feedforward clipper

If Vch and Vcl were low, say +-2 volts, the opamp will go open loop when the transistors saturate and clamp the output. The zeners would then conduct and keep the opamp closed-loop.

The 10 volt values don't make a lot of sense, but some sort of anti-open-loop thing could keep the opamp from going open loop and winding up, which would slow down recovery.

Dang! So it does, thanks. My opinion of Doug Self just dropped.

piglet

Well, that circuit is pretty convoluted. Anything that complex must have something wrong with it.

Yes, the circuit is working fine, I could not find faults in it.. As long as A1 and A3 stay below the forward of the diodes there is no current between the two of them and thus does not effect the gain at A2.

Increase the input enough to where D2 and D4 start clipping the unwanted side, then you'll get current between D1 and D3 and this will cause a gain drop at A2.

R2, R5 and A3 related R's set the gain at which A1 and A3 will place those sections in clipping state verses the input level.

That's the basics of it. Personnally I don't see why a single stage with two back to back zeners won't work? I do that for 10 volt references using 9.2 volt zeners.

That is a fixed type. A variable type can be done using a jfet to clamp the feed via a precision rectifier, which is actually a very good way to do it.

Jamie

That works directly if your op amp works normally from the required thresholds (as power rails). Or you can set enough gain and offset that the input thresholds map to the power rails, then attenuate and bias to get any lesser range. It takes more resistors that way...

I've often wanted a window/overload detector, so comparator-based solutions seem more 'normal' to me. Two comparators, an analog 3-in MUX, it could easily be made into an IC that clips to any pair of levels in the input range.

Also, a pair of (TL431) programmable zeners is a minimum-parts solution that ought not be overlooked. Just series-connect, if you can tolerate the reverse-diode drop uncertainty; for audio, that'd work fine.

There are a couple of "clamp amps" around, opamps that have explicit clamp input pins. The HFA1130 is the classic, but it's a kind of nasty beast otherwise. The AD8036/AD8037 aren't as fast but clamp much better.

Might be some strange recovery time issues since the TL431 is essentially a one-legged OpAmp plus a reference; but it could be worth a try. ...Jim Thompson

-- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | San Tan Valley, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

Can you explain a little more how this would work? I'm a little dense today.

One use for the TL431 that might be kind of cool is for an output-muting circuit in audio equipment...you want to keep the line outputs of some piece of gear clamped to ground at power-on until the filter caps charge and the rails stabilize so you don't get a lot of pops and thumps in downstream equipment.

Maybe I'll play around with that idea if I have some free time.

I can do a screencap of part of a schematic of one synthesizer I have which uses a circuit made from discretes to engage a clamp on the outputs while the rails stabilize. I don't understand exactly how it works, but it looks fairly convoluted and I don't think it could be all that fast.

Bandgap regulators (and zeners, what with capacitance) can be slow.

The TL431 is not simply a zener, it contains gain elements wrapped around a bandgap to make an adjustable reference, thus it has loopgain and recovery time issues. Sending current up thru the substrate junctions if you did these devices "back-to-back" puts the bias in undefined conditions.

I used a Mercury-wetted time-delay relay to disconnect speakers while everything was charging up ;-) ...Jim Thompson

Ah, yes, I actually did know that the TL431 was kind of like an opamp with a reference attached to one of the inputs. In a pinch I've pressed it into service as a filter amp using the multiple feedback topology in a 5 volt system, though the bias current requirements on the input of the really cheap bipolar variety are pretty high compared to an opamp.

Not so RoHS friendly anymore, unfortunately. Haha.

Back in the days when Mercury was good >:-} ...Jim Thompson

The data sheet allows +150 to -100 mA cathode currents, so it's pretty much rated to do forward-diode conduction just like a 'real' Zener. There's a diode clamping the cathode to anode, in the datasheet schematic, for what that's worth, but no indication of any recovery-time from reverse bias.

That diode is the substrate diode.... every NPN has such a diode from substrate (anode) to collector (cathode). What I would be concerned about is the recovery time for the "OpAmp" to settle, once the voltage is in the "right" direction. ...Jim Thompson

hmm, yeah increasing the forwards drop of D2 and D4 would fox that. (eg add a 9.1V zener to each, series opposing)

yeah, That's gotta be it. thanks.

If the reverse (call it -0.5V) bias recovery is an issue, and the TL431 has 2.5V (the minimum) setting, you've got 3V of swing. For audio, at less than 1V/us, there's a few microseconds of recovery time. It'd be possible to look for even harmonics of a clipped sinewave, as an indicator of any settling-time issue.

It's not as simplistic as that... the TL431 _may_ overshoot recovering from such a reverse bias... the compensation cap will take a while to slew from reverse bias. Something that needs to be tested in the lab... the datasheets and Spice models suck. ...Jim Thompson