fz-2 clone PCB

[imJonWain]

Schematic was wrong, ignore this post . Please delete the thread and I'll start a new one when I have a 100% correct fz-2 schematic!

[Jero]

Boss enclosure inner dimensions:
2 & 3/8" W x 4 & 1/2" H...1 & 1/8" D at the top, and 9/16" at the bottom.

[imJonWain]

thanks!

[imJonWain]

Does anyone on here have an original Boss FZ-2 or a hyped fuzz clone built? I have a clone running and am getting oscillation when I have the gain and Hi controls maxed in boost mode. The oscillation starts at the high end of the audio range moves up to about 60kHz and then down to ~34kHz (aka I can't hear it and audio cuts outs). It's pretty much square wave.

No such "problem" in the fuzz modes, which are pretty out there anyway and do semi self oscillate at max gain. The original manual actually warns about oscillation in boost mode when the "depth" is too high.

It's not in a box yet and the leads aren't stellar since I have the PCB sitting on a bread board so I'm sure that doesn't help....

[eatyourguitar]

there are many gain stages in this fuzz. some of them are very high gain. some of them create added upper harmonics. I think one of your outputs is bleeding back to one of your inputs. this can be through induction or through capacitive coupling in the breadboard. your breadboard is made up of metal fingers all packed tightly next to each other. a capacitor is essentially two metal plates with an air gap. when you have the AC coupling between gain stages, that is a high pass filter. all the parasitic capacitances from the breadboard and the long cables or jumper wires is like a cap to ground. this would be an RC low pass filter. if you put all your phases, gains, and frequency responses into a computer, you will find a resonant peak at 34KHz to 60KHz.

this will likely not be a problem when you do it on a 2 layer PCB with a ground plane and power decoupling caps placed close to the opamps. also, if you are using 2N2222, stop using 2N2222. they have crazy high bandwidth. you would need to add capacitance with 22pF from collector to the negative supply to bring the bandwidth back to < 20KHz. for PNP it would be collector to the pos supply. don't take my word for it. you would want that on a switch or socket. you may need to play with the value or even add series resistance to stop the oscillation without totally making it muddy. I also think that 10uF is going to have very low impedance and also therefor the gain could be too much thus creating oscillation. I would suggest switching 10uF to 1uF though you might not need to if you can get the other methods working to stop the oscillation.

I don't even breadboard anymore. I go for the custom PCB. the breadboard is a lie. I only use it for testing LDR or LED brightness. all the DC stuff that just works and must be calibrated IRL for each sample (LDR) or each batch (LED). I use it to test frequency response of one transistor gain stage. bias of one transistor. rise time of one transistor. stuff like that. I still use it for testing fuzz face germanium but I am in the process of building a fuzz face transistor tester PCB for this anyway.

[imJonWain]

Thanks for the input! This is actually a PCB I made and I'm just using the breadboard for external wiring. I'm using 2n3903s (half the hfe 3904) which looking at the datasheet isn't much lower bandwidth then the 2n2222 I guess and I'm using 2n5485's for the jfets. I'm going to experiment with it more today when I have time. While probing with a meter and I scope I found with the input open/floating I have a pretty perfect and small triangle/sharkfin oscillation coming out of the q1 buffer and a large resistor to gnd on the input kills that and also cuts output oscillation in boost mode down to nothing except until the gain is basically maxed... My layout is pretty compact so I wouldn't be too shocked about ac coupling... My old laney amp had that between channels due to layout.

Any idea why boss does not bother with an input R to gnd on their pedals?

I didn't do a full gnd plane, it's more of a U shape which I know isn't ideal for current return distance but it's very similar to what boss did on their PCB layouts back then...

