Audio filter placement

[ki4nr]

I been experimenting with my audio gear and have a question on were is the best place to put the Bandpass filter.

My setup is preamp, EQ, compressor, adjustable positive/ negative peak clipper to custom 455khz two quadrant multiplier for the modulator in the IF strip of my Yaesu 1000MP, perfect scope picture and responds just like plate modulation.

I call it synthetic plate mod. One very cool feature is by changing the bias you can adjust the symmetry. I have it set at 95% on negative baseline, it goes positive to 150 percent.

Back to the filter, My thoughts are either between the preamp & EQ  or between EQ & compresser. Ive tried both ways and I can't really tell much difference. The bandpass filter is from my car stereo days, high end gear 24db per octave. Not are good as DSP or switched capacitor, but decent for analog.

I have the filters set at 3 db down at 80hz and high end at 3.8khz. It controls the bandwidth fairly well and my total RF bandwidth on the SDR play waterfall is around 9khz

[K8DI]

If you only have one filter, you want it after the clipper. Clipping causes distortion of course, and those sharp clipped waveforms contain harmonics which do end up on the air. Even better is two filters, one early so you’re not processing sounds you’re going to filter off anyway, and a second low pass after the clipper.

Ed

[ki4nr]

Yes that is my setup, I do have a 5khz filter after the clipper I forgot to include. The clipper is just to set the negative audio and catch any transient peaks that get by the compressor. I have the clipper threshold slightly above the compressor threshold. The compressor is doing 95% percent of the work. The clipper is mostly transparent.

I'm thinking filtering the audio early as possible is the best way to go. If I place the filter right after the mic preamp it would only have to deal with flat consistent mic audio, If it's after the EQ then it will have to handle the EQ curve and may not be as sharp as it could be with the extra burden of EQ boost.

[K9MB]

Check out this thread.

http://amfone.net/Amforum/index.php?topic=47699.0


The title refers to W3AM’s All Filter circuit.
This circuit lowers the peak to average ratio so that modulation density is improved.
I corresponded for months in 2021 with Jim Tonne who has spent his life developing speech circuits and modulators for AM transmitters.
Jim has an active low pass software on his sight that allows one to see the ringing of a filter when driven with a very high rise time signal like clipped audio.
Unless you put this antialiasing filter after the clipper, the harmonic energy in the clipped audio plus the overshoot (ringing) will give your signal a wider bandwidth and alienate your neighbors on the band..

Here us Jim’s site:

http://tonnesoftware.com/
Lots of free software and articles

http://tonnesoftware.com/appnotes/pwm/PulsewidthModulators.pdf

I helped edit this paper, or rather proof read it. Jim updated not only his pwm articles but also talks about overshoot and also active filters design for “undershoot”

http://tonnesoftware.com/downloads/SpeechProcessor-As-Submitted-To-QEX.pdf

His speech processor monograph offers a complete solution for a speech system



I developed a filter under Jim’s direction that can completely eliminate the overshoot that these active low pass filters normally exhibit while amplifying square or trapezoid waves[ the worst case scenarios that we want to avoid to prevent splatter.
See the images below:
Note that the response curve is trailing off. It is necessary to get the antialiasing “undershoot” characteristic needed.
In his speech processing paper, he has as his first preamp- a filter with a peak at about 2800hz to maximize intelligibility and it gives the boost needed to make the overall response perfect.

Note that in this active filter, the capacitors are fixed and the resistors are all the same. By making the resistor values higher or lower, a cutoff that is desirable is obtained and the characteristics of the filter do not change.

I have been meaning to publish an entire summary of this work, but have not had time. This is only a portion of it. Read Jim’s articles to obtain all you need to know.
73, Mike

[ki4nr]

Thanks Mike for the info.

I have read some of Jim's papers, all very interesting. I have experimented with phase rotators before and it just did not have the audio like asymmetrical mod does. I have way more average modulation by using asymmetry.

Maybe in broadcasting it works because they are limited by FCC. But ham radio is not. I have set up my radio with 100% symmetrical modulation with a phase rotator and looked at the bandwidth on the SDR play waterfall, then set it back up at 150% asymmetrical and there was no change in bandwidth. The big change was the asymmetrical modulation was so much louder and just as clean.

Great on the anti-aliasing filter, I have just a simple 5khz 12db per octave filter after the clipper. I will breadboard that schematic you posted to replace the one I am using, it's simple enough. In my case it probably won't do much as my clipper is just there to shear off the transient peaks coming out of the compressor.

I have a super fast peak detecting compressor but it still has some overshoot. The clipper threshold is higher then the compressor and it stops that and causes no distortion. But if I ever wanted to drive the clipper harder it will then make a difference over my simple filter.

[K9MB]

Thanks Mike for the info.

I have read some of Jim's papers, all very interesting. I have experimented with phase rotators before and it just did not have the audio like asymmetrical mod does. I have way more average modulation by using asymmetry.

Maybe in broadcasting it works because they are limited by FCC. But ham radio is not. I have set up my radio with 100% symmetrical modulation with a phase rotator and looked at the bandwidth on the SDR play waterfall, then set it back up at 150% asymmetrical and there was no change in bandwidth. The big change was the asymmetrical modulation was so much louder and just as clean.

Great on the anti-aliasing filter, I have just a simple 5khz 12db per octave filter after the clipper. I will breadboard that schematic you posted to replace the one I am using, it's simple enough. In my case it probably won't do much as my clipper is just there to shear off the transient peaks coming out of the compressor.

I have a super fast peak detecting compressor but it still has some overshoot. The clipper threshold is higher then the compressor and it stops that and causes no distortion. But if I ever wanted to drive the clipper harder it will then make a difference over my simple filter.

Ok, I know other guys like Steve who uses asymmetrical modulation and Steve sounds very good. There are opinions both ways.
I recommend that you download Jim’s Active Low Pass program and enter your present filter and the one I posted and run a square wave through both and see how much overshoot you see.
I entered some published lpf citcuits from AMfone and they rung pretty bad with square waves. Admittedly, that is a worse case scenario and most operators do not clip that hard.
Check out Jim’s compressor in his speech article. It uses a photo resistor and an LED to give very fast attack.
Of course, one can go with DSP and beat any analog filter, but these analog circuits look pretty good if you stay “old school analog”.