Decompression Overdrive - another approach to diode clipping

[samhay]

Preamble: I came across Mictester's 'Special Sauce' clipper recently on the other forum and it got me thinking. His design uses 1 pair of diodes to forward bias a second pair of diodes to change the way they clip the signal (to ground). Cool idea methinks and it sounded pretty good on the breadboard. Not content to stop there, I started playing with the idea of changing the behaviour of clipping diodes. This is what fell out. As there is nothing new under the sun, I expect someone has thought of it before. Either way, it is worth playing with, if this is your sort of thing, as it certainly can sound quite different to straight clipping diodes.

Compression Overdrive: We have two op-amp gain stages with a clipping stage in between. Nothing special about the op-amp stages - band pass filtering and gain control on the first stage, which also rolls off a little treble at high gain, and a little make-up gain on the second stage, which is also a bandpass filter. These can be tuned to taste and my tastes lean a little to the dark. I have tried TL072 and NE5532 op-amps here and pretty much any dual op-amp should work and have minimal effect on the sound as long as the gain of the first stage is kept low enough to prevent the op-amp from clipping.

The clipping stage uses a pair of diodes that clip to separate capacitors. As the caps charge up, the DC on them reverse biases the diodes, stopping them from conducting. The result - as the note rings the clipping gets progressively reduced and the signal cleans up and starts to ring through.
The 'decompression' pot, which gives this effect its tentative name, allows the DC on the caps to bleed away. Higher resistance values prevent this bleed off and increases the DC on the caps as well as their 'attack' and 'decay' (cap charging/discharging) times. At minimal resistance, you essentially have the same effect as clipping to ground - a compressed signal with peak-to-peak voltage determined by the diode's forward voltage drop. As I used Ge diodes (OA91s), this causes the signal to get very compressed and a bit fizzy. It is a DEcompression pot as I don't have any reverse log taper 1M pots, so a log taper pot will allow you to increase this resistance and decompress the signal as you turn up the pot.
The sym/asym switch switches out one of Ge diodes for a 1n4148 to give it a asymmetrical clipping and is worth experimenting with.

This is still on the breadboard. It should fit in a 1590A if you are foolhardy and I have a vero layout that I will try to verify in the next few days. Once it is built, I will try and cobble some sound clips together.



Edit: typos fixed

[midwayfair]

Another one of yours I need to build.

Mictester's original circuit, as I recall, had some issues going between different guitars and signal levels. Any of that here?

[samhay]

Jon - this was a little procrastination design while trying to get my head around other things like your tremolo.
Mictester's Special Sauce has a fixed gain stage before the clipping section that could quite easily be overdriven. While this is an easy fix, this circuit doesn't do that as the first gain stage is variable and the make-up gain quite modest. If the op-amps are clipping, I can't hear it, and I have tried overdriving it with a reasonably hot humbucker. Of course, if you put a booster in front of it, anything is possible. However, even in this case, the clipping stage should then take the edge of any op-amp clipping anyway.

[midwayfair]

Jon - this was a little procrastination design while trying to get my head around other things like your tremolo.
Mictester's Special Sauce has a fixed gain stage before the clipping section that could quite easily be overdriven. While this is an easy fix, this circuit doesn't do that as the first gain stage is variable and the make-up gain quite modest. If the op-amps are clipping, I can't hear it, and I have tried overdriving it with a reasonably hot humbucker. Of course, if you put a booster in front of it, anything is possible. However, even in this case, the clipping stage should then take the edge of any op-amp clipping anyway.

Thanks. Do you think there's a minimal gain required to get the effect? I'm wondering if I could shoehorn your idea into a transistor/diode combination like the Harmonic percollator's diode clipping section. This is a good idea with lots of different possible uses if it's "moveable."

[samhay]

Thanks. Do you think there's a minimal gain required to get the effect? I'm wondering if I could shoehorn your idea into a transistor/diode combination like the Harmonic percollator's diode clipping section. This is a good idea with lots of different possible uses if it's "moveable."

I am using a slightly improvised gain pot at the moment, but you get effect with most of the travel, so I would guess you could set and forget the gain quite comfortably - particularly if you use Ge diodes, which should significantly clip with pretty minimal gain. This was my initial intention before mission creep set in.
More generally, the clipping stage should be quite transplantable and, I would imagine, doesn't have to be driven between op-amp stages - I know how you have a penchant for discrete transistors.

Edit: fixed another typo and had a look at the Harmonic Percolator schem - this clipping stage will work, but may need a buffer after it. Will have to give it a try the next time I get a quiet moment.

[kurtlives]

Cool idea, I will have to try this on the breadboard. I would think the idea is transferable to other hard clipping diode circuits but it will be interesting to see.

As for the gain stages they are pretty standard non-inverting setups, no? The first op-amp has a gain of 11 (21dB) while the second op-amp has a gain of 2 (6dB).

[samhay]

Cool idea, I will have to try this on the breadboard. I would think the idea is transferable to other hard clipping diode circuits but it will be interesting to see.

