening there !!
Because I began to pay attention to cotton.

[midiez1997]

https://soundcloud.com/user-703662903/chorus
The first couple of times I turned on in mode 1. Then in vibrato mode.
I hope this will clarify ..
This track you can download and load it into the editor..

[Rob Strand]

If you remove the resistor and the capacitor circuit works. She also works with them.

OK thanks for clarifying that.  So the conclusion is adding the resistor and capacitor does not change the problem. 

https://soundcloud.com/user-703662903/chorus
The first couple of times I turned on in mode 1. Then in vibrato mode.
I hope this will clarify ..
This track you can download and load it into the editor..

Excellent idea!

That's great. I can see the gap when going from clean to vibrato.  I can measure the gap in samples and it is 50ms (not 0.5s = 500ms).    One thing that makes me suspicious is the 50mS is the time constant of the RC nework  (R61 =1M, C31=47n, RC = 47ms).   That might be a fluke.

When you have a small gap of 50ms it is very difficult to know if there is problem is Mode I.  In Mode I the sound switches from Clean + Delay to Clean.    You always hear sound.  If the Delay signal did not appear until 50mS after pressing the footswitch it might be hard to know it was delayed by 50ms.     In Mode II the signal switches from Clean to Delay.  If Delay is delayed by 50mS there is a gap of no signal so that makes the gap easy to hear. 

Mode I:   Clean, Clean for an extra 50mS, Clean + Delay       ; always has sound
Mode II:   Clean, No sound for 50mS, Delay                          ; gap easy to hear.

So if we go back to Reply #29 and Reply #30.   When you added a 1M resistor across R61 the delay should have changed from 50ms to 25ms.    With the 1M added in parallel R61 becomes 1M in parallel with 1M = 500k,   then R61 * C31 = 23.5ms.    Even though the gap is smaller it is still audible as a gap and that's perhaps why it did not do anything.

So removed all the mods.   

One issue with the existing circuit, which I noticed before, is the RC time constant of R61 and C31 does not match the time constant of C9 and R12.  However, the problem is more than that:  The voltage where the JFET turns on and off is at around 4.5V on the gate this makes the JFET turn-on slower than the turn-off.   So even if the RC time constants match there may still be a mismatch when one JFET turns on and the other turns off.
(There is a way to fix this but wait until the result of the next test)

Put a 470nF (0.47uF) capacitor across C9.  A film cap will be better but an electrolytic cap will also work provided you connect the + side to R12.    What this change does is it delays the turn off Q3.  If you can make that change see what effect it has.  Also it would be very useful to do another recording to see if the size of the gap has changed.

One *very* interesting thing I noticed:  The Boss VB2  Vibrato pedal does not use JFET switching!  It must be the only Boss pedal not to use JFET switching.     That must be done for a reason.  I suspect they could not get the turn-on and turn-off to match and gave up.
------------------
If you have time you could try  R61 = 100k,  (C31, C9, R12 all normal values, 470nF cap removed).  A recording would be useful.

[Rob Strand]

I simulated the circuit with different values of the voltage on C20 set using the Bias Adjustment trimpot.  I have called the voltage on C20 "Vb".

Click twice to enlarge:


This diagram shows the signal gap at one end and overlap or a gap at the other end.  When the voltage is more than 100mV that means both switches are on at the same time.  When the signal is zero that means both switches are off at the same time.

When I set the trimpot voltage to 5.3V I get similar behaviour to what you are seeing.

When the trimpot is set to a high voltage the gaps gets larger at *both* ends.   At the end, there is overlap but if Vb is increased the overlap decreases then at some point you will get a gap also.

So I think what you are seeing is "normal" for that circuit.   Also I do believe the problem does occur in Mode I but it is hard to hear the problem because there is no gap like in Mode II.

I will work on a fix.     The problem is the gap and overlap depends on the trimpot setting.

[Rob Strand]

If you have time you could try  R61 = 100k,  (C31, C9, R12 all normal values, 470nF cap removed).  A recording would be useful.

I think you should try this mod.   Record the results so we can see the reduction in the delay.  If the reduction in the delay agrees with the simulation then we are on the right path.