[eatyourguitar]

if the ground plane covers at least one side of the PCB edge to edge all the way around then that is a full ground plane. even if you have a PCB shaped like a horseshoe it is still a ground plane.

the 1M from input or output jack tip to ground is for anti pop when you true bypass that is why you only find them on true bypass pedals. boss pedals are always buffered even in bypass mode. the switching is done with FET's and a transistor flip flop. sometimes they use FET's in the flip flop and the buffers. sometimes they use NPN in the flip flop and FET in the switched buffers. it depends on the year of production. you are always going through a FET buffer no matter if you are bypassed or not. the input output caps never have any change in DC bias since they are in the signal path the whole time.

now if you have 1M to ground AFTER the input cap then that is usually to bias the first opamp gain stage. this 1M would go to virtual ground 4.5v. the kind you see in a voltage divider on the +9v, 0v two 10K resistors and a cap to ground. this is there because the jfet input of a non-inverting opamp is super high impedance. it will self bias to any random leakage current while the input floats. the cap is high impedance with silence. when the audio comes in it could be with a crazy DC bias that immediately makes it clip on one side. the DC bias comes from inside the opamp. the leakage is very small but because there is no 1M to overcome this random DC bias, it persists. adding a 1M to DC bias in the middle of the +9v and 0v rails (4.5v bias) is enough current (4.5uA). that should hint at how small the 1458 or TL072 leakage is.

P.S. that footprint for the DIL8/DIP8 opamp has very small pads. you should modify the part library or use a different library. it will be easier to assemble. there are tutorials online how to build a part in eagle.

[imJonWain]

Cool, Yeah I understand all that. What I was wondering was doesn't the bias network (divider or whatever) or having an antipop resistor add somewhat to input filtering of noise?

Yeah I need to clean up my Eagle libraries, they are a mess.

[eatyourguitar]

What I was wondering was doesn't the bias network (divider or whatever) or having an antipop resistor add somewhat to input filtering of noise?

these are two questions so we have to take them one at a time.

What I was wondering was doesn't the bias network (divider or whatever) add somewhat to input filtering of noise?

what you call the bias network is actually probably called virtual ground which is also called a split rail supply. this is for people that want to have dual supply opamp circuits powered from a cheap single output DC power supply like a boss 9v. it is hard to say if it is better or worse with respect to power supply ripple. a lot of that has to do with what 9v supply you use and what filtering the pedal has on the power entry. then we have to pick the dual output 4.5v power supply we will use for comparison. basically they don't exist because everyone uses a voltage divider. how much filtering or regulators you will add to that basic design varies quite a lot. we need to look at two specific schematics if we are going to draw conclusions about power supply ripple or conclusions about a pedal circuits ability to reject power supply noise.

regardless though, to try and answer the question you asked, you really don't have a choice if you are using a circuit that requires input protection, DC blocking, input bias, or any reference to the virtual ground. opamps are differential by default. any inverting gain stage will need the + input tied to virtual ground (to measure the DIFFERENCE in voltage between two inputs). adding 50K from + to virtual ground there will reduce noise very slightly. you just can't have an inverting gain stage without virtual ground anywhere. this is probably true for non-inverting gain stages also since they will float up to static electricity bias.

the only conclusion we can draw from a 10K+10K voltage divider across the power supply is that it will load the power supply for 20K all the time. you can use this to make other calculations with specific circuits and draw conclusions from there. the quality of your virtual ground greatly affects your circuits noise floor so it is definitely an important consideration during the design process. this is why you will see caps on the 9v and 4.5v rails. this creates an RC low pass filter on all the power rails.

What I was wondering was doesn't having an antipop resistor add somewhat to input filtering of noise?

YES! changing the corner frequency of your input filter does have an effect on noise. it is a high pass filter with or without the 1M or 5M to ground before the input cap or after the output cap. high pass because you have a DC blocking cap on your input and your input is AC coupled. the power supply has an RC lowpass filter to remove AC and pass DC. it can be a complicated subject but to sum up, that 5M there will reduce some of the hum because it reduces the impedance of induced hum to ground. it may also attenuate the bottom end, increase high frequency, therefor increasing the user perception of high frequency noise. usually the noise is very small in guitar circuits but the high gain and shitty power distribution leads to pretty bad noise floor for most guitar players. it is possible that most people can't hear the difference with or without a anti-pop resistor. that has been my experience most of the time. it depends on many factors. guitar pedal designers are sending these things out into unknown conditions. we like to have protection on things. we always filter out anything that is beyond the carnegie chart for a guitar. there will always be low pass filters and high pass filters everywhere.

the trend in high end audio is to DC couple everything with a servo balance to remove DC offset manually but in an automated way with a perfectly opposite counter DC offset to null the DC component. this is how SSL has always done it.