As for the gain stages they are pretty standard non-inverting setups, no? The first op-amp has a gain of 11 (21dB) while the second op-amp has a gain of 2 (6dB).

Thanks and yes - there is nothing special about the gain stages and you should be able to modify/replace them without any major issue. The first stage has a gain of 1-11 and the second a gain of 2. I would not suggest you use much higher gain on the second stage (if you use one) as it will clip with big signals when the decompression pot is maxed.
The clipping stage should be drop-in replacement for other clipping stages that have diodes to ground. Make sure you keep a series resistor (like R4) with it though. This will require some testing though.

Edit - first gain stage has a minimum gain of 1

[kurtlives]

Cool, I was thinking of maybe trying this in a Rat/Dist+/DOD-250 or something really simple just so I can hear how well the control works.

What's R4 doing exactly, little bit of current limiting/high end roll off?

[slacker]

Clever idea, interesting to see where it goes.
Won't it only work if the signal is DC coupled to the previous stage, like in your circuit? So if you're dropping it into existing designs you might have to remove any coupling cap, replacing it with one after the clipping stage depending on what follows it.

[electrosonic]

Looks cool I don't have much to add but I think C5 is in backwards.

Andrew.

[MR COFFEE]

Electrosonic is right... C5 is backwards.

Interesting idea - thanks for sharing.

mr coffee

[GFR]

Something I simulated a while ago (just the clipping bit):


picture upload


photo hosting

(The multiple signal sources are just to generate bursts).

I din't try it on the breadboard, tough.

Kind of the opposite of your decompression: as the caps charge, it clips harder. So the headroom decreases after the attack. The idea was to emulate an amplifier with a sagging power supply.

[samhay]

Thanks for the kind words everyone.
I have a couple of quiet evenings coming up, so will try and get this thing built. I am now thinking I will do away with the diode switch and replace it with a pot in series with one of the diodes. This will give the option of asymmetric clipping (like in the Harmonic Percolator etc) that can be fine tuned and, more importantly, will make life a little easier when I try and cram it in a 1590A.

Cool, I was thinking of maybe trying this in a Rat/Dist+/DOD-250 or something really simple just so I can hear how well the control works.

What's R4 doing exactly, little bit of current limiting/high end roll off?

R4 is a current limiter and determines how much current flows through the diodes and thus how quickly the caps charge. I used 10k so that even fairly low-current op-amps will play nice. It is a good place to start tweaking.
It is also, I guess, a low-pass filter with the two 10uF caps. However, this would have a corner frequency of ~1 Hz (hence why they charge up with an audio signal) and I don't think you will have much joy using it as a tone control.

Clever idea, interesting to see where it goes.
Won't it only work if the signal is DC coupled to the previous stage, like in your circuit? So if you're dropping it into existing designs you might have to remove any coupling cap, replacing it with one after the clipping stage depending on what follows it.

I think it should work (not tested yet) with a signal swinging either side of ground. In this case, the polarity of C5 will have to be reversed. The input impedance of whatever it is driving should also ideally be 10x or more the value of R4.

Looks cool I don't have much to add but I think C5 is in backwards.

Andrew.

Electrosonic is right... C5 is backwards.

Interesting idea - thanks for sharing.

mr coffee

Nope - in this design, as Slacker mentioned, the two op-amps are DC coupled so the signal is swinging either side of Vb (~4.5V). If C5 is increased to 1000's uF AND you can sustain a note for minutes-hours then you MIGHT be able to pull the DC on this cap below ground. If this is likely to be a problem then you can decrease the values of C4 and C5 and/or use non-polarised caps.


If the clipping stage (including R4 or equivalent) is placed after a decoupling cap (so the signal is swinging either side of 0V) then the polarity of C5 needs to be reversed. In this case, it should be a drop-in replacement for the clipping diodes in the Harmonic Percolator or DOD-250. I'm not sure how nice it will place with stock Rat of Dist+ circuits due to the tone controls in the Rat and mix knob in the Dist+. As slacker mentioned, one option would be to put this before the output cap. Either way, I guess some experimentation will be required.
One thing to try to make it more easily transplantable is to reduce R4 to e.g 1k. However, you will then have to increase the values of C4 and C5 to get comparable charge/discharge times and thus decompression but these should be tuned to taste anyway.

[samhay]

Something I simulated a while ago (just the clipping bit):

Cool idea. Also take a look at Mictester's 'Special Sauce' design (you will have to go looking on the other forum) for something similar to mine that is designed, I think, to clip more as the caps charge up.

[kurtlives]

Tried this last night.

Only changes were 4558 op-amp, silicon clipping diodes and a linear taper decompression pot.

I couldn't really hear the decompression pot doing anything. I played for a while at different gain settings (mostly high gain to hear the diodes in action) and I varied the decompression control and like I said didn't notice any difference.

Maybe I need to play with that 10K resistor and the two 10uF caps, mess around with the time constants there.

