Fuzz Pedal Oscillation issues

[RandyBeast]

The most useful tool I have for tracking down oscillation problems is my finger. Poke around the PCB touching wires, parts, traces, pads, until you find a spot that at least changes it or kills it. That will help you to focus on an area that is providing the necessary positive feedback to oscillate.

Hmmm, interesting, I will give this a try and report back. thanks for the tip! So if my finger changes or stops the Occsillation, does that mean I would need to move the location of the component? for ind a better grounding solution?

Thanks!

[RandyBeast]

That's why I have a test point, you plug the meter into that spot and adjust the trimmers to get 4.5V at T2 collector. I have no problems at all getting the LEDS to conduct correctly, and it sounds great. but this is beyond the scope of this conversation.

Thanks

That 4.5v is a DC 'Q-point' or quiescent point voltage. It is nothing to do with the maximum AC signal voltage that you will be feeding to the green LEDs which is determined by the potential divider formed by the resistors in the collector of Q2 and the ratio of them. In your circuit, the maximum AC voltage swing of the output from Q2 feeding the diodes is never going to exceed about 1 volt peak-to-peak and typically when biased correctly will be more like 350mV peak-to-peak. If you want to drive those diodes with a larger signal then you need to increase the value of the 470 ohm resistor considerably and reduce the BIAS2 pot resistance or just take the signal directly off of the collector of Q2 to feed to the diodes.

You are right, I get an AC around there, I can't tell you exactly why it works, but it does. if I bypass the 500K trimmer, tons of feedback and Buzz comes through, this circuit cleans it up considerably. I can send videos to prove it. But thanks for your Feedback.

[johngreene]

The most useful tool I have for tracking down oscillation problems is my finger. Poke around the PCB touching wires, parts, traces, pads, until you find a spot that at least changes it or kills it. That will help you to focus on an area that is providing the necessary positive feedback to oscillate.

Hmmm, interesting, I will give this a try and report back. thanks for the tip! So if my finger changes or stops the Occsillation, does that mean I would need to move the location of the component? for ind a better grounding solution?

Thanks!

Your finger is basically a big capacitor with some resistance thrown in depending on how much you perspire. So it will either kill it or at least change its frequency. Depending on what you touch to cause this will mean different things. If it is a trace on the PCB you may have a layout error or short on the PCB. If it is a wire, then you need to redress and/or shorten it. Chances are if it is a wire, just moving it around will stop the oscillation.
Most importantly it just tells you were the circuit is "hot" for providing feedback.

[antonis]

R2/R3 is High Values to keep any Current from entering my Bias signal.

I've to admit I can't get you..
1M (R15) is big enough to prevent any current "entering" into ICB non-inverting input..

C2 helps stabilize the VB and I don't have an R12

What I was proposing is to delete C12 and R15 (not R12, my bad..) thus saving 2 items and further "isolating" signal path from VB..
(after all, IC1A output is low enough for needing a buffer after it..)

I know the role of C2..
Could you plz compare 100k/100k voltage divider output impedance with the IC1B one..??

What I'm saying..

[RandyBeast]

Thanks for the clarification, do you think the Voltage Bia is what's causing the Oscillation?

thanks

[antonis]

IMHO, no..

Most suspect is PCB layout and IN/OUT wiring..
(you see, there is plenty of room on breadboard and items/wires/jumpers are sparsely populated so no interference/crosstalk etc..)

[RandyBeast]

Welcome to the forum!  Electronics, like music, has a steep learning curve but once you get better at it, it becomes enjoyable.

I'm not happy with the lines going to the base of T1 crossing near the output jack and the back-to-back location of the input and output jacks almost guarantees oscillation.

Typically, the Fuzz Face uses a 100 K resistor where you have the feedback pot - the pot leads are connected to the input and they can pick up output signal from almost everywhere.  A fixed resistor may help with this.

You have an op amp gain stage with a gain of about 23 at the output of a Fuzz Face stage that was not necessary in the original Fuzz Face.  This would definitely boost the signal to the point where oscillation would be probable with the output so close to the input.

The output and input signals are near each other at the H1 connector which I presume is the cable to the bypass switch.  Although the leads are not adjacent, the input and output run in close proximity for the length of the cable.

The lead for the base of the first transistor winds all the way up to the top of the board to pick up C6 and the 250K pot then comes back down to pick up C3 and C4 and they go to the big cap switch in the middle.  That's a long lead covering a lot of area.

I'm not surprised your breadboard worked but the PCB oscillated.  You may need to revisit the layout to get the input signal and output signal a bit more separated from each other.

Hello amptramp,

I took some of your thoughts on the layout and did a new one. let me know if you think this will get better results. I separated each of the sections the best I could and made the leads much shorter for travel distance.

