A slightly weird limiter circuit, any chance of some schematic review?

[dmills]

Hi all,
Designing a de-esser card for my cutting chain, and have come up with a filter/gain cell that is a bit weird, is there any chance someone could give me a second opinion?

Limiterl.PNG

U1A forms a second order MFB LPF at about 2.2kHz, it is inverting.

In the inactive state U2A has a large amount of attenuation (-40dB), meaning it contributes some noise but negligible distortion to the output, only when limiting is its gain increased to 0dB (at full limiting) where it subtracts high frequency signal from the output. This is done to minimise any possible influence of the VCA on the non limiting case (When not limiting the chain is three inverting opamps and a 100nF C0G cap, pretty blameless.

In the non limiting case (VCA effectively out of circuit) the LPF output is subtracted from the input at C9 (Note this is only a first order response because of the phase shift not being accounted for, but that is fine), this is inverted by U3B then summed with the LPF output in U3A giving a full bandwidth output at unity gain and a flat frequency and phase response.

With the VCA fully open (0dB gain, but note that it is both current mode and inverting) it subtracts the HPF signal (made by R6,8) from the HPF signal bypassing the VCA (C9), and these currents cancel, meaning the output of U3B is zero and the final output is just the LPF output (A second order LPF).

There is a clamp on the control amplifier output to ensure that the VCA gain is never driven above unity (which would make the HF level come back up!).

Note the control law is nothing very useful, but this is a feedback design with the control amplifier looking at the output of the IRIAA board that follows this thing, so providing a reasonable amount of loop gain is present (there is plenty) the shape of the control response pretty much does not matter.

There will probably be a second set of this going over at maybe 8kHz or so with a simple differentiator driving the control loop preceding the 2.2kHz job looking at the IRIAA output to chop the really high band down and reduce the amount of limiting needed further down.

Would there be any virtue in feeding the threshold control with the linear velocity by means of a pot reading the carriage position? This would let you cut better HF near the outside of the disk, rather then having to go with a fixed threshold.

Thoughts?

[dubcutter89]

Hey Mr Engineer

Interesting design!
Reminds me of some classic limiters for just that purpose, not with THAT chip but quadrant multipiers and stuff. A split band limiter?
I also once started a very similair thing from a commercial unit having the same architecture (VCA chip for subtraction "gain") just with a notch instead of band response.
The original unit was pretty useless but a a bent up resistor and a cable gave straight access to the VCA section that then could be controlled from wherever you think it makes sense.
Unfortunately never got to the point putting it into the rack and do cuts...

Regarding your design - at a first very quick look I was not able to find drastic errors.
Can't see the "clamp" in the sidechain right now (maybe that's the EC- of the VCA taking care of. The Control Ref and Output 100R connected to ground. or something else...)
I'm more the classic, would strap all refs to gnd and use a diode or so for clamp...

Thanks for interesting reading!

Lukas

[dmills]

The clamp is on the page that deals with the sidechain (Not shown), and I am a big fan of defining exactly where things connect to the awfully nebulous 'ground', my kit does NOT hum (There is no excuse for that shit).

Split band indeed, I am procrastinating between two and three bands, probably do 10kHz, 5kHz, and 2.2kHz (final one closed around the IRIAA card so it picks the LF stuff up properly, the others just differentiators to give the falling 6 and 12dB/octave thresholds), depends a bit on what fits on my default single eurocard.

I considered a jfet or photo jfet limiter, but prefer designs that do not need trimming, something nice about not needing to fart around with tweaking and 'select on test'.

What do you think about the notion of making the threshold track the linear velocity at the cutting diameter? No point in enforcing the same limits at the outside of a 12" 45 RPM, as at the centre of a 33RPM single that I can see, nice thing about doing your own machine is that you can add the required signals and pot on the carriage.

I will post the complete design when I get that far.