Cap in parallel with diodes

[Christoper]

Good morning all,

I have a RAT2 that I bought with the intention of modding only to realize that there is very little I would care to change about it. One thing I would like to change is the fact that the cleaner the signal on a RAT the thinner it gets. Ruetz mod changed the character of the distortion too much, but I found a forum post that said a 220 nF cap would thicken up the cleans without changing the character of the distortion too much?

If I do a small/large signal and high/low frequency analysis it would stand to reason that low frequencies are untouched regardless of volume, but the less the diodes are conducting the more treble is getting bled to the ground. Am I understanding this correctly? What other circuits does this idea show up in? Has anyone tried it?

[GGBB]

The 220nF cap acts as a low pass filter in addition to the existing "filter" control. I don't consider that to be a "thicken up" effect but rather "dull down" - might be exactly what you want - I don't know - RATs can be awfully bright at some settings. I would think that if the filter knob doesn't get you what you need then the 200nF won't either, but I have not tried it. Neither will actually restore low end loss which is the real problem.

At lower distortion/gain settings the RAT gets thinner because of the high pass filter design in the gain stage. The reutz mod changes that but rather disruptively as you have discovered. Another - IMO better - option for adding "thickness" is my BB mod: https://www.diystompboxes.com/smfforum/index.php?topic=130988.msg1271759#msg1271759.

[antonis]

Hi & Welcome..

the less the diodes are conducting the more treble is getting bled to the ground.

That happens due to LPF action (op-amp output series resistor & diode pair shunting cap..)

At extreme (diodes fully conducting) parallel cap serves only for clipping top/bottom smoothing..
(its capacitive reactance is much bigger than diodes dynamic resistance..)

edit: I was just about to call for Gord..!!! 

[Christoper]

The 220nF cap acts as a low pass filter in addition to the existing "filter" control. I don't consider that to be a "thicken up" effect but rather "dull down" - might be exactly what you want - I don't know - RATs can be awfully bright at some settings. I would think that if the filter knob doesn't get you what you need then the 200nF won't either, but I have not tried it. Neither will actually restore low end loss which is the real problem.

At lower distortion/gain settings the RAT gets thinner because of the high pass filter design in the gain stage. The reutz mod changes that but rather disruptively as you have discovered. Another - IMO better - option for adding "thickness" is my BB mod: https://www.diystompboxes.com/smfforum/index.php?topic=130988.msg1271759#msg1271759.

So a more subtle HPF into the opamp? Look like just the ticket, I will have to try that out.

While I'm here, what are my options to lower the gain? I never put the dial past noon on my main guitar, but sometimes on my lower output instruments I need to kick it up a little higher.

[GGBB]

While I'm here, what are my options to lower the gain? I never put the dial past noon on my main guitar, but sometimes on my lower output instruments I need to kick it up a little higher.

The distortion pot is 100k log, so noon is somewhere between 10-20k. A "little higher" might be double that. A 50k or 25k pot in place of the 100k (I use 50k myself) will probably work better for you. You can simulate the value with a parallel resistor on the 100k pot (100k - 50k, 33k ~= 25k, or somewhere in between) although that will flatten out the pot taper a bit (no big deal).

[Christoper]

Thank you, I will try your mods and report back

[Mark Hammer]

The RAT uses two "ground legs" to set gain, in conjunction with the pot in the op-amp feedback loop.  One of those ground legs is a 560R/4U7 pair while the other is a 47R/2U2 pair that effectively provides up to 100x more gain for content above 1500hz.  If you would like a decent "clean-ish" setting, without irritating highs that create the thinness you dislike, the easiest solution is to include a toggle to simply lift/cancel the connection between that 47R/2U2 pair and ground.  Break the connection anywhere along that path that is easiest, and you'll have a more restrained gain.  Reconnecting it with the toggle will give you a stock Rat once again.

[Christoper]

While I'm here, what are my options to lower the gain? I never put the dial past noon on my main guitar, but sometimes on my lower output instruments I need to kick it up a little higher.

The distortion pot is 100k log, so noon is somewhere between 10-20k. A "little higher" might be double that. A 50k or 25k pot in place of the 100k (I use 50k myself) will probably work better for you. You can simulate the value with a parallel resistor on the 100k pot (100k - 50k, 33k ~= 25k, or somewhere in between) although that will flatten out the pot taper a bit (no big deal).

If I put a 33k resistor in parallel with the pot, 0 ohm will still be zero ohm and the max setting will be 25kish, middle setting will 8k? I know not to think in terms of numbers with regards to pedal design, but will this spread out the narrow sweet spot the RAT has at at roughly 9 o clock?

[GGBB]

If I put a 33k resistor in parallel with the pot, 0 ohm will still be zero ohm and the max setting will be 25kish,

Yes.

middle setting will 8k? I know not to think in terms of numbers with regards to pedal design, but will this spread out the narrow sweet spot the RAT has at at roughly 9 o clock?

It will spread out the overall range, but a narrow sweet spot - a specific section of the range - might not get spread out as much as 25k vs 100k might suggest because the affect that a parallel resistor has on the audio taper will be to make it behave more linearly overall. The smaller the parallel resistor, the greater the affect. A real 25k pot will work a lot better.

[Christoper]

Any good source for a A25k pot that will fit a RAT pcb? Preferably one the same size as the others

[GGBB]

It might be impossible to find an exact match in A25k. You can probably fit a standard pot with solder lugs and then wire the connections, but the bushing needs to be long enough for the enclosure, nuts, and PCB which is longer than typical. Note that USA and Chinese RATs use different pots and PCBs - match to whatever you have.

