Schematic critique, please - multi distortion/boost pedal

[patricks]

Hi everyone, I'm requesting some eyes on a schematic I've put together for feedback and critique.

The pedal is the "Cerberat", a three-headed Rat-inspired distortion monster. Based on a JRC2902 I pulled out of a microscope illuminator board and some advice from the board here, I learned the op-amp has similar (poor) audio performance to an LM308. I breadboarded up a rat circuit using one of the four op-amps in the package and it does the Rat thing quite nicely. So I have four to play with!
The Cerberat uses all four op-amps in three circuits that can be used independently or cascaded into one another.
Section A is a more or less standard Rat circuit, with options for Si or LED clipping diodes. Section B is a similar circuit, but feeding an additional amplifier stage that's limited in gain. Section C is a dirty boost, with less gain than section A and without clipping diodes.
It starts with a voltage doubler, not for reasons of clarity but to give sections B and C more chance of dealing with the signal from A and/or B without turning to mush. Clipping diodes in section B, the resistor/capacitor combos in feedback loops of B and C, and all JFET output buffers are biased to Vref (+9V) to be able to handle more signal before clipping.

I downloaded Diptrace over the weekend and got stuck in, and here's the first attempt at a full schematic (updated based on feedback):


I'm planning to test various things including the Reutz mod in Section A, the value of R20 (resistor in the feedback loop of the second op-amp in section B), whether the bias network of the clipping diodes in section B works, and a few other things while I have it on the breadboard, so it's early days still, but because I've put it together from building blocks I'm asking for feedback in case I've drawn out things that won't work, don't make good design sense, etc.

Particular questions I'm hoping to answer are:
1. Is the network after the MAX1044 necessary? Will the Vref supply be improved for the smoothing or can I just take Vref from the low voltage pin and delete a few components? (Assuming taking Vref from the +9V input to the circuit isn't a good idea in case it needs smoothing)
2. In section B, is the resistor/capacitor network in the feedback loop necessary? I understand this network shelves low and high frequency, so is it redundant to have it on both op-amps in this section or will it help to prevent high and low frequencies from distorting too much and making things sound harsh or flubby?
3. Is the connection between op-amps in section B okay, or do I need to revise part values?
4. Is C19 an appropriate value? I based this arrangement on the OCD circuit and just used the existing cap value from the Rat circuit (https://www.analogisnotdead.com/data/medias/circuit-analysis/fulltone-ocd/fulltone-ocd.png?1601279645413)

Any and all feedback is most welcome, and this is a big circuit so feel free to ignore if it's too much work to look over. I'll be testing things as I go and happy to report back if anyone's interested 

[antonis]

1. Better leave it as it is 'cause +18V might be not exactly +18V..
2. Yes it is, 'cause it sets each individual op-amp LPF corner frequency..
3. IMHO, it's OK..
4. Make R12 10k and maintain C19 1nF..

[patricks]

Brilliant! Thank you, I really appreciate it 

[duck_arse]

The pedal is the "Cerberat",

Clipping diodes in section B, the resistor/capacitor combos in feedback loops of B and C, and all JFET output buffers are biased to Vref (+9V) to be able to handle more signal before clipping.

jfets - they are biased to ground, connected between +18V and Vref, 9ishV. that can't be right.

[Steben]

The pedal is the "Cerberat",

Clipping diodes in section B, the resistor/capacitor combos in feedback loops of B and C, and all JFET output buffers are biased to Vref (+9V) to be able to handle more signal before clipping.

jfets - they are biased to ground, connected between +18V and Vref, 9ishV. that can't be right.

Perhaps the intention here is to connect the Rsource to GND and Rgate to Vref.

[antonis]

For what I'm really curious is the current drive capability of Q3..

[patricks]

Thanks everyone!

The pedal is the "Cerberat",

Clipping diodes in section B, the resistor/capacitor combos in feedback loops of B and C, and all JFET output buffers are biased to Vref (+9V) to be able to handle more signal before clipping.

jfets - they are biased to ground, connected between +18V and Vref, 9ishV. that can't be right.

Perhaps the intention here is to connect the Rsource to GND and Rgate to Vref.

I read in another thread about biasing the jfets to Vref:
https://www.diystompboxes.com/smfforum/index.php?topic=114524.msg1065205#msg1065205
I'm just copying and pasting building blocks from other circuits, though, so I'm not sure whether I've done it right (or whether I shouldn't worry about it and just bias to ground)

For what I'm really curious is the current drive capability of Q3..

I'm not sure about that. The original schematic has a 2n5457, I've got a 2n5952 on the breadboard at the moment and sound comes out the other end, but I'm not sure whether I should choose another type of jfet?

[antonis]

For what I'm really curious is the current drive capability of Q3..

I'm not sure about that. The original schematic has a 2n5457, I've got a 2n5952 on the breadboard at the moment and sound comes out the other end, but I'm not sure whether I should choose another type of jfet?

I was talking about R35 value..

[patricks]

Aaaah, I think I see what I've done now...

The pedal is the "Cerberat",

Clipping diodes in section B, the resistor/capacitor combos in feedback loops of B and C, and all JFET output buffers are biased to Vref (+9V) to be able to handle more signal before clipping.

jfets - they are biased to ground, connected between +18V and Vref, 9ishV. that can't be right.

Perhaps the intention here is to connect the Rsource to GND and Rgate to Vref.

The bias to the jfets is set by the 1M resistor on the gate, isn't it? Schematic updated with that connection made and the connection to the drain resistors changed to ground.
I changed the value of R35 to 10k as well, that should fix up the current drive capability, yes?

