FM Audio Broadcast Processor Project

[darklife]

There was an old project of mine that someone brought up here I designed around 2012 back when I was learning about
audio processing for broadcast in a time when it was hard to find information on these topics.

I no longer support that ancient project as it was riddled with problems and misconceptions in design and after seeing it
linked here I felt compelled to redesign it properly.
Well after crafting a 3 band AM audio processor I have started to rework that old FM processor too.

So far it's the basics much like the original idea I had (loosely based off Orban 8000a) but implemented properly after all
these years, a simple wideband limiter that splits off to a high frequency pre-emphasis limiter, then combined into a clipper,
then into basic filtering stages. Filter ringing can be cleaned up too if anyone is that anal about it, I have circuits for that I can post
if curious. The MPX circuitry and other stuff other people can fill in if you really want to recreate a similar full all in one processor
box.

There is a ton that can be done here and I'm at a point where it's time to release schematics for those who want to build
it as is or develop something similar. Just please give credit where due, Kage Electronics, Ken.
Been in the pirate community for decades now and like to give back ideas when I can.

Anyway schematics may change, you can visit updates over at.. https://darkliferadio.proboards.com/thr ... ts-project

Schematics... (right click and click open image for full size)



Any ideas or schematic changes are appreciated. I realize there is a lot that can be modified, especially since I was working
with what I had in the bin while developing it.

Two ideas I have rolling around in my head right now are a possible bass band audio limiter ahead of the wideband one with a 200Hz split like the Orban 8100, and possible dynamic time release circuitry that changes release time dependent on aural activity over time to sound more natural instead of breathing/pumping.

BTW this is for 75uS pre-emphasis. It can easily be converted to 50uS since I know most of you are not in the US like I am.
Don't mind the dirty power supply, in fact don't mind any of it being dirty lol. Just happy to hear the thing work.

[darklife]

Lol, it's a prototype rebuilt from another prototype
So soon I plan to drill out a center faceplate hole for the multi-meter selector switch to choose L, R, L+R, G/R, and possibly bass G/R if designed in. After some paint and labels it might actually look the part. Sound is all that matters though. All from scrap. Think the chassis was a DVD player and faceplate cut with tin snips.

[darklife]

Wow rereading my own post I just realized a fatal mistake. I need a three prong cord grounded to chassis. Funny how we see things after taking pictures The whole time I was building this and I just noticed something so stupid dur. I even included the fuse. How the hell... total blind spot wow.

[jvok]

Nice work, tbh I didn't even notice the missing ground I just assumed it was there.

I like to use diode-connected BJTs instead of diodes in the clipper. Gives a much sharper onset which means near zero distortion up to the clip point, whereas diodes start to distort a fair way before they clip.

[Albert H]

Jvok - that's a good point, and something I discovered (the hard way) a long time ago!

Most LEDs also have a very abrupt "edge", and I've made use of the property of bidirectional Green LEDs in that they crack over at exactly PPM6. Roger Howe used this trick in "his" Limiter+ (which was actually a rework of the fast Limiter published in ETI in the early 80s). Roger just added an overpriced SSM debalancing chip at the input, and added passive pre-emphasis, slightly simplified the sidechain (by removing a pretty critical adjustment that was necessary to overcome component tolerances), and modified the original PCB design to accomodate his "fixes". He also added the LED clipper at the output to try to shave off the inevitable over level transients that would get through because they were faster than the sidechain.

As supplied, the Limiter+ worked reasonably well, but could pump horribly on bass-heavy materiall!

That ETI Limiter made use of transconductance amplifiers as variable negative feeedback paths for good quality op-amps. The nastinesses of the TCAs was largely overcome by the cleaning effect of negative feedback. The problems with the circuit become apparent when trying to handle really wide-ranging level, because there's only about 22 - 24 dB of attenuation available at most. In one product I designed, I overcame that problem by cascading two of the gain cells in each channel....

After many years of experimentation, I found that the most effective (and easily constructed) gain control just uses a FET or CMOS switch-based chopper, switched at ultrasonic frequencies, by a variable pulse-width oscillator. PWM control is inherently balanced from channel to channel, and is really cheap to implement. It also has vanishingly small distortion if your post-chopper filters are well designed. In some products, I've used chopping frequencies in excess of 150kHz (the filters are easy) or, in later designs, I've gone for a multiple of the 19 kHz pilot (often 114kHz) derived from the stereo coder clock.

Kage - take a close look at PWM - it's a really simple approach, and your 15 kHz LPFs will certainly prevent breakthrough of a really high chopping frequency, without further modification.

[darklife]

Nice work, tbh I didn't even notice the missing ground I just assumed it was there.

I like to use diode-connected BJTs instead of diodes in the clipper. Gives a much sharper onset which means near zero distortion up to the clip point, whereas diodes start to distort a fair way before they clip.

Nice idea. Luckily the clipping diodes I used are in a spot where I can easily swap them out for BJTs so I'll have to test that soon.
That is a sore spot in my circuit I need to look into since the diodes seem to soft clip rather than chop the transients as intended with continuous signal but work okay with occasional limiter overshoots. The limiter attack times are very fast with this design luckily so it's not a huge concern but it can be seen on a scope if observing closely with pulsed signals.

The soft clipping does have the benefit of being able to drive the clipper slightly hot giving a louder more punchy audio if desired up to a limit where it becomes audible but that's really only useful for AM processing since FM doesn't need loudness to get through the noise. Some stations used it to good effect in the early days to sound more full and present on the air if dialed in properly so not to cause fatigue.

I was only able to get harder clipping by driving them with a lower resistance but it strains the OPAmp too hard so I will experiment with your idea soon.
Any particular transistors I should try in that configuration you'd recommend?

[jvok]

I used bc546 mainly because its my go-to part but I think any small signal audio transistor will work. I found around 10k is good for the feed resistance - too high resistance and you start to get soft clipping like with diodes, too low and the clip point starts to shift

[Albert H]

Aother trick that's commonly applied is to bias the clipping trasistors so that the clipping point can be fine tuned by altering the bias.

Variable Clipper.jpg

I used any small NPN / PNP pair with roughly equal characteristics.