If we are on the right path I have a number of solutions to the problem.  The problem is it may require changing about 5 parts and adding a resistor and diode.    The "right" part values depend on the setting of the trimpot and the JFET characteristics.

I can decrease the gap without so many changes.   That will be an improvement but not 100% fix, you might still have a very small gap.

[midiez1997]

so! I replaced p61 with 100k.
C31, C9, R12 left unchanged.
Here is the result of this mod:
https://soundcloud.com/user-703662903/chorus-r61

[Rob Strand]

so! I replaced p61 with 100k.
C31, C9, R12 left unchanged.
Here is the result of this mod:
https://soundcloud.com/user-703662903/chorus-r61

Thanks for doing that.   I'm now convinced the switching overlap/underlap is the cause of the problem.
(Not sure why the level is going up and down.  The change from 1M to 100k won't cause that.  Maybe it's your hand moving closer to the enclosure when you press the footswitch.)

To me it is better than before.

So the problem now is *sometimes* you get a click.    The click is actually a small gap (or a bad join),  probably about 10ms.  It's small enough that we don't perceive it as a gap like the 50ms case.  I can't actually see it on the waveforms.

The question is how many mods do you want to do to reduce that gap?   Also, there a small risk after those mods you might still have small tick.   It is a delicate operation to swap waveforms without any ticks.  The reason is part tolerances can cause a gap even when "in theory" there is a small overlap (no gap).

I had three mods:
- The one you did.  Very simple.
- A slightly more complex one.  In practice I don't know if it will have less click than the first.
- A more complex one.   However this one lets you adjust the parts to tweak the overlap/gap.

Are you happy with it as is?
How many mods are you prepared to do?

[Rob Strand]

OK here's the simplest circuit which lets you tweak the timing.
It should do the job.

Please let me know if you think that is too many changes.

Click twice to enlarge:


So,
Change R11 from 1M to 10M
Change R12 from 100k to 1M
Add two series networks 270k + 1N4148 across R61 and R12.

[midiez1997]

What is the first one? Only with the replacement of r61?
tell the other two, we will try!)
The one that was attached to me below, now I will test!
oh/ not sure that i have a resistor on 10M resistor...
And instead 270k, i solder a 240k!

[midiez1997]

"Change R11 from 1M to 10M
Change R12 from 100k to 1M
Add two series networks 270k + 1N4148 across R61 and R12."
I did like this.
True, I'm not sure that the resistor is 10M, since my multimeter measures a maximum of 2000K.

[Rob Strand]

What is the first one? Only with the replacement of r61?
tell the other two, we will try!)
The one that was attached to me below, now I will test!

Yes.

I tried about 10 ways.   At first I tried to only change a small number of parts but the solution was not reliable.  So I went with solution #2 which can be adjusted.   It should be possible to get #2 working.

#1) Change R61 only
#2) Attached schematic "V1.0" , 2019/03/27
#3) Minor improvement, similar to #2 except change also R58 and R59 from 56k to 33k and change 2x 270k to 2x1M.

In case 2 and case 3 you might need to adjust the resistors in series with the diodes.   The reason is the values I showed are theoretical but you can do better by adjusting on the real unit.  The resistors in series with the diode don't both need to be the same value.  You can adjust each one separately to reduce gaps.

oh/ not sure that i have a resistor on 10M resistor...
And instead 270k, i solder a 240k!

There is a trick,
If you have a 1M resistor, measure the 1M resistor,  call that measurement R1.
Then measure the "10M" (R2) and the 1M in parallel, call that measurement R3.
Then calculate R2,
R2  = 1 /( (1/R3) - (1/R1) )
So if you measure R1 = 1M and R3 = 909k,  R2 = 1/(1/909k - 1/1M) = 9.99M
and if you measured R1 = 1M and R3 = 825k, R2 = 1/(1/825k - 1/1M) = 4.71M

I did like this.

Do you hear any problems?  gaps or ticks.     

It would be good if you could post an mp3.  It is hard for me to use the waveform plot.   At "Vb" it looks like the waveform has a small gap.

[midiez1997]

https://soundcloud.com/user-703662903/chorus2803