Help with the simulation of Phase 90 on LTSpice!!! JFET's giving 0.3 Volt sweep

[savethewhales]

The "separate" lug appears to be the screen connection, so removing it wasn't a good idea.  It's easy enough to work out which lug is connected to which connection using a DMM.  Having an extra switch connection isn't helping.  Try to find stereo jacks in future with only three connections.

Exactly!!! That is sold as a stereo 6.3 socket. But with the multimeter I've tried but couldn't find where it connects, except for where it is touching, as we can see on the image. If that wasn't yet chopped, I wouldn't use it anyway..

Anyway am I thinking right about the connections of the other lugs?

I just wonder what the hell is happening. Maybe I'll have to buy another 6.3 stereo socket, but a good one this time, and hope it works.

[bluebunny]

If you can't determine with your multimeter what lug connects to what part of a plug, then you need a new multimeter too.

As for the lugs you left behind, the one in the middle (it's a bit smaller than the others) is connected to the useless switch contact.  This was the one to remove.  The furthest one in the picture connects to the tip.  That leaves the nearest lug which must connect to the ring.

[savethewhales]

If you can't determine with your multimeter what lug connects to what part of a plug, then you need a new multimeter too.

As for the lugs you left behind, the one in the middle (it's a bit smaller than the others) is connected to the useless switch contact.  This was the one to remove.  The furthest one in the picture connects to the tip.  That leaves the nearest lug which must connect to the ring.

Hmm alright! What about the sleeve?

I tried, with the multimeter, connecting the battery and seeing where it connected. The middle lug (which you say connects to the useless switch), connected to the "leg close to the tip", which disconnects from the tip when the input jack is inserted.

[bluebunny]

Hmm alright! What about the sleeve?

I tried, with the multimeter, connecting the battery and seeing where it connected. The middle lug (which you say connects to the useless switch), connected to the "leg close to the tip", which disconnects from the tip when the input jack is inserted.

The "sleeve" (i.e. screen) connection is the lug you cut off.

Forget a battery for the moment.  Also forget that smaller middle lug.  Put your multimeter on a continuity or resistance setting.  Put one probe on one of the other lugs you have left, and leave it there.  Now touch each of the two springy contacts (that would make contact to a plug if one were inserted) to find out the correspondence between lug and contact.  Repeat for the other lug.

TBH, you're only going to get so far with a butchered stereo jack.  Just throw it out and replace it with a new one.

[savethewhales]

I am sorry for taking too long to post, it's because the pedal worked 100% and I ended up staying more time on my university to make all the needing tests. In the end I came home too late for posting right after.

The "sleeve" (i.e. screen) connection is the lug you cut off...
...Just throw it out and replace it with a new one.

I tested it with the multimeter and the middle one goes to the biggest "lug" which is touching the tip when the jack isn't connected. Anyway this is confusing me a hell of a lot. After I'm done writing the report (final project from college), I will go and buy a good stereo jack and see if it works. If not, bye bye battery.

Now for the pedal behaviour itself, those clicks I was mentioning before were happening when the box wasn't closed, it was kind of the skeleton of the pedal yet. So it was happening a problem of grounding I believe, because:
1 - The frequency response, as I put here before in a post, was actually flat where the notches don't exist and there was no trace of clicking problem;
2 - When I closed it, the clicking was way better or didn't even exist. If I tried, I would hear it, but whenever I'm playing the guitar, the noise disappears. So I guess the "grounding" is way better now (not perfect yet).

About the resolution of the plots, I was able to test with more points and saw that with the problem of the squared curve disappeared/diminished a lot when testing with more points. Nevermind the scale of the graph or size of the image:
With 60 plot points:


With 200 plot points (the maximum I could because it takes too long):

[savethewhales]

Hello!

I'm having some oddities with my LFO.
Here's the circuit I'm using:



And this is the response I'm having at the input of the JEFT gates with the minimum LFO frequency:


I didn't get this response when I was designing the circuit or testing in breadboard, so that's why I'm finding it strange..

