Questions about bass distortion design

[Jubz]

Hi.

Ive breadboarded a RAT distortion to play with my G&L active bass and would make some improvements before trying to make a perfboard layout for it.
I usually find bass distortions too muddy or muffled for my tastes. So i decided to design a circuit based on some simple ideas.

Rolling off the bass and activate the treble boost active mode of my bass usually restores some brightness.
I like RATS distortions for bass (especially the grindy ones like the fuzzrocious).
Ive made myself a breadboarding test box with +15/-15V supply.

It uses only simple and well documented building blocks (necessary because i have hard times understanding how it works). The signal is splitted after a jfet buffer. One path has a bass cut (density control) then opamp treble boost similar to the one my bass has, LM308 gain stage (lower gain than a traditionnal rat especially for higher frequencies) and clipping diodes to ground, 2 bands tonestack and recovery gain stage. The other path has a treble cut then a slight bass boost. The 2 paths are passively mixed then buffered, without any coupling caps (the way darkglass b3k or tech 21BDDI do it).

Ive learned a lot (essentially on this forum) and had lot of fun trying a lot of different types of EQ post clipping (gyrators, Big muff, baxandall) and pre clipping (notch filters, 2nd order high pass) and finally the voicing of the pedal satisfies me with "simpler" rc filter and passive tonestack. There are always lot of points i cant understand

1) Which value is best for the source resistance of the JFET buffer? I dont see a lot of JFET buffers using dual supply. The ESP article on voltages followers shows 1k (but the author says its not optimised) and one other article ive found on this forum was showing 5,1K. Ive settled to 4,7k by ear but it bothers me to not understand how to choose it.

2) Clipping of gain stages which are meant to stay clean. Even with 30V headroom, the bass boost of the clean path clips. With a gain of 2,2! I suspect (but im not sure) that the tonestack recovery gain stage is clipping too with gain of 2,7 applied to a signal which cant exceed 0,7V - tonestack loss. It is normal or do i have a design flaw here ? For the recoveray gain stage, i dont even understand how it bias. Ive assumed C13 is the coupling cap and the serie resistance to ground of R18 and the bass and treble pot references to ground the input...

3) the density tonestack. Ive layouted it like that and liked how it sounds but i dont know how it works. I think it pans between a high pass filter made of C4/density pot and a shelving high pass filter made of C4/density pot and R6 but im not sure.

4) Coupling caps. Yeah a lot of coupling caps are missing. Ive tried a lot of different values (associated with resistor to ground) between the treble boost stage and LM308 but in the end with higher values the bass content was overwhelming and it sounded too farty, and too low values made the sound lifeless to me. Didnt manage to make it sound as good with a coupling cap than without and cant understand why.

5) design with dual supply. I went with the circuit shown in the LT1054 datasheet, but i dont know if i need and how i need to filter the dc lines. Filter 9V before LT0154 and/or filter V+ and V- after 1054? I ve included 100n cap to ground for each opamp V+ and V- it is overkilled? Do i have to filter V+ and V- with bigger caps? even inductors like ive seen in some designs?

6) General gestion of bass content pre clipping. I was myself surprised to see how much i had to remove bass and boost treble before clipping before enjoying the sound. I know it comes to personal tastes, but how do people more experienced in bass distortions design handle it? In my design, even with the density control at minimum there is a slight bass loss (at maximum bass and mid disappear). I couldnt manage to make bass frequencies clip without sounding muddy and farty. Ive tried the notch filter which can be seen in the darkglass and tech21BDDI. They cut a lot of signal at 320 and 700 (if i remember correctlt) hz. I enjoyed a clean signal with those filters but didnt manage to make them work well with the following distortion.

7) Tonestacks and input/output impedance of gain stages. When i make a high pass filter a send a resistor to ground but it also affects the impedance of the following gain stages. Sometimes i have small value (from 1K to 2,2k) to ground which, in parallel with any other resistance, will set the impedance. So the input impedance of the stage will be no more than 1K. The design mainly has opamp gainstages with low output impedance. Should i ignore the problem and consider with 1k resistor to ground i dont load the signal after an opamp ? Or should i avoid low values resistance to ground?

Thanks in advance for your answers

Here is the circuit drawn on diylc

[marcelomd]

Hi,

This design looks nice =)

I have a G&L 2500. Their pickups are monsters. Even in passive mode they are easily the strongest pickups I've ever seen. They can overdrive almost anything.

