Original designer of the circuit here (Kage). This is one of my very early attempts at audio processing for broadcast and can easilyjvok wrote: ↑Thu Nov 21, 2024 7:38 pm Looks like a decent design in general. Couple of issues I can see though:
- The rectifiers only detect negative peaks. If you have audio with asymmetric positive/negative peaks (more common than you'd think especially with speech) the positive peaks won't get limited. I'd add an inverting buffer driving a second 2N3906 to each limiter stage.
- The cutoff frequency of LPF1 will shift depending on the Drive Gain control setting because the pot is effectively in series with the 47k resistor. I'd add a buffer after the pot to stop this.
- You'll get quite high distortion with the FETs connected like that. You can bring it down a fair bit by feeding the audio signal on the drain through to the gate with 6dB attenuation, but you have to be careful to avoid control voltage feed through (shifting dc level causes a thump/click when the limiter kicks in). This circuit (figure 2) works quite well: https://sound-au.com/project67.htm
- If you build it in stereo you'll want to link the control voltages between the two channels or the stereo image will shift when the limiter kicks in. Also you'll want to match the FETs between the channels. Be prepared to buy lots of FETs and spend loads of time testing to find ones that match.
be modified to obtain higher quality standards.
I agree with the points you brought up and have modified the design slightly over the years including
adding distortion cancelling feedback to the gates. I never intended to push it very hard with audio where it was brick wall limiting anyway, it was more designed with limiting as final protection and tried to keep the circuit as simple as possible for small hobbyist pirates while still creating its own pre-emphasis limiter, pilot tone notch filter and so on.
Since then I have designed a far better audio processor employing an AGC, multiband compander, high-pass rumble filter, a clipper and low-pass filter that is designed to catch peaks at the tail end of the chain and reintroduce them with phase change so not to have any overshoot from occasional hard clipping through the filter as it "predicts" overshoots before the final filtering is done.
This can be duplicated for stereo FM use if the FET gates are tied together and the low-pass filter values changed from rolling off at 9kHz to 14kHz and adjustment of the overshoot conditioner circuit pots.
This also implies the builder knows that some features can be bypassed for FM use like the asymmetrical limiter adjustment that is only useful for AM...
https://darkliferadio.proboards.com/thr ... tch?page=2
Schematics and my build can be found there. Someone was kind enough to draw up a PCB layout too.
Have to scroll through to find the schematics and other info. The original FM processor from 2012 also has its own thread on there.
Hope this helps anyone who wishes to modify for their own use.