Below there's the reponse with the maximum LFO frequency.


It's good! I don't know why it changes so much between this frequency and the lower freq.. I'm guessing duty cycle, but then, shouldn't that strange wave be the higher frequency one, because there wouldn't be enough time for it to complete the integration? 

Also, I don't know where these strange clicks come from:


I'm also seeing a change on the offsets and the range on the different LFO frequencies without even changing anything besides the speed:




What's going on here? How can I have the same offsets, same voltage range and same waveform independently of the frequency? Thank you very much for your attention!

[Rob Strand]

And this is the response I'm having at the input of the JEFT gates with the minimum LFO frequency:

I'd be thinking losses in the electrolytic caps 100uF+ 10uF.   Maybe one is in the wrong way.  Maybe they are different parts (brand type) to the ones you previously tested.

What's going on here? How can I have the same offsets, same voltage range and same waveform independently of the frequency?

One of the electrolytic caps around the wrong way might do it.  So too could be the fact to are using two different values in series.

I think both of these problems *could* be related to the electrolytic caps.

Also, I don't know where these strange clicks come from:

Ticks are fairly common with circuits that have an LFO.

I'd say two common causes would be:
- using dual/quad opamps where different opamps in the same package are used for audio and for the LFO.
- Not enough bypassing on the supply at the LFO.  Try adding, even temporarily,  a 100uF cap across the pins of the LFO (specifically the Schmitt trigger opamp).    If that doesn't work the next step is to put a resistor between the main power and the power for the LFO circuit.   Something like 10ohms to 100 ohm.  You will need to keep the 100uF cap on the LFO power.   The idea here is to isolate the LFO power.

There's other possibilities like the LFO wiring is near or touching audio wires in the enclosure,  PCB tracks on the LFO are close to audio tracks.    Ground tracks sharing LFO and audio signals.

Another one is the LFO is causing fluctuations on Vref or Vbias and that is injecting click into the audio; as the any variations in Vref will appear at the audio output.

[savethewhales]

I'd be thinking losses in the electrolytic caps 100uF+ 10uF.   Maybe one is in the wrong way.  Maybe they are different parts (brand type) to the ones you previously tested.

Ok. I think the ways of both are correct: Reversed to each other. However I think I used (as far as I remember) the exact same caps as I am using now on the circuit. In a first stage of testing I used a single cap with a different value, but I think I remember testing with these two the way they are now.

One of the electrolytic caps around the wrong way might do it.  So too could be the fact to are using two different values in series.

I think both of these problems *could* be related to the electrolytic caps.

Hm... I get it. It all comes down to the 2 electrolytic in reverse series.. Which is something I didn't want to happen because I can't really explain how they work, except for saying that this is widely used.
Again, I'm 99% sure they are the right way. But on saturday I'll open the circuit and check that. It's just I do'nt want to hang with the open circuit till there..
Anyway here goes the real schematic on EasyEda:



We can see that they are in the right direction, but let's see if I didn't screw up on assemblying.

Ticks are fairly common with circuits that have an LFO.

I'd say two common causes would be:
- using dual/quad opamps where different opamps in the same package are used for audio and for the LFO.

I think there is no dual op amp here used for LFO and audio at the same time, as far as I see on the EasyEda schematic..

- Not enough bypassing on the supply at the LFO.  Try adding, even temporarily,  a 100uF cap across the pins of the LFO (specifically the Schmitt trigger opamp).    If that doesn't work the next step is to put a resistor between the main power and the power for the LFO circuit.   Something like 10ohms to 100 ohm.  You will need to keep the 100uF cap on the LFO power.   The idea here is to isolate the LFO power.

Fair enough.. it seems something like the P90 LFO that has that capacitor and that resistor that you said were just controlling the rises/falls of the schmitt trigger (that weren' that important). I guess I'll do it on the LTSpice simulation, because my circuit is already soldered. I would just solder anything on top if I were sure that it would work. If not, I'm afraid of damaging the circuit.
Just for the record, what's the "main power" and the "LFO power"? 