[anotherjim]

I don't want to reverse the polarity of that cap because it would explode with the positive voltage hitting its negative end. however, I am curious to change the VA end on just the cap to GND to see what happens.

I don't think you understood. The gain bypass cap doesn't have to point "down" as it does on the original positive ground PNP fuzz face, it should point to ground which is now a negative ground. So if the gain pot is now at a positive voltage, then it connects to the positive end of the cap and the negative end of the cap goes to ground instead of VA. Why would it then blow up?

[amptramp]

The new layout definitely looks better.  If you don't want to go through the expense and delay of getting custom boards made, there are prototyping boards available in various sizes and descriptions like this:



These are designed for IC's but they can be used for random analog components as well, and you have both.  I have built most of my stuff on this type of board and you can get boards of various sizes with 3, 4 or 5 pads on either side of the power lines.  But you may not get the density you want with these.

In addition to my comments on the circuit from yesterday, it may be worthwhile to add a small capacitor across R14 to maintain stability.  A cap of 100 pF would give you a rolloff at 3388 Hz so reducing the capacitor or resistor would raise this frequency.  A 47 pF cap would give you a rolloff at 7208.5 Hz and that would be adequate.  This capacitor adds a feedback lead to counteract any capacitance to ground at the inverting input.

[RandyBeast]

I have tried the cap between R14 before but ended up not liking what it did to the tone. I did a 470pf to help with high end buzz though.

I will look into these boards and try this out.

I will report back once finished. It might be a couple of weeks, so hang tight. Thank you so much for your great input.

Thanks

[RandyBeast]

Ok Everyone,

Good and Bad news.

Let's start with the good news:
No more oscillation!

Bad News:
weak distortion.

I had to do more troubleshooting on this. I changed the Transistors to a silicon base 1N3906 these are easier to get and have an average hfe of 250. this helped, but still was missing the sauce that my breadboard version had. I hooked up my line 6 delay to loop a rift through so I could play with the bias and listen to the sound. But by doing so, the pedal was working perfectly! I didn't have to do anything to the bias. that told me there was a buffer on the output of this delay pedal. so I got the breadboard out and created a buffer.

Going straight out of my buffer, the pedal worked, but oscillation was back stronger than ever. I put a resistor in series on the buffer output, a very small value like 330 ohm and oscillation stopped and the pedal sounded great!

Now for the part I need help understanding.

All the fuzz pedals I've seen, you want nothing at the input. it should be straight guitar and pedal. why would mine only work once the guitar is buffered? and why is the Layout changes on the PCB make the pedal react sooo different between designs?

[johngreene]

.....!

Now for the part I need help understanding.

All the fuzz pedals I've seen, you want nothing at the input. it should be straight guitar and pedal. why would mine only work once the guitar is buffered? and why is the Layout changes on the PCB make the pedal react sooo different between designs?

The first obvious answer is input impedance. Your design is more stable with a lower source impedance driving it.
You are probably running into the old "converting a Fuzz Face to a Negative Ground" problem. There has been a lot of debate on why they tend to oscillate when this is attempted but what I found is that the problem has to do with the power supply not having a low enough internal impedance allowing a small amount of "sag". This sag creates a potential difference between the base and emitter when the transistor turns on hard due to the "anti-pop" resistor being referenced to the "-" and the emitter is referenced to the "+". The input cap should prevent this but when the transistor suddenly switches "on", that transition gets through the capacitor and is amplified by the transistor which further increases the sag until the transistor clips. At this point there is no more "change" in the power supply and depending on the time constant of the input cap, the transistor jerks back to its bias point creating a transition in the power supply in the opposite direction which is then amplified by the transsitor shutting it completely off. Thus an oscillation whose frequency is dependent on the size of the input cap. When the "anti-pop" resistor is referenced to the same rail as the transistor emitter, this transition on the power supply is in phase so therefore no positive feedback.
It takes a very, very, very low ESR capacitor(s) of sufficient capacity to prevent this sag on the supply rail enough to prevent this.
A resistor in series with the base of the input transistor will help considerably but it usually takes a lot away from the sound. I wasn't able to find any solution, other than on-board voltage regulation, to stop the oscillations that didn't have a negative effect on the sound for this design.

[FSFX]

Another point regarding stability of negative ground PNP Fuzz Faces is the the decoupling capacitor on the fuzz pot on the emitter of Q2 should go to ground rather than be connected to the +ve supply. That way you have the lowest impedance path to ground. If you have it connected to the +ve supply then the emitter decoupling capacitor and the power supply decoupling capacitor are in series and form a capacitive voltage divider. This means that some of the signal from Q2 emitter (which looks a bit like an emitter follower and thus has a low output impedance) gets feed back onto the +ve power supply. A good recipe for instability, oscillation and motorboating. 

