Audio limiter to quiet loud radio commercials

[Frankchie]

I have an internet radio who's volume gets annoyingly loud during some commercials and I am trying to figure out a simple way to automatically reduce that volume. The output of the radio goes to an amplifier via a standard headphone jack on the radio. So inserting a limiter of some kind in that output wiring is easy.

I've tried clamping the audio line with back to back silicon diodes and that does help, but it's still too loud. I'm still playing with this technique and may try germanium diodes, etc. I don't think I need much help with this technique. BTW, I mostly listen to talk radio and the P-P voltage is about 1.2v at normal listening levels. With talk the signal looks like lot of spikes with lots of dead intervals.

Anyway, I was thinking of another technique that doesn't need a power supply. I'm thinking of rectifying and filtering the audio voltage and using that voltage to somehow control the audio feed to the amp. For example, there may be enough power to activate a sensitive relay that inserts a resistor divider circuit into the audio line. Or maybe there's an led/ldr optocoupler that might work, although, I'm not sure there would be enough power to activate the led.

So If anybody here has some other ideas or suggestions on the above.I would love to hear them.

Thanks, Frank

 

[AnalogKid]

Are you sure the commercials are louder? Not apparently louder, but Vpp louder? The audio for commercials is more heavily compressed, which increases the perceived loudness and "punch" without increasing the voltage peaks. In other words, the peak to peak value is the same as pre-compression, but the RMS value is higher.

ak

 

[Harald Kapp]

but the RMS value is higher.

Therefore a limiter based on the RMS value could help. The design of such can be arbitrarily complex. Here's an example of imho medium complexity. In any case there'll be a short delay from the onset of the high RMS commercial until the limiter has evaluated the RMS value (which takes time) and compression or limiting sets in.

 

[Frankchie]

Yes, good point about the "compression" factor...That probably is the situation since I did not see a big amplitude increase with the commercials...Although there was some increase in amplitude and I still want to fine tune the clipping diode technique, hopefully it will moderate the high volume enough to make it tolerable.

I agree the RMS technique may be the better method, but I am disappointed at the potential complexity. Given that nobody commented on my simple rectify/filter/relay technique, I suspect it may be somewhat "hairbrained". But I really would like to avoid a power supply and a lot of circuitry.

I just had a fleeting thought to use a combination of my two techniques. That is, to develop both a plus and minus DC voltage from the audio and then use those voltages to modify the bias on their respective clipping diodes. Thus bringing those diodes strongly into play upon loud audio. I don't I care if the commercials are distorted as long as the volume is lower. I haven't given it much thought so I may have overlooked something, but I think it may be viable.

Thanks to all for the help...if you have any other thoughts, please keep them coming.

Frank

 

[Audioguru]

Your diodes would cause severe distortion to loud speech and loud music in addition to loud commercials. Usually a compressor rectifies the signal and feeds the DC to the gate of a Jfet that cuts the peak levels of the audio. The original commercial compressor reduces levels of the peaks which increases the RMS levels but if it is used again for you to reduce the RMS levels its reaction times would be very slow.

An audio compressor circuit takes time to activate then the first syllable of loud speech or first parts of loud music blasts at max volume before the compressor reduces the next levels. Then the compressor takes time to get its gain back to normal so the first parts of normal levels cannot be heard. The levels are noticeably pumping up and down all over the place.

 

[AnalogKid]

Given that nobody commented on my simple rectify/filter/relay technique, I suspect it may be somewhat "hairbrained".

Not hairbrained at all. But it's activation is based on the peak value of the waveform, which already is controlled at the signal source. The circuit technique is very common (Maxim has a microphone preamp chip with it built-in), but adjusting it to the sweet spot between the two types of audio material will be difficult.

Here is an old example that will trigger the 741-bellyache syndrome:
http://blog-marlin.blogspot.com/2016/05/compressor-audio.html

ak

 

[Frankchie]

Thanks, AK,
Your circuit looks like it would do a good job for me, but I'm hoping for something simpler, without a need for a power supply.
If i can't do it simple I may give your circuit a try.
Frank

 

[Frankchie]

Thanks, Audioguro,
Yes, I agree my diodes will likely cause severe distortion....I don't think I would care for a commercial, but it's possible that would it be very annoying, Likewise I think I can cope with the slow recognition of the commercial, but if it cuts in and out that sounds bad.
Frank

 

[Frankchie]

I lashed up a quick rectifier circuit on the audio channel and was pleasantly surprised that the DC voltage without a commercial was fairly steady at about 0.4 volts, however when a loud commercial came on it jumped to over one volt and stayed there, again pretty steady. Some commercials caused almost 1.5v...no wonder it's so loud.
The bad news is that less than a 1 ma load caused the voltage to drop to zero. Not much hope of operating a relay with that supply. However, using that voltage to bias the clipping diodes who are downstream through a 1000 ohm resistor still might work satisfactorily.
If anyone has a better idea on how that voltage swing from 0.4v to over 1.0v might be used to control or shutoff the audio I sure would like to hear it. Again, without an external power supply.
Frank

 

[Audioguru]

Years ago, an incandescent low power light bulb to ground was used as a level control in an audio oscillator. But it was very slow and when a loud signal was the input then the levels bounced until it became steadily reduced.

 

[Harald Kapp]

Not much hope of operating a relay with that supply.

Not a relay, but the voltage can be used to control an electronic attenuator (e.g. figure 17 here)

 

[Frankchie]

Harold,
Great idea..Although it's been about 45 years since I designed anything using a FET...Anyway I found this design (below link) that seems to fit my situation. I don't think I need that center FET because I think that's used to open the "perfect switch" and I need some conduction to hear the muted audio. Also, the -18v switching voltage is higher than I have available, but again I don't need a perfect "on" condition. Worse case maybe I can insert a battery to boost my switching voltage. That battery should have a very long life. I'm off to search my junk box for FETs.
Thanks,
Frank

Scroll down to the "High Frequency Switch" on page 13.
http://www.ti.com/lit/an