3 - The arrangement you used changes between two high pass filters (C2/(R6+Density) and C4/Density) and shelving + high pass as you said. The best way to analyse these filters, for me, is to remember capacitors are short circuits at high frequencies and open at low frequencies. Then look at the extremes of the pots one at a time (do it for every combination of extremes). It is a linear circuit, meaning every "state" is simply "superimposed"(for lack of a better word) to the others. Exactly at which frequencies these extremes blend can be calculated later.

4 - Coupling capacitors are there to address differences in DC levels between stages. At any point in the circuit, the signal is swinging around a bias point. In your case, every pat of the circuit is biased to 0V, so no decoupling is necessary. Someone correct me on this.

5 - I would use a big capacitor+small capacitor at the 9V input (ex. 220u and 100n) and a pair of a not so large + small one for the outputs (say 47u and 100n. The 100n here is probably overkill). A 100n capacitor for each supply pin of each opamp is ok. The circuit will work without all these capacitors, but it is a good habit and doesn't cost too much.

That is all I got for now =)

[PRR]

> Which value is best for the source resistance of the JFET buffer?

Depends entirely on the load it has to drive, limited by JFET's Idss.

Do you know what load(s) you are driving?

[12Bass]

Bit of tangent perhaps...

The Tech 21 dUg Pinnick DP-3X uses a crossover to separate into two bands, with the highs going through a clipping stage and the clean lows through a compressor.  Works very well! 

Might give some ideas.

https://www.youtube.com/watch?v=kkCpA2FAv_U

[Jubz]

Bit of tangent perhaps...

The Tech 21 dUg Pinnick DP-3X uses a crossover to separate into two bands, with the highs going through a clipping stage and the clean lows through a compressor.  Works very well! 

Might give some ideas.

https://www.youtube.com/watch?v=kkCpA2FAv_U

Hi. Sure a cool idea.

In this video the guy get very huge tone using something similar. For the moment im too lazy to try to add a big part of circuitry to the distortion and the few things i have read about compressor circuits were way out of my comprehension. I ve been toying with the idea of using a jfet as variable resistor as a AMZ warp control but it was a disaster. I didnt manage to make a usable envelope detector and figure how to send just the right voltage to the gate of the fate... Dont feel so ready with circuits blocks employed in compressors.

Hi,

This design looks nice =)

I have a G&L 2500. Their pickups are monsters. Even in passive mode they are easily the strongest pickups I've ever seen. They can overdrive almost anything.

3 - The arrangement you used changes between two high pass filters (C2/(R6+Density) and C4/Density) and shelving + high pass as you said. The best way to analyse these filters, for me, is to remember capacitors are short circuits at high frequencies and open at low frequencies. Then look at the extremes of the pots one at a time (do it for every combination of extremes). It is a linear circuit, meaning every "state" is simply "superimposed"(for lack of a better word) to the others. Exactly at which frequencies these extremes blend can be calculated later.

4 - Coupling capacitors are there to address differences in DC levels between stages. At any point in the circuit, the signal is swinging around a bias point. In your case, every pat of the circuit is biased to 0V, so no decoupling is necessary. Someone correct me on this.

5 - I would use a big capacitor+small capacitor at the 9V input (ex. 220u and 100n) and a pair of a not so large + small one for the outputs (say 47u and 100n. The 100n here is probably overkill). A 100n capacitor for each supply pin of each opamp is ok. The circuit will work without all these capacitors, but it is a good habit and doesn't cost too much.

That is all I got for now =)

Mine is a L1505. Love the sound (really bity) but yeah its hard to get effects that can deal with the high output. Sure we could roll down the volume but we bassists just CANT. Seriously i dont even know why basses have volume pots. I will walk through the stage in the dark, stumble on all the cables to roll down the volume on my amp before even starting to think i can do it on my bass too.

If i try to calculate what happens in the first filter with the pot all the way down (the maths for the shelving filter are found here http://www.linkwitzlab.com/filters.htm) :

(10K + R6) // C4 and nothing going to ground. I guess it has no effect on the signal. But at the same time i see a hi pass filter made of C4 and the value of the density pot (10k). So lets say a hi pass with cutoff frequency of 1592 hz (thank you amz rc filter calculator). Im suspecting there is something else happening with R6 and the 10K of the pot but cant figure what.

With the pot all the way up :

R6//C4 and 10K going to ground. So high shelf with fc low = 3386 hz _  fc high = 4978 hz and 3,3db loss between fc low and fc high.

It can explain why i settled by ear on this filter. In my quest of feeding the lm308 with the maximum treble boosted clean signal I wanted to attenuate a little a very large band of frequencies at all time because i didnt manage to boost treble without clipping low mids/mids. They were less boosted than treble but enough to muddy the signal. The fc low and fc high seem anyway very high i will try to lower them.