There's other possibilities like the LFO wiring is near or touching audio wires in the enclosure,  PCB tracks on the LFO are close to audio tracks.    Ground tracks sharing LFO and audio signals.

Interesting cause ground tracks most probably are the same.. But the wiring seems to be correct, even though I will double check it.

Another one is the LFO is causing fluctuations on Vref or Vbias and that is injecting click into the audio; as the any variations in Vref will appear at the audio output.

This is what used to happen in the P90 LFO, in the simulations actually. I tried to isolate the parts as much as I could with all those op amps on the LFO for this not to happen but I guess it's not that easy after all.

[Rob Strand]

Hm... I get it. It all comes down to the 2 electrolytic in reverse series.. Which is something I didn't want to happen because I can't really explain how they work, except for saying that this is widely used.

I guess one thing that's not common is putting two different values in reverse series.    It kind of asking to expose the complexities of putting electrolytics in series.   Why not just use two of the same values?   You might need to experiment with the value to get the same LFO frequency.

As far as debugging the problem, it could be worthwhile replacing the two electro's with equal values anyway.

I could speculate that when you built the circuit it did work fine but after testing for long periods the one of the caps has lost its oxide and gone leaky.

Fair enough.. it seems something like the P90 LFO that has that capacitor and that resistor that you said were just controlling the rises/falls of the schmitt trigger (that weren' that important). I guess I'll do it on the LTSpice simulation, because my circuit is already soldered. I would just solder anything on top if I were sure that it would work. If not, I'm afraid of damaging the circuit.
Just for the record, what's the "main power" and the "LFO power"? 

Yes, that's another trick.      It can also be used on a circuit like yours.  Checkout C26 and R38 on the Boss BF2 Flanger,
https://www.hobby-hour.com/electronics/s/schematics/boss-bf2-flanger-schematic.gif

Interesting cause ground tracks most probably are the same.. But the wiring seems to be correct, even though I will double check it.

Putting a big cap across the LFO/Schmitt trigger opamp power rails can help here as well.   It keeps the LFO current pulses local to the LFO opamp and cap and stops the pulse going down the ground tracks.   

Best to concentrate on the power rail caps stuff first since that's probably the most common fix.

[savethewhales]

I guess one thing that's not common is putting two different values in reverse series.    It kind of asking to expose the complexities of putting electrolytics in series.   Why not just use two of the same values?   You might need to experiment with the value to get the same LFO frequency.

As far as debugging the problem, it could be worthwhile replacing the two electro's with equal values anyway.

Yeah that's right.. It was because I didn't even had the right values of electrolytics caps to do this method (not even worth mentioning the non-polarized). In that way I was left with a 100uF and a 10uF.
It could be causing those problems but actually the equivalent capacitance is equal to 9.09 uF as the simulation on Spice agrees with real life testing.

I could speculate that when you built the circuit it did work fine but after testing for long periods the one of the caps has lost its oxide and gone leaky.

I thought something like this but.. it was only one week.. how is it possible?

Yes, that's another trick.      It can also be used on a circuit like yours.  Checkout C26 and R38 on the Boss BF2 Flanger,
https://www.hobby-hour.com/electronics/s/schematics/boss-bf2-flanger-schematic.gif

Sure.. it's the exact same from the P90 circuit.

Putting a big cap across the LFO/Schmitt trigger opamp power rails can help here as well.   It keeps the LFO current pulses local to the LFO opamp and cap and stops the pulse going down the ground tracks.   

Best to concentrate on the power rail caps stuff first since that's probably the most common fix.

Nice!!! You talk about connecting the big cap literally to the 9V and 0V points of the op amp?

[Rob Strand]

I thought something like this but.. it was only one week.. how is it possible?

I do suspect it's leakage, and it's likely to be caused by using two different cap values, but the detailed cause is a lot of speculation.   Some caps are likely to be more prone to problems than others.
o bad again.

