Dynamic clipping threshold control with regulator

[teemuk]

If anyone's interested, here's a sketch of an effect I've been cooking up lately. Thought it might come handy for those that are keen in experimenting various effects stuff or just needing inspiration for new ideas.

The circuit is basically a simulation of a tube push-pull output stage with distinctive voltage sag. There are two parts: 1) a clipping circuit emulating the operation of a push-pull gain stage and 2) a voltage-controlled regulator driven by a precision rectifier.

The regulator's output provides a DC reference for the clipping diodes, hence controlling their clipping threshold voltage. The regulator's control samples its driving signal from the guitar signal and rectifies it. When the mean amplitude of the signal increases the regulator is driven to decrease its output voltage.

The clipping stage is basically a gain stage and a phase splitter, followed by two shunt diode clippers that clip the peaks of positive half waves. The signals are then combined with a differential amp. When summed, the two differential signals "correct" each other's errors, thus somewhat softening the clipped portions.

The injection point to voltage-controlled regulator was chosen after the clipping stage so that the sagging effect can reduce when clipping decreases the mean signal amplitude.

There you have it in a nutshell.

The following figures show a crude schematic and a waveform capture of a real audio sample driven through the SPICE engine that demonstrates the circuit's operation.

In the latter, the light and dark grey waveforms show the signals at the anodes of the two clipping diodes. (Aside clipping taking place) they are basically invert images of each other. The medium grey DC signal is the regulator's output and the cathode reference of the clipping diodes. Each time this voltage drops circa 600mV below the anode signals the signal is clipped. One can see that the regulator circuit reacts to high signal amplitudes and gradually decreases the reference voltage, thus the initial picking transient is not clipped (though this depends on amount of "sag" currently taking place). Sustained high amplitude signals will result into clipping that increases in magnitude until the signal's amplitude has become low enough to not drive the regulator to introduce more voltage drop. At lower magnitude signal portions the reference voltage "recovers" allowing initial picking transients to be amplified cleanly again.





I deliberately made the effect quite subtle since I most likely intend to use this circuit only as a little extra "spicing". A dedicated clipping circuit can do the heavy lifting – just like the preamp usually does it in modern hi-gain tube amps. The potentiometer in the regulator control can adjust the circuit's sensitivity. The circuit is fine tuned for an input voltage of approximately 1Vpeak but tweaking that should be easy since the circuits are pretty basic.

[Brymus]

Wow thanx for sharing that Teemuk.
It sounds very promising.

[kikey]

 Interesting!
I have done some work with a overdrive/sag-circuit that is based on the same idea.

I suppose you are 'the' Teemu that wrote the ss-amp book !?
I am really impressed by your work!! 

[teemuk]

I pretty much re-designed the whole thing...
- Ditched the regulator, which was too complex solution for such a simple job. Now the reference voltage for diodes is generated by subtracting the rectified and filtered output signal from a constant DC bias level.
- The input section now has a true differential phase splitter and a negative feedback loop. There's also a filter in the output that simulates tube amp's response to a reactive speaker load.
- The circuit section emulating power tubes has grid clipping emulation and the varying DC offset and clipping levels will now also increase the amount of crossover distortion during overdrive - just like in most push-pull tube amps.



Here's some simulation results of overdriving the amp with sine waves that increase in amplitude by twofold in each plot window: The first set of plots are the circuit's output as it would behave to a resistive load, the second set of plots (labelled "output") are the actual output after the speaker-load emulating filter and you can see how it has a drastic effect on the overall signal. The graphs also illustrate how the clipping threshold becomes lower and the amount of crossover distortion becomes higher as the diode's reference voltage "sags".

[joecool85]

Good to see a project continuing on.  Have you built a prototype yet?

[teemuk]

Not yet, looking forward to it, though.

[J M Fahey]

Very good Teemu, :tu: Thanks for sharing.
Go on and post some MP3s pleez.

[teemuk]

I don't think the ~ 3 - 5 second sample clips I can extract from the SPICE sim really do any justice. You can't really capture the interaction with the circuit & your playing with pre-recorded samples processed by the software. It would make more sense to give the readily built thing (whenever it will be in that state) to some good guitarist to demo, but yeah, I can post.

Also, it's just a "power amp" emulation so it's a much subtle effect in processing things than something like a full-blown preamp. Keep that in mind.