I also wanted to pan between a classic "fat" rat sound and a chainsaw style distortion. The simplest way i found, rather than putting gyrators everywhere, was to make bass AND mids disappear before clipping. So i guess this filter gives me a classic hi pass too without having to add other components. But Im still thinking im still missing someting the way i look at the schem. And still dont understand what really happens when the pot is not at max or min.

I think too too much filtering is better than too little. On the breadboard its very noisy and i would like to improve it. Im afraid just boxing it wont solve everything.

> Which value is best for the source resistance of the JFET buffer?

Depends entirely on the load it has to drive, limited by JFET's Idss.

Do you know what load(s) you are driving?

Hum... not sure of my answer. I would say I split the signal in 2 paths each going to a following opamp gain stage. So 2 loads in parallel but there is a "third" path made by the shelf filter. I see the signal going, after the 10 u cap

1) R5 to ground > R6 > density pot value to ground which varies > 10K - density pot value which varies > IC2A +input

2) R5 to ground > R7 > C5 to ground > IC1A +input

3) R5 to ground > C4 > 10k (the full density pot resistance) to ground > IC2A +input

The 3) totally messes the way I understand my schematic. Can I simplify by considering it as a part of the "general" load going to IC2A ? Can the input of IC2A see the density pot go to ground twice ? The full 10K via C4 and a value depending on the rotation of the pot via R6? Does the "third" path doesnt exist and I just see it because of the way Ive drawn this part of the schematic and I like to overcomplicate things ?

In every case the input of IC2A, 300k input resistance typical, would be paralleled with 1M//10K (or worse 1M//10K // 10K or less) so I guess the JFET would have to drive a 10K load at best only for this path? This 10K at best would be paralleled with the load seen by IC1A so 1M // very high. Anyway I have the impression the JFET buffer would have to have very low output resistance to drive 10K (and I imagine the reality is far lower).

[marcelomd]

Hi,
Have you tried simulating this circuit? Should help clear things a bit. I think pure math is a bit dry in this case =)

[vigilante397]

The Rat is pretty well my favorite dirt pedal in existence. Have you checked out the Idiotbox Blower Box? It's a really excellent bass Rat with a Baxandall EQ. The also use the JFET at the end as a gainstage instead of a buffer to make up for losses in the EQ. It's a fairly simple build and is my go-to for bass dirt now.

[George Moore]

  12Bass depicted the [split signal HF/LF, mixed again after...] for the LF's compressed, HF's clipped concept.   
  It's like[?] when viewing through a fan, in the right light, at high speeds it appears to rotate C-Clockwise, as the blade spin slows, the reverse rotation slows, stops, re-reverses [creates percieved clockwise] effect, then at further slowing [lower freq.] all you see is clockwise rotation, as it slows more you actually see through between the fan blades a PWM affair..on, off, on, off.
  The best analog I know of to explain how clipped HF's sounding somewhat swirly, Mids not so much and at the bassy end of the range, starts to sound "buzzy". Instead of the 'complex harmonics interacting', overlaps become stretches of linear wave dotted by events of clipping, hence the 'separate' clipping [percieved as buzzy tone] and bass frequency tones. 

[PRR]

> not sure of my answer.

Let's plot it out.

Worst case, all caps are short and all pots turned to least resistance.



Of the four paths, I just ignore the 1Meg because it is likely far higher than the rest.

Path C4 to Density, 10K.

R6 to min Density, 5K. (Count on thumbs first to get a rough idea. Then refine any critical calcs as needed.)

R7 to short C5, 10K.

So we have 2.5K load on the JFET. (2.416k nominal exact.) (And the 1Meg is the flea on the elephant.)

> Which value is best for the source resistance of the JFET buffer?

The JFET pulls up and its source resistor pulls down. With a BJT here we have ample pull-up for any small-audio chore. A JFET needs to be investigated for pull-up ability, which may be marginal, depending what FET is used. I don't see it specified? So For pull-down, the extreme is when the JFET is driven "off". Now we have a simple 5K+2.5K voltage divider to -9V. Clearly the lowest it can swing is -3V.

Now if this was guitar, I'd feel ample with 3V. But I hear bass is bigger. And your pickup may be a monster. It might be interesting to tap signal from the right side of C2 directly to your instrument amp. This should be clean. If it isn't, don't be looking downstream. Either beef-up this buffer (an opamp is the swiftest path) OR turn-down that over-large signal out of your monster axe. There's really no reason to have over 1V signal on stage. Except to be louder than the other guy's (other company's) axe on the same amp settings.