[johngreene]

Another point regarding stability of negative ground PNP Fuzz Faces is the the decoupling capacitor on the fuzz pot on the emitter of Q2 should go to ground rather than be connected to the +ve supply. That way you have the lowest impedance path to ground. If you have it connected to the +ve supply then the emitter decoupling capacitor and the power supply decoupling capacitor are in series and form a capacitive voltage divider. This means that some of the signal from Q2 emitter (which looks a bit like an emitter follower and thus has a low output impedance) gets feed back onto the +ve power supply. A good recipe for instability, oscillation and motorboating.

I disagree. The emitter is referenced to the +ve supply rail. The pot should be connected to the same reference rail as the emitter of the transistor it is connected to.

[FSFX]

Another point regarding stability of negative ground PNP Fuzz Faces is the the decoupling capacitor on the fuzz pot on the emitter of Q2 should go to ground rather than be connected to the +ve supply. That way you have the lowest impedance path to ground. If you have it connected to the +ve supply then the emitter decoupling capacitor and the power supply decoupling capacitor are in series and form a capacitive voltage divider. This means that some of the signal from Q2 emitter (which looks a bit like an emitter follower and thus has a low output impedance) gets feed back onto the +ve power supply. A good recipe for instability, oscillation and motorboating.

I disagree. The emitter is referenced to the +ve supply rail. The pot should be connected to the same reference rail as the emitter of the transistor is it connected to.

OK then we will have to agree to disagree on this one. From that comment I don't think you have a full understanding of what is really happening with this type of circuit.

I have worked with many of the 'inverted' PNP designs and can assure you that the emitter decoupling capacitor needs to be connected direct to ground. Any circuit analysis or simulation will confirm that having it connected to the +ve power supply will be far less stable.

I stand by my assertion 100%.

[johngreene]

Another point regarding stability of negative ground PNP Fuzz Faces is the the decoupling capacitor on the fuzz pot on the emitter of Q2 should go to ground rather than be connected to the +ve supply. That way you have the lowest impedance path to ground. If you have it connected to the +ve supply then the emitter decoupling capacitor and the power supply decoupling capacitor are in series and form a capacitive voltage divider. This means that some of the signal from Q2 emitter (which looks a bit like an emitter follower and thus has a low output impedance) gets feed back onto the +ve power supply. A good recipe for instability, oscillation and motorboating.

I disagree. The emitter is referenced to the +ve supply rail. The pot should be connected to the same reference rail as the emitter of the transistor is it connected to.

OK then we will have to agree to disagree on this one. From that comment I don't think you have a full understanding of what is really happening with this type of circuit.

I have worked with many of the 'inverted' PNP designs and can assure you that the emitter decoupling capacitor need to be connected direct to ground. Any circuit analysis or simulation will confirm that having it connected to the +ve power supply will be far less stable.

I stand by my assertion 100%.

[FSFX]

The emitter is referenced to the +ve supply rail. The pot should be connected to the same reference rail as the emitter of the transistor it is connected to.

The emitter is not referenced to the +ve supply rail.
That is just the source of the DC current to the transistor.
For AC, the signal is referenced to ground. That is why you should always decouple the +ve supply rail to provide a low impedance path to ground.

[johngreene]

The emitter is referenced to the +ve supply rail. The pot should be connected to the same reference rail as the emitter of the transistor it is connected to.

The emitter is not referenced to the +ve supply rail.
That is just the source of the DC current to the transistor.
For AC, the signal is referenced to ground. That is why you should always decouple the +ve supply rail to provide a low impedance path to ground.

If you consider the transistor circuit as it's own domain with a flipped power supply, then externally to that there should be only 4 connections: power (-9V), ground, input and output. From the "outside" the input and output are "decoupled" with capacitors so "ground" is whatever you connect to the "ground" connection.  This means the volume control can reside outside this box if it is to also serves as the "anti-pop" reference for the output. But everything else is contained inside this power "domain". Since the distortion pot is capacitor decoupled it could technically live outside of this domain but that would mean another external connection which is an invitation for a ground loop.
The pot carries AC current because of its connection to the emitter. However, because it is out of phase with both the input and output, any "leakage" would result in negative feedback and therefore increased stability at the cost of slightly lower gain. 

[FSFX]

All I know is that doing it my way results in a 26dB reduction on noise at 1kHz feeding through to the VA supply as well as reversing the phase of the noise voltage. That is using Nichicon capacitors for the 47uF and 22uF capacitors.
So a reduction of 26dB of potential noise feeding back to create instability.

No point arguing anymore. Point proven.

[johngreene]

And off topic of the original post. But thanks for sharing.