RAT Mod - Slight Return

[WGTP]

For all you Rat modders, I was simulating the dual RC network that comes of the feedback loop and discovered that basically the 4.7uf/560 part boosts the bass around 5db at 100Hz while boosting only 1db at 1000Hz.

So, using the existing parts, you can:

1.  using a switch disconnect the 4.7uf/560 for less bass and more smoothness thru the clipping section

2.  leave it stock, or

3.  add a switch for a FAT mod.  By moving the 4.7uf cap to the 47ohm resistor with the 2.2uf cap, you now have a 6.9uf cap with the 47ohm resistor.  This doesn't boost the highs much, but jacks up the mids and bass as much as 6db.  Dual Turbo Rat.

I have only simmed it, and it may start squealing as Rats do, but I think it will work.   

[MartyMart]

Interesting WGTP, I never "got" that dual path process, infact I removed one on my clone
so I think that I only have a 2u2 and 47 Ohm path now !!
Do you really get "two" sounds from having both networks connected to the same node ?

...... still dont get it 

MM.

[Traintrack]

Is this what you are talking about?

http://www.diyguitaramp.com/rat.html

[WGTP]

That explains it nicely.  What I was doing was switching out the 560/4.7uf network instead of the 47/2.2uf network. 

The 47/2.2uf has more gain and rolls off the bass around 1500Hz.

The 560/4.7uf has less gain and rolls off the bass around edit: (stupid decimal points) this should be 60Hz not 100HZ.  When added to the other RC network, this acts as a bass boost which is unusual prior to the clipping.

The 2 combined has the most gain, but not by much, with the extra coming mostly in the bass.  It sort of creates a very mild dip in the low midrange.

The FAT mod adds the 4.7uf to the 2.2uf for 6.9uf to the 47 resistor for more boost in the mids, as well as bass.  The roll off drops to around 500Hz.

I guess you could also add the 2.2uf cap to the 4.7uf cap for 6.9uf to the 560 resistor, but that is adding mostly bass lower than a typical guitar.  Might work for dropped D or C tuning.

The mod in the article would probably have more bass that I would prefer, but might work well with single coils.

Come to think of it, you could switch the 2 caps and resistors around as well, for different curves.

Mark Hammer proposed using a small value pot, like 500 to pan between the 2 and 1/2 the standard resistors IIRC.

With the parts that are there, all you need is some switching.   

[WGTP]

I messed with this some more last night and you can definitly hear the gain jump up on the bass when playing a low E by adding the second "leg".

My simulations (which may be subject to error) show that the relationship betwee the 1st leg - 47r/2.2uf and the second leg 560r/4.7uf produces about the max amount of bass increase that can be had without boosting the highs as well.  It appear the values used in the 2 legs can be converted to different values for more or less gain, as long as the ratios stay the same.  The same response curve can be maintained.

For example, I used: 1st leg 1K/.1uf - second leg 12K/.22uf.  The resistance of the second leg is 12 times the first leg resistance, and the capacitance is 2.13 times the first leg capacitance.

The debate over the poor slew rate effect of the LM308 giving the Rat a distinctive tone should not omit the additional "knarliness" added by the increased bass thru the op amp (which I think clips a lot) and the clipping stage.   Maybe the slew rate effects the highs and the 2 leg RC network effects the lows. 

[PenPen]

WGTP, if you are still fooling with simulating this, could you do me a favor? I'm curious as to the response curve of the tweaked Rat I made, compared to the stock unit. I have been considering one more mod, but I'm unsure as to how it will affect the overall response.

Values, I used a single RC network to ground in the feedback network, 220Ohm and 10uF. I also changed the bypass cap to 1nF (0.001uF). Additionally, and this may make some difference, I changed the compensation to a 100pF cap to ground from pin 8, which is supposed to limit the bandwidth. The stock compensation was far too buzzy for my tastes. Would you mind simulating this setup for me? How does it compare to the stock values? The mod I've been considering is changing the resistor to 470Ohm, to reduce some of the gain. Will this change the overall response, and how much?

[WGTP]

I am using a modified version of the James Duncan Amp Tonestack.  I tweaked the impedance values until it acted like the standard RC formula said it should.  I'm not sophisticated enough to "sim" the slew rate mod you have made with the 100pF cap.

I think the RC values are in the ballpark.  The 22or/10uF network produces roll off at around 70Hz and 3db more bass at 100Hz than the stock circuit.  In the treble however, it produces around 20db less creating a flat response from 100Hz to 10KHz.   Adding in the .001uF further reduces the treble (in the 3KHz area with the 100K pot at max) another 5db.  It should be less buzzy for sure.

Trading the 470 for the 220 reduces gain 6.5db in the mid-range and produces a roll off of 35Hz.  This is similar to the mod linked above.  Might hear slightly more deep bass in relation to the mid-range.

Jack Orman used a 100r/1uF circuit in his AMZ Distortion Pro that produces a roll off of around 1.6KHz (same as the Rat without the second leg) which may be a little bass shy without post distortion EQ, but produces a nice smooth distortion. 

IIHO a lot of bass thru the distortion can make it sound more like a Fuzz.  Hope that helps.   

[Mark Hammer]

Note that all this discussion centers around what parts of the spectrum are exceeding the clipping threshold more, not the tone EQ per se.