If you replace the caps with two equal values and it all works then you can just move on

If you want to know why that's a separate project.  You could put the two caps on a DC supply to re-plate them.  Look up electrolytic capacitor "reforming".     That's only going to confirm the problem.  It won't fix the cause.   The caps could, and maybe will, go bad again.

Nice!!! You talk about connecting the big cap literally to the 9V and 0V points of the op amp?

Yes just solder it across.   Sticking big caps across the power rails is a quick way to narrow down many problems.  It stops the need to think of more crazy ideas.

[savethewhales]

I am very sorry for taking this long.. I'm writing down the project, being awake till 3h30 morning every day, so as this wasn't a very urgent subject (because the pedal works still), I didn't went to look further on it. 

I do suspect it's leakage, and it's likely to be caused by using two different cap values, but the detailed cause is a lot of speculation. Some caps are likely to be more prone to problems than others.
o bad again.

If you replace the caps with two equal values and it all works then you can just move on

Unfortunately I'm unable to do this now.. I guess I'll accept it.. It's not audible on the pedal anyway (if it was I would for sure desolder and try to fit one new non polarized cap there).

If you want to know why that's a separate project.  You could put the two caps on a DC supply to re-plate them.  Look up electrolytic capacitor "reforming".     That's only going to confirm the problem.  It won't fix the cause.   The caps could, and maybe will, go bad again.

Ok.. thanks!

Yes just solder it across.   Sticking big caps across the power rails is a quick way to narrow down many problems.  It stops the need to think of more crazy ideas.

For sure this was something I could have done but totally forgot to do it while there was time, and now (and for some time) I don't seem to have time to do it. If the pedal works, screw this for now.

[savethewhales]

Hello everyone! This is the part of the circuit that represents the LFO:



NOTE: that the integrator caps are like they are for a matter of showing what I build into the PCB, but the equivalent cap value should be around 9 and 10 uF for the calculations. Also, R32 is a 500k potentiometer and Vref sits at 4.78 Volt.

Now, while I was searching for the integrator cap and resistor values, I did simulations, no math. In that manner I was able to get the frequencies I wanted, between aproximately 7 Hz and 0.3 Hz. I'm writing my project down and I needed to know what are the equations for the frequency of the LFO.. Can someone help me? I've searched online and what I found didn't give me a reasonable answer (when I would calculate, it would always give me like 14 Hz or 0.05 Hz, with these values, which is far from the measurement and the Spice simulations).

Thank you very much, cheers,

Fred

[savethewhales]

Hello!! Please ignore my last post (I couldn't find the modify button to erase it).

So this is part of my final phasing circuit:



Here at the start, R3, C1 and R4 serve to make a high pass filter on the frequency of 7.13 Hz. Also 470 kohm helps getting the input impedance high enough for a pedal.
 
However I took strong inspiration on the P90 LFO and ended up putting the 470 kohm connected to Vreference (which is the fixed voltage coming from the Zener). But I don't actually know why it's done like that (the connection to Vref).. I imagine it's like a coupling? But why do I need to make a coupling at the start of the circuit?

This is what most bugged me at the time of designing the circuit but I just accepted the fact then.. If somebody could help me I'd be very much thankful.

Cheers, Fred

[PRR]

> why it's done like that (the connection to Vref)..]

The power is zero and +9V, right?

But audio swings both ways around "zero". How can the amp do that?

Our usual trick is to "bias" the amplifier DC condition to "about half of supply". Now audio can swing both ways. At the output another cap re-references to zero volts DC.

Is that what you are asking?

[savethewhales]

Is that what you are asking?

Exactly! Thanks!
So what the 470 kohm really makes is getting the non inverting input of U6 to being Vref right?

[savethewhales]

So I got 17 out of 20 in my project of the phaser pedal.. I am very very happy about this!!!

Just have to thank specifically:
- Rob Strand
- PRR
- Everyone that helped me throughout this project (Eb7+9 - JCM, 11-90-an, bluebunny, others)