FWIW: another path is to raise these impedances. The opamps they feed are essentially infinite, there's no strong reason to be down in the k-Ohms. 1nFd 47k 100k would give the same response but now the FET+R fights 24k not 2.4k. With the same 4.7k source resistor it could swing 7V peak. BUT IC1A has gain of 2+. If fed the same 9V rails it makes max 8V out so <4V its input. In fact it looks like for some settings IC1A and JFET clip at near the same point. So changing one won't change the overload much at all.

The other path has gain up to 5,500. Even a very soft strum will bang the rails. And the universal hiss will be as loud as a soft strum. I will not listen to complaints about that side clipping. It is supposed to be brutal.

[Jubz]

> not sure of my answer.

Let's plot it out.

Worst case, all caps are short and all pots turned to least resistance.



Of the four paths, I just ignore the 1Meg because it is likely far higher than the rest.

Path C4 to Density, 10K.

R6 to min Density, 5K. (Count on thumbs first to get a rough idea. Then refine any critical calcs as needed.)

R7 to short C5, 10K.

So we have 2.5K load on the JFET. (2.416k nominal exact.) (And the 1Meg is the flea on the elephant.)

> Which value is best for the source resistance of the JFET buffer?

The JFET pulls up and its source resistor pulls down. With a BJT here we have ample pull-up for any small-audio chore. A JFET needs to be investigated for pull-up ability, which may be marginal, depending what FET is used. I don't see it specified? So For pull-down, the extreme is when the JFET is driven "off". Now we have a simple 5K+2.5K voltage divider to -9V. Clearly the lowest it can swing is -3V.

Now if this was guitar, I'd feel ample with 3V. But I hear bass is bigger. And your pickup may be a monster. It might be interesting to tap signal from the right side of C2 directly to your instrument amp. This should be clean. If it isn't, don't be looking downstream. Either beef-up this buffer (an opamp is the swiftest path) OR turn-down that over-large signal out of your monster axe. There's really no reason to have over 1V signal on stage. Except to be louder than the other guy's (other company's) axe on the same amp settings.

FWIW: another path is to raise these impedances. The opamps they feed are essentially infinite, there's no strong reason to be down in the k-Ohms. 1nFd 47k 100k would give the same response but now the FET+R fights 24k not 2.4k. With the same 4.7k source resistor it could swing 7V peak. BUT IC1A has gain of 2+. If fed the same 9V rails it makes max 8V out so <4V its input. In fact it looks like for some settings IC1A and JFET clip at near the same point. So changing one won't change the overload much at all.

The other path has gain up to 5,500. Even a very soft strum will bang the rails. And the universal hiss will be as loud as a soft strum. I will not listen to complaints about that side clipping. It is supposed to be brutal.

Hi

Thanks for your inputs.

Ive added filtering on the supply. I suppose I have to add resistances in serie with the caps to make a filter but dont know how to choose values. Or do the diodes already can act as resistance when they conduct here? The 47 ohm resistance on the +15V is present on the breadboard at the moment. The + and - lines werent symetrical and add this resistance seems to reduce the imbalance.

Ive also raised the values of the first high pass filter and replace the jfet buffer with opamp as suggested by PRR. The main purpose of the buffer is to drive the filter so I feel more comfortable with the opamp because its output is so low I dont have to care about it (and because I havent started to interest in transistors for the moment, just opamps).I abandonned the idea of the low pass filtered and bass boost paralleled signal and just went with the regular clean signal. In fact I prefer the unflitered clean which preserves fret noises and add hi mids attack too.

I didnt really liked the EQ so retried gyrators. This time with bipolar transistor instead of OPAMP. Work quite well on the breadboard but I think it has something wrong and can be improved. I just transfered the mantic vitriol gyrator which is made with opamp and tweaked the capacitor values and Q (thank you AMZ bandpass calculator). But I guess with transistor and +15/-15V supply I have to choose new values with other considerations in mind than just frequency response. I wonder if the values of the whole EQ stage components have to be tweaked when going from opamp to transistor gyrator.

5500 gain was clearly over the top. I tried to lower IC1B gain by swapping R6 to 6,8K and C3 to 15nf (so from 11 to 4 gain) to bring back the total saturated path gain back to 2000 (Proco rat has 2128 gain for frequencies above 1540hz) but after it barely had distortion except with gain maxed, was farty and lacked dynamics which surprised me. So I took to PRR advice (nevermind if it clips at this stage) and re upped IC1B gain to 8 and lower the cut off frequency to 1000 hz. Its far better now for me. The total gain is 3841 which seems really over the top but  I guess I still lose more signal than I think between the input buffer and the LM308. And I saturate more high frequencies (Read on this forum that they need more gain to saturate than low low frequencies).