You have to wait a sec though so I can figure out where to host the files...

[teemuk]

Damn I hate these filehosting services... none of the free ones seems to allow direct linking.

Clip #1. http://www.datafilehost.com/download-585ea821.html
This features two different riffs I used as a test signal for the software. First sample is clean and completely unprocessed, basically what I used as the input wave for the software simulation. The second is the same sample processed by the simulation with moderate overdriving and the third one is processed by even more overdrive. Basically, the third setting is already in the gain/overdrive territory that is preferably handled majorily by a good preamp circuit instead of the power amp, leaving the power amp to merely do a bit of soft clipping compression.

Clip #2. http://www.datafilehost.com/download-d0b70eed.html
This features the entire clip #1 run through a cabinet simulator circuit to cut out high-frequency harshness you can hear at higher levels of overdrive in clip #1. Especially in the very last sample. This is closer to what you could expect to hear when listening to an amp played through an actual speaker cab.

As mentioned previously, these clips are just the result of running some pre-recorded test clips through a power amp sim so don't expect it to sound like an entire amp doing it's thing. The tone is somewhat buzzy and harsh etc. because the circuit deliberately has a very wide frequency response in comparison to most preamp designs and there is no drastic low- or high frequency cutting - nor frequency peaking or notching - going on to voice the overall tone - specifically an overdriven one.

The "sweet spot" of this circuit is at the moderate overdrive levels where most of the clipping is clearly audible (like in generic tube amps) but not obnoxious sounding. So it's sort of a "harmonically rich" limiter circuit with a bit unlinear frequency response and not much anything else. Basically that effect is best illustrated by a screen cap of wave editor showing the first half of clip #1 with it's tree samples: The second one is still almost clean sounding but you can clearly see how compressed it is in comparison to the (first) non-processed sample. You can also see that despite the clipping the signal still retains a lot of its original dynamics when it comes to initial note attack.

[J M Fahey]

Hi Teemu, thanks for posting.
The effect is noticeable, although subtle as you said.
Agree that it is not a "distortion pedal" but an element to get somewhat closer to classic amps.
I think that using it as the front stage of a power amp, driving a real guitar speaker and miking it will show its true colors.
Thanks.

[teemuk]

I'm planning to put this thing to a preamp that can also do a complete amp + speakers simulation. Something that I could use while practicing with headphones on, or just throw in to drive a power amp if needed (I don't currently have a band nor gig so I have more use for headphone setups for practicing and recording than for high-power loud amps with big speaker systems).

But come to think of it, just alone something like this might be an ideal effect to throw in to the effects loop in between the pre and PA. In many designs you could even control the amount of overdrive with master volume control because that one is usually before the loop. For this kind of application the design likely needs a gain recovery stage with a trimmer adjustment for fine tuning the output signal level to match the input sensitivity of the used power amp, though. (In current design, everything above approximately 200 mVpeak is starting to clip though the output voltage may still swing to approximately 800 mVpeak).

Needless to say, this kind of effect will naturally tax on the headroom and clean output power. Quite considerably too. Good news is, if you don't mind "dirty cleans" you'll make it up in apparent loudness. And it can always be switched off too.

[MR_STENCH]

Hi, teemuk! Thanks for sharing. It's realy great idea. Please excuse my bad English...I think there is one thing in schematic which can be improved. In that design... "saging" (bias shifting) hapens all the time... i mean from very low levels... practically from zero. But, in real tube amps it occurs when Grid starts to draw curent at high positve levels and charging coupling capacitor, shifting Bias voltage that way. I can't explain it good with my poor language  but i'm tniking for litle corecton in rectifier... something like this will be much more realistic:



The voltage drop at R6, R11 is smal but it can be compensate with more gain in V5 inverting stage. 

[phatt]

I have done some work with a overdrive/sag-circuit that is based on the same idea.

Love yah workk, anymore ugg schematics with cold cathode bias or something really out there would be most interested in the incredadbull saggy Ugg circuit.
Phil.

[J M Fahey]

Hay Pilll Y olzo loik UGG chwemoiytiks,
Have you seen our new live snakeskin boots?
TRENDY !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!  :lmao:

[phatt]

Arrh tis Good,,,
Teemu's *Dynamic Threshold Control Regulator* seems to to capable of catching out any kind of short circuit.  (Big Cheesey Grin) :tu:
Phil.