distortion cancellation for diode limiter

I've been modeling a diode hard limiter as part of a circuit that I describe in another thread:
http://www.diyaudio.com/forums/subwoofers/229141-lt-gain-dependent-boost-limiter-new-idea.html

The diode limiter is a last resort stopgap in case the other circuit is implementing a "soft" response and the user wants to prevent excess voltage being sent on down to the amplifier.

The problem with hard limiting that uses diode clipping is that the clipping squares off the tops of the waveform, and that introduces lots of harmonics as distortion products. While looking into this kind of thing recently, I happened to stumble upon a technique that uses a differential amp to compare the signal before and after clipping. The output from the differential amp represents the "error" e.g. the induced distortion. This in fed back out of phase to the clipped signal to cancel the distortion components.

The circuit is briefly described in the text below Figure 11 near the bottom of this web page:
Limiters & clipers

In the description the differential signal is low-pass filtered, but it seems that the distortion from a squared-off waveform will be mostly made of up higher frequencies. It's not clear to me why the LP filtering is needed/used, but I haven't modeled it and maybe that would become readily apparent.

Does anyone have experience with this approach? How effective is it in practice?

-Charlie
 
That circuit does the opposite of what you want: it undoes the clipping (by adding back the removed LF signal), but retains the distortion!

I would treat that web page with some caution. It appears to claim (under fig 3) that germanium diodes clip with more even-order distortion. Not true: any balanced clipper will produce only odd-order distortion whatever type of diode is used. An unbalanced clipper will add even-order too, whatever type of diode is used.
 

jcx

Member
2003-02-17 7:38 pm
..
soft limiting means that the compression distortion starts at a lower signal level

maybe it is acceptable in some situations - like when for some reason you simply can't have the required headroom

but if you just want a simple analog clipper to prevent amplifier output stage clipping recovery issues then you still have the option of using enough power, high enough speaker sensitivity to keep the clipper from being audibly obnoxious

Self, Cordell appear to actually want to reduce lower level distortion vs a simple diode clipper with their input signal clipper circuits - which means "sharpening" the clipping characteristic

Geddes may go the other way - doesn't worry much about high SPL distortion in loudspeakers - so presumably not in amps either
 
Last edited:
soft limiting means that the compression distortion starts at a lower signal level

maybe it is acceptable in some situations - like when for some reason you simply can't have the required headroom

but if you just want a simple analog clipper to prevent amplifier output stage clipping recovery issues then you still have the option of using enough power, high enough speaker sensitivity to keep the clipper from being audibly obnoxious

Using a higher amplifier power is not really an option in this case. This is for a Linkwitz Transform type application, with high levels of boost (e.g. 24+dB) already applied to the lowest bass using very high power levels (e.g. 500-1000W).

-Charlie