ping John Woodgate - "soft clipping" ?

In message , dated Fri,

No, the project folded before I found a good, simple solution.

Thanks.

Did you ever try zener soft clipping ?

Graham

In message , dated Fri, 25 Aug

Yes, but not in depth.

Back in yonder '60s we had some (valve) radio transceivers using some really soggy solid state diodes as "soft clippers" to limit Tx mod. Google for "westector" and you may find some stuff.

In message , dated Fri, 25 Aug 2006, rebel writes

I know, but those are copper-oxide rectifiers, not now obtainable.

In article , rebel wrote: [....]

A bridge rectifier feeding a small MOSFET with its gate hooked to the drain, does a fairly soft clip.

Philips makes several DC controlled gain amplifiers. If you make a window comparator (2 comparators to detect min / max) and an RC filter connected to the gain control input, you have an audio compressor / limiter.

In message , dated Fri, 25 Aug

Indeed; that is no problem IF you can afford it. In my case, the cost aspect was so exceedingly important that the project failed.

I took a look... interesting device. Nice articles about them on

Thanks.

I haven't seen anything useful, either. A simple and accurate non-linear component *should* exist, but I haven't seen anything other than what was in the Jan 3 2005 earlier discussion.

Thanks for your reply, mw

Is a log amp out of the question? ISTR, years and years ago, when I was reading The Radio Amateur's Handbook, and mags like QST, that clipping had been pretty much covered - for voice, it's amazing how much clipping can be used and the voice stays intelligible. And it doesn't even need to be all that soft. And hard clipping can be softened by low-pass filtering it, which always smooths out corners and stuff.

Or, if you've got a critical app, just use an opamp with different feedback resistors and pairs of antiparallel diodes, such that the gain decreases with increased input.

Cheers! Rich

I'm using clipping in an existing product to limit the crest factor of speech because the speaker I am using is a dinky little thing and normal speech (with that high crest factor) sounds very very low on it. By bumping the effective gain on the signal and compressing, I get much better response, albeit with some distortion, but as you say speech can be compressed quite a lot and still remain intelligible. Because I have so much in the way of digital processing on the board already, I moved this issue into the digital domain where it is (in some ways) much simpler to solve.

Cheers

PeteS

In message , dated Sat, 26 Aug 2006, Rich Grise writes

Quite right; I've recently sent a client recordings of speech unclipped and with the gain increased to 15 dB beyond clipping, and no intentional low-pass filtering. The clipped signal sounds different, but you can understand every word. It doesn't work if the amplifier has hysterics with that much over-drive.

I've done it with zeners and various series R's. This kind of technique is actually similar to basic log amp thinking, it's just that the log function only operates above a certain signal level.

It works fine.

Graham

No, a log amp or something similar (the diode clamping) isn't too bad. I was just wondering if John had come across something better. Every approach has its downside, as you can see when you read through the thread that started this thread (150 messages).

Thanks, all.