In fact I originally intended to build a blowerbox (If the schematic was available somewhere I would have tried the Dunwich volt thrower too the youtube demos seem impressive) but once Ive breadboarded the LM308 gain stage and add diodes I started experimenting because it was the most fun for me. At the end, it went its own way and I never finished to breadboard the blower box.

[Gus]

Something I posted in the past. Look up inverting opamp and note what you can change with increasing R1 and/or decrease R15 value
https://www.diystompboxes.com/smfforum/index.php?topic=103110.msg916963#msg916963

[Jubz]

Re.

Several weeks later, finally managed to finish a perfboard layout. I will try to verify it as soon I will have some 100n MLCC. The charge pump is on a daughter board. Its the first layout I make and I tried to follow basic rules Ive read here. Tried to keep high signal traces away from the input stage, divide and isolate building blocks. But I wanted board mounted pot and avoid jumpers so the traces are not the most direct. Obviously, there might tons of others good practices that Ive not followed. Can you see some which can improve the whole stuff?

The layouting really drove me crazy. I will try to learn to use eagle and try to layout a 2 layer board for this circuit.

The schematic has not moved, though the values of the bass-low mid gyrator and output coupling caps (maybe they are a little too big) will probably be tweaked again to avoid too much artificial boominess. Ive added 2 1K resistances which are basically jumpers (yeah no jumpers but ive cheated).

Again thanks for your help here. In a few months I will probably think it sounds like shit and try to completely redo it but in meantime I hope to have some fun with it.

[Jubz]

Ok spot mistake 1. I ve made the whole @#$%ing layout looking at the pots the wrong way...

[rankot]

This looks really promising, do you have any demo of it?

[ElectricDruid]

Ok spot mistake 1. I ve made the whole @#$%ing layout looking at the pots the wrong way...

No! you just soldered the pots to the wrong side of the board is all!

T.

[Jubz]

Yeah. I was uber proud of my no jumper first "big" layout and 1 day after finishing it I start drawing a box template and ... I have no clue of how Ive layouted the pots. Just tried do connect them in the more convenient way on the perfboard not even figuring which way they rotate. So back on the breadboard to verify it :

GAIN, VOLUME, DENSITY are layouted the wrong way. No problem for VOLUME in fact after trying B100K I ve sticked with it. I ve reupped the GAIN pot to 100K (still not enough distortion...) and I think C100K could be use. Same for density. But I dont have C100K pots, it would work but would be ugly,

Lucky me, EDGE and PUNCH were alright.

and while I was on the breadboard I decided the DENSITY control was still not cutting enough bass. So I replaced it with 2 simple high pass in serie connected by 1 pot to ground. The resistances to ground are paralleled 47K + 250K pot value. I decided to cut the big coupling capacitor and 1M resistance after the first buffer, if I understand I dont really need it.

And I ve had some free time so I decided to learn to use eagle and layout the circuit on a double sided PCB. I was totally stuck with the perfboard layout, at this point I didnt manage to move a single component and not finish stuck. Eagle is not very intuitive for me but there is ton of ressource on how it works here. And 2 side layout seems so easy when we already have struggled to perf the layout, so the charge pump is now onboard. But I have some questions on eagle design rules :

1) I went on 50mil traces for power lines, 24mil for signal, seemed a compromise between a lot of values quoted in different thread here. But the clearance is only 12mil between signals and 10 mil between signal and component pad. Do you think the clearance is enough or I should make the trace a bit thinner to up the clearance?

2) Ive tried ground plane ?... heu ground pour ? I dont understand the difference. the only one place someone explains it is the gaussmarkov tutorial but... I didnt understand. I have the vague intuition I should separe signal and power ground, especially with a charge pump. So I have ground planes/pours for GND (signal) and AGND (charge pump and decoupling caps). I should connect them in one point but dont know how to do it, eagle does not let me, and anyway I cant decide which is the best point to connect them.

3) The ground planes/pours design rules seem less documented. I went with 10 mil width and 10mil isolation but I am unsure about values. There are thin strips on grounds on the edges which seem the only way to connect some ground pads.

4) I have another vague intuition I shouldnt connect all the signal grounds in the same plane/pour, but I dont understand enough about grounding to choose which one go with which one. The breadboards experiments tell me that at least all the grounds which are before the LM308, so before any serious high gain occur, should be separated from the others. But anyway, I dont know how to connect several grounds together in eagle.

This is terrible, the more you give me answers the more I come back with new problems to solve. Here the new eagle drawn schem and the great spaghetti monster. If someone interested in building it, keeping in mind that is overly harsh and over the top if you dont play fingerstyle bass on a 15 without tweeter, I will join the complete eagle file when completed.

[Jubz]