Another option is to keep the 100k pot and "scale" the rest of the gain stage. So to get the equivalent of 25k, keep the 100k but change 47R and 560R to 4x that and 2.2uF and 4.7uF to 1/4 that. Therefore 220R, 2.7k, 470nF, 1uF respectively. If you are doing the BB mod adjust those values accordingly. You should also change the 100pF to 22pF but that probably won't make much difference.

[GGBB]

If you would like a decent "clean-ish" setting, without irritating highs that create the thinness you dislike, the easiest solution is to include a toggle to simply lift/cancel the connection between that 47R/2U2 pair and ground.  Break the connection anywhere along that path that is easiest, and you'll have a more restrained gain.

I think this is the same thing as the original Reutz mod. Later versions used a trimpot instead of completely disconnecting the 47R leg, so as to lessen the impact, but it still changes the character of the sound more than some people like. That can depend on how you set up your RAT though. Losing that >1.5kHz boost seems to take away the "RAT" sound.

You can apply the principal of the BB mod to the 47R leg and achieve the same result. Instead of raising the gain of the 560R leg, lower the gain of the 47R leg. But keep the knee frequency the same. So instead of 47R-2.2uF try 100R-1uF or 220R-470nF. This would also help with the gain sweet spot - 220R-470nF will make the 100k pot more like a 25k pot. The primary reason that the BB mod works in the 560R leg rather than the 47R leg is because you can't add components in parallel on the 47R leg with a switch and get it to work, so modding an stock RAT is much easier with the 560R leg. Also - modding the 560R leg doesn't reduce max gain.

[Christoper]

And to clarify, BB mod without a switch is double cap value, half resistor value, and then for 1/4 gain, quarter cap and resistor value?

[GGBB]

Depends on which leg you are modifying. For the 560R leg, you reduce the resistor and increase the cap. For the 47R leg it's the opposite direction - reduce the cap and increase the resistor. The key is to keep the same factor but inverted between resistor and cap. And the factor can be any multiple you like. The switched BB mod factors are 2 and 3. My personal recommendation as a permanent mod to the 560R leg would be 180R and 15uF which is equivalent to about 3. On the 47R leg that would be about 120R and 860nF which reduces the gain pot to the equivalent of about 39k. You could stretch that to 150R-680nF for a more easily available cap value which pushes the gain pot down to 32k equivalent.

[Christoper]

Ah, now I see. I was looking at the schematic wrong. How do you know so much about this circuit? Just about every thread I can Google on this topic on this forum has you imparting quality info.

[GGBB]

LOL. Not sure exactly. The RAT was the first pedal I bought something like 40 years ago (really wish I had kept that one). I guess I'm loyal. When I started into building pedals I was naturally curious about the RAT so I just did a lot of reading about it - mostly here - and accumulated as much as I could.

[Christoper]

I tried a 1 uF / 100R pair for the 2.2 uF / 47R leg, and I think it's just about perfect. Not too much bass, not too little gain, and it still sounds very much like a Rat.

Do you have any other tweaks that you like doing? It seems like most people just mess around with the HPF RC network or the clipping diodes. I'm probably gonna leave it as it is now, but it's always nice to know.

[GGBB]

I'm a fan of LED or MOSFET clippers instead of silicon diodes. The higher threshold means gain has to be a bit higher for the same amount of dirt, which adds a bit more low end and to me seems a little more touch sensitive. Along with that, modify the output JFET bias to the setup used in the Turbo RAT - 2.2M from gate to ground PLUS 2.2M from drain to gate (R10 and R16 below). Although I haven't really heard this myself definitively, I've read multiple user reports that this can reduce/eliminate fartiness/fizz from the JFET being driven to the edge of cutoff by the larger output signal due to the higher clipping threshold (apparently even with silicon diodes in some cases). Because Pro Co deliberately added it to the Turbo RAT, it follows that it is beneficial to the circuit. Also note that in early Turbo RATs which used the same PCB as the RAT2, the additional resistor was tack soldered to the bottom of the PCB, which I take as an indication that it was a very important addition. YMMV - because JFETs vary a great deal, your specific 2N5457 may be fine without it when someone else's is not.

https://www.mediafire.com/file/4pdvsczbyvlusgx/Multi-RAT.png/file

[Christoper]

Just a little circuit theory here real quick.

At maximum gain the stock circuit produces a gain factor of 100,000/(47 || 560) = 2300ish for all frequencies above the 1500 Hz cutoff. All frequencies between 1500 Hz and 60 Hz get boosted at most 100,000 / 560 = around 200 times.

With the 100R 1uF pair I get around 1200 times gain above 1.5kHz and a still at most 200 for the other frequencies.

Doing the logarithms, stock has an 11 dB boost above 1.5kHz and the mod has 8 dB of boost.

So if I've got the math right, my mod boosted the bass by 3 dB and effectively turned my 100k pot into a 50k pot, meaning that I lost the top 1/3rd of the gain control but the bottom 2/3rd is now better spread out.

[PRR]

gain factor of 100,000/(47 || 560) = 2300ish for all frequencies above the 1500 Hz cutoff.

Also the LM308 with stock 30pFd compensation has gain of 1,000 at about 400H, falling to 300 by 1kHz (250@1,500Hz) and 30 at 10kHz.
https://www.analog.com/media/en/technical-documentation/data-sheets/lt0108.pdf  page 4, "Open Loop Freq Response"

So while the gain-set parts may appear to set high gain, it probably doesn't ever hit a "thousand". (Which is still a LOT of gain to apply to a signal.)