In this sense, the dual path arrangement of the Rat is a distant cousin of the TS-9/808 in that both try to tailor how much gain is applied to certain parts of the spectrum (hence how much distortion and additional harmonic content comes from that input content).  The TS-9 rolls off bass at the input and applies uniform gain across the spectrum resulting in some content being closer to clipping than other content.  The Rat lets everything in, but applies more gain to one range of the audio spectrum than to another.

Seems to me we can productively transfer that principle from the Rat back over to the TS-9.  That is, why not let all bass into and out of the TS-9 but use dual paths to ground to provide range-adjusted gain?

Note as well, that the harmonic content added by clipping is separate from whatever filtering might be superimposed n that to play up or play down parts of the "new modified" spectrum.  I suspect that variations on the simply mid-scoop filter of the Univox Superfuzz would make for some excellent Rat sounds.

One needs to think of this in terms of additive vs subtractive synthesis.  In additive synthesis, you build the tone of the note by adding different proportions of harmonics to the fundamental.  the old Yamaha DX-7 was a good example of this.  In subtractive synthesis, you start with a harmonically rich waveform and carve away parts of it with selective filtering.  The differential clipping of the Rat is a form of additive synthesis in which the dual-path network adds more harmonic content from these fundamentals than from those.  The filter control, or any added mid-scoop now carve away or subtract from the harmonic content one has in order to achieve different tones.  Of course, with subtractive synthesis, it becomes more interesting and "sculp-tive" (is that a word?) the richer the the signal to be processed is.  That's but one of the many reasons why most analog synths would include features tomake those waveforms more interesting.

[PenPen]

Thanks for the info. I'm glad you could confirm what I heard, I could only imagine what the curve looked like (I'm a visual person) and that helps a lot. Basically, the response was exactly what I was trying to do, when I began I planned to use a modified BMP tone control, and I wanted to get a somewhat flat response curve from the chip so that all of the frequencies would be a there to be shaped. However, by the time I GOT that somewhat flat response, I loved it too much to put in a tone stack, so I'm leaving the control out, or at least I'm going to make it switchable so that I can remove the tone stack. Currently it is tone-stackless. Turned out to be a perfect compliment to my stock Rat, which boosted the highs.

I like to use the Rat at less than half gain, and having both on at once at this setting added 'balls' to the overdrive into distortion tone, making it very useful.

Thanks again for running that. Can I ask where/how you modified the DTSC to allow custom simulations? I've been wanting to do this to test out various tone filters and controls I've thought up. I'd love to be able to SEE what I came up with.

EDIT: Mark, very good point. I came to that realization a while back before starting my clone. You guys hear me talk about Rats a lot because I'm a big fan of it, and its the one circuit I've studied over and over until I *THINK* I understand how it operates completely. The facinating thing to me is exactly what you described, the fact that nothing is exactly removed from the signal, just certain parts are boosted over others. There's a lot of play and sounds to squeeze out of that circuit just by changing the response-boost curves.

[WGTP]

Mark, good point, I should have labled it "pre-clipping EQ" and of course that threshold will vary with the clippers used.

The Rat tone control reduces the treble at 6db/oct. and that is not factored in here either.

(Warning, I'm not really sure about all this, but it seems to match the calculations typically done for RC networks)  Here is what I'm doing to sim the frequency response of the stock Rat dual legs (parallel RC networks) off of the feedback loop.  Using the Duncan Tonestack Calculator, the JAMES tonestack with the following values:

Z=.001ohm
R1=47r leg 1 resistor
R2=10M BASS turned all the way down
R3=10M
R4=1r
R5=1r
R6=560r leg 2 resistor TREBLE turned all the way down
C1=2.2uf leg 1 cap
C2=1pf
C3=4.7uf leg 2 cap (make this value 1pf to just look at a single leg)
C4=10uf  this approximates the 100pf bycass cap, but with the impedance value used requires it to be multiplied by a factor of 100,000.  I'm not sure about the interaction between the high pass and low pass networks, so this is the one I'm least confident in.

Set it to Auto Y Scale

Setting caps to 1pf stops the circuit at that point.  The same with 10M.

Yes, I very much understand the Visual thing.

Please let me know if I'm all wrong about this deal, but it is fun to play with.   

[WGTP]

Edit:  Well, I was messing with this again on the breadboard (with a modded BMP tonestack) and found something else cool you can do.  I think of this as a "Character" adjustment, since it effects the harmonic levels generated by the distortion, where as post distortion EQ effects the levels of those harmonics already present.

Replace the 560 ohm resistor with a 2K pot (probably log, but I'm not sure which way)

With the pot at:

2K - the 4.7uf cap is pretty much out of the circuit, so you get 3db less bass

1K - the bass is increasing

560 - stock Rat setting

250 - 3db more bass/2db more treble than stock

125 - 4db more bass/6db more treble than stock

62 - 5db more bass/9 db more treble than stock

22 -5db more bass (overall increase of 10db)/10db more treble than stock (overall 10db gain)

Of course you can use a fixed resistor if you choose.

Summary:  as the resistance drops, first the bass increases, then both bass and treble increase, then the treble increases and the EQ is close to the original, but with 10db more gain.  (as if the Rat needs that).  Inbetween the stock and +10db settings are values that result in a different bass/treble balance.

For a more reasonable gain level, multiply the resistors by a factor of 10 and divide the caps by a factor of 10.  This should also work on Distortion+ and TS's with a Pot value ranging from 100K to 500K for the feedback loop.