eti-446 limiter/compressor

[duck_arse]

attached is the original circuit for the eti446. it was presented as novel at the time because of its use of the N mosfets in the 4049UBE as the level control element(s). it also features low distortion due to the gate voltage being modulated by the signal, via C2. this is claimed to reduce distortion from 3.5% to 0.8%.



(I dunno if any images will work, but .....)

edit: they didn't

originally intended for stereo work, its specs are:

freq: 20Hz - 20kHz +-3dB, input range: 1mV - 10V, max gain: R2=4k7- 26dB, R2=47k- 40dB, max attenuation: 55dB.

I've breadboarded this, and watched it working with a signal generator, but I haven't listened to it. (I can't hear compressors work, as far as I can tell).

[duck_arse]

and here is a slightly modernised version. it may contain errors, I'm hoping one of the more qualified types might pass by and give it a once over. I have breadboarded this version, and watched it working, but again not listened to it. it is still split supplies cause I'm not sure of what to tie where, real ground and Vref wise. I stuck the input buffer on because it seems like a good idea.



(..... back to imageshack I suppose)

the original article notes that only the 4049UB will work because of its input protection. I don't have a 4069 to try, but it looks like it *might* work, if wired proper. I also tried the 4007, but it would only amplify when wired as inverters. I haven't tried it with the N channels in isolation, but I did try a single 2N7000, and it seemed to work ok. a VN10KN also worked, but showed less attenuation. which leads me to think using the 2N7000 would be the easiest way to implement this circuit.

the values shown for the pots are my guesses, but should work ok. I'm not keen on all those unused inverters, which can't be used as inverters because the chip has no Vdd voltage. maybe I could cobble together a modified ea tremolo from the same ic .....

.... and now I see ~arph's mosfet cardinal on the other thread, I can probably forget all about his circuit, and try that one instead.

[Mark Hammer]

I made the original, so many years ago, from a construction project article in ETI.  It was severely "limited" in its ability to handle peaks with the specs indicated.  As a result input signal had to be modest, which, in turn, impacted on S/N ratio.  It was too long ago for me to have any opinion about the sound quality.  For all I know, it may be one of those now-depopulated what-the-hell-WAS-that PCBs that sit in my bin o' boards without any identifying marks on it.

I know I still have the special projects issue that contained the circuit and board layout, but I suspect that a more contemorary version, using something a little more tolerant of peaks than a 4049, will be a better choice.

[PRR]

0.8% THD is not at all hard for simple JFET limiters with comparable gain and noise performance.

Gain/noise: assume the output is 1.23V (+4dBu). By R7/R6 the level at the internal inputs is 6mV. That's mighty low.

Good catch to lose the BJT pair. They were "needed" to get the huge idle gain with 20KHz with low THD. (Though I'd think a decompensated '301 would have been fine.) The '072 has more GBW, guitar doesn't need so much, the extra transistors could be dropped.

> I'm not keen on all those unused inverters

Use them in the other channel. Don't have one? Parallel them with the others for 6dB deeper gain-reduction or increased input overload. IAC, they are 3 cents each and I can't see a "waste" here.

[StephenGiles]

I agree with Mark, and like so many ETI guitar pedal projects was quite useless. The trouble was that the magazine paid peanuts to authors for articles so what they got was usually sustandard.