[samhay]

Tried this last night.

Only changes were 4558 op-amp, silicon clipping diodes and a linear taper decompression pot.

I couldn't really hear the decompression pot doing anything. I played for a while at different gain settings (mostly high gain to hear the diodes in action) and I varied the decompression control and like I said didn't notice any difference.

Maybe I need to play with that 10K resistor and the two 10uF caps, mess around with the time constants there.

Thanks for giving it a try.
Most of the action of the decompression pot happens at fairly low resistance and you should hear a difference going from full CCW (no resistance) to maybe 10k or so of resistance - it is a bit twitchy and a log pot is definitely recommended here. The other thing you can try is removing the pot and then alternatively adding/removing a jumper in its place - you are going from no to infinite resistance and should hear something. I was initially going to put a SPST switch here but seem to have become infatuated with pots.
Actually - measure the resistance of your pot when it is hard CCW. If it is not very close to 0R then that might be the problem.

I haven't tried it with a 4558 or two Si diodes but can do so tonight. I take it you could hear the diodes clipping so I guess it must be able to drive enough current to do something. I would try playing with the value of R4 - don't be conservative and try 1k and 100k perhaps.
One other thing to try is to compare it to the same diodes clipping to Vb. It should sound a fair bit softer going through the caps to ground. if not then I guess I need to have a think and do some more testing.

[samhay]

I have now tested it with 4558, TL072 and NE5532 op-amps and a pair of Si (1n4148) diodes. In all cases, it works as intended, is louder, and sounds pretty good. Ge diodes are definitely not required and Si may well suit some people better.

I have now built one of these things on vero (but not boxed yet) using an NE5532 and OA91s.
Among a few tweaks I have bumped the gain up quite a bit by dropping R3 from 10k to 2.2k. This ups the max gain from 11 to 46. The op-amp will likely be clipping well before we get to max gain, but the clipping stages does a nice job of smoothing this out.
I managed to get C4 in backwards while building it. Apart from sounding awful, this prevented the decompression pot from working. Getting C5 the wrong way round will have a similar effect.
The asymmetrical clipping pot (A100k pot in series with one of the diodes) also works nicely. There is a bit of interaction with the decompression pot, but they both serve their purpose.

The vero layout and an updated schematic will come in the morning.

[kurtlives]

I have now tested it with 4558, TL072 and NE5532 op-amps and a pair of Si (1n4148) diodes. In all cases, it works as intended, is louder, and sounds pretty good. Ge diodes are definitely not required and Si may well suit some people better.

I have now built one of these things on vero (but not boxed yet) using an NE5532 and OA91s.
Among a few tweaks I have bumped the gain up quite a bit by dropping R3 from 10k to 2.2k. This ups the max gain from 11 to 46. The op-amp will likely be clipping well before we get to max gain, but the clipping stages does a nice job of smoothing this out.
I managed to get C4 in backwards while building it. Apart from sounding awful, this prevented the decompression pot from working. Getting C5 the wrong way round will have a similar effect.
The asymmetrical clipping pot (A100k pot in series with one of the diodes) also works nicely. There is a bit of interaction with the decompression pot, but they both serve their purpose.

The vero layout and an updated schematic will come in the morning.

Cool findings. I didn't get to play with the circuit on the bread board anymore tonight but I am anxious to experiment some more.

With the gain adjusted to 46 you're definitely going to get that first stage clipping. Were using pretty standard op-amps here that sure aren't able to swing from rail to rail. Still I think with the increased gain it's good because then the clipping diodes are having more effect (well when the op-amp isn't clipping).

[samhay]

With the gain adjusted to 46 you're definitely going to get that first stage clipping. Were using pretty standard op-amps here that sure aren't able to swing from rail to rail. Still I think with the increased gain it's good because then the clipping diodes are having more effect (well when the op-amp isn't clipping).

I set the max gain using my fairly weak bridge single coil. I stopped dropping R3 when I could get this to clip the second op-amp stage with the diodes pulled. You can only really hear this clipping when you pull the clipping diodes. Normally the diodes will smooth off the ugly clipping the first op-amp does and, as long as the 2nd op-amp is not clipping (hence why mine only has a gain of 2) then you are all good.

Fore reference, here the approx max gain for some similar designs (ignoring the freq response of various caps & assuming I found the correct schematics):
ProCo Rat:  100
Tube Screamer:  116 (without clipping diodes)   
MXR Distortion+:  214
DOD 250:  214 (same as Dist+)
   
My max gain of 46 is actually quite modest, but then I was trying for a gentle overdrive rather than a nasty distortion.
Of course, the gain is variable and you also don't have to have it dimed all the time.

[samhay]

The schematic in the first post has been updated.

A build doc can be found here: https://dl.dropbox.com/u/11996927/DC_Overdrive_Build_Doc.pdf (4.5M)
This has a BOM and a verified vero layout. The layout will fit in a 1590A, so it is a bit cramped.