[antonis]

I changed the value of R35 to 10k as well, that should fix up the current drive capability, yes?

[antonis]

Just a further suggestion..

It should be good for all "unnecessary" AC grounds (like C26, C27, C28, C29, C33 & C34) to be routed to GND (like C24/C25) to keep Vref as clean as possible..

Also, in the cost of two DC blocking caps (on U2.2 output and right after SW2 middle lug) and a bias resistor (on U2.3 non-inverting input), a perfectionist should also route clipping diode pair to GND (instead of Vref)..

[patricks]

Thanks very much!
I'm learning a lot, and grateful for the improvements to the circuit

The thoughts of referencing the clipping diodes in section B and those caps to Vref comes from the OCD schematic plus a couple of threads here. One that said the author's preference was to do that (https://www.diystompboxes.com/smfforum/index.php?topic=114524.msg1065205#msg1065205) and another that said someone had tried it and it caused motorboating until they tied the caps on the non-inverting input of the op amp to Vref (I can't find the link right now, sorry).

My understanding from the reading I've done so far is that this might raise the clipping threshold?
If that's the case, would it be an option to reference these points to the low voltage pin of the 1044 to keep Vref clean?

I also read in another thread that RG was saying he prefers lower value bias resistors than 1M, and even more than 470k, to avoid introducing noise to the circuit
https://www.diystompboxes.com/smfforum/index.php?topic=67499.msg538069#msg538069
Is that worth considering for the clipping diodes and the op amps here?

[PRR]

...I also read in another thread that RG was saying he prefers lower value bias resistors than 1M, and even more than 470k, to avoid introducing noise to the circuit https://www.diystompboxes.com/smfforum/index.php?topic=67499.msg538069#msg538069

You seem to be putting Jay Doyle's words in R.G.'s mouth; and everybody misuses "noise".

[patricks]

Oh, whoops! My bad, thanks for that

[amptramp]

I also read in another thread that RG was saying he prefers lower value bias resistors than 1M, and even more than 470k, to avoid introducing noise to the circuit
https://www.diystompboxes.com/smfforum/index.php?topic=67499.msg538069#msg538069
Is that worth considering for the clipping diodes and the op amps here?

The input stage coming straight from the guitar should be high impedance since the guitar pickup acts as an inductor in series with a generator.  The inductive reactance rises with frequency so if the input impedance is low, there will be a loss of treble.  Some pedals lose the treble deliberately, especially if they are distortions or fuzzes, because these pedals generate exact harmonics that will exhibit a beat frequency with the guitar output which  deviates from true integer multiples of the fundamental frequency.  In an electric guitar, these partials are slightly sharp compared to a true harmonic.  In an acoustic guitar, partials 2, 3 and 4 are usually flat whereas the 5th is very close to an actual harmonic.  The exact deviation is caused by string stiffness and delivery of power to the pickups or in an acoustic, to the resonating hole.

With a pedal input, there are parallel resistances that reduce noise generation.  If the guitar has a volume control and a tone control, these are in parallel with the input resistance of the pedal to the extent allowed by the coupling capacitors and of course, the pickup has a series resistance.  At low frequencies, the coupling capacitor tends to isolate these impedances but that doesn't matter - you aren't going to provide much gain for low-frequency signals anyway because the lower limit of a guitar is the 82 Hz E string and for a bass is 41 Hz.  If you have a 500K volume control and a 10K pickup resistance, your highest impedance is 255K where the resistance seen from the volume control slider would be 255K or half of the total.  A tone control in parallel will reduce this further where the capacitive reactance of the tone cap permits.  Therefore, a 1 megohm input resistor on the pedal doesn't have much effect other than to preserve treble.  If the volume is turned up all the way, the resistance is just the pickup winding resistance in parallel with the volume control, tone control and input resistance and at a typical 10K, it dominates all the other resistances.  The pickup inductance is typically on the order of 3 to 5 Henries.

After the first stage, once you are operating from low-impedance sources, it pays to use low resistances to keep noise down but this should be balanced with the requirement for low power drain and the limits of the devices used.  You don't always want to drive an op amp output to 10 mA just because you can - this adds stress and power consumption without much purpose.

I used to do noise calculations for thermal imaging viewers and there are days when I did eight calculations of variations on a circuit and that took a full day. 

[antonis]

My understanding from the reading I've done so far is that this might raise the clipping threshold?

Definitely not..!!

What raises clipping threshold (for a given diode V_FWD) is the addition (somehow) of a voltage source in series with diodes..
(e.g. a resistor in series with diode pair raises threshold voltage up to V_FWD plus current through resistor times resistor value..)

For a hard clipping configuration (diodes in shunt with signal axis), both diode pair edges should be lie on the same DC potential..
(leaving space only for AC addition/subtraction..)

So, in case of op-amp output DC blocking capacitor absence (like U2.2), diodes MUST be connected to Vref..
(I let you guess the reason why.. )

[patricks]

I love this, I'm learning a lot, thanks everyone
I'm going to have to think on this, it's late in the evening here in Aus, and my brain's tired

[antonis]

it's late in the evening here in Aus, and my brain's tired

A generous amount of Sachertorte could help..

[bluebunny]

Wrong "Aus" I think... 

(Although I don't see why they shouldn't have Sachertorte down under!)

[antonis]

Wrong "Aus" I think... 

I'm not sure 'cause we both had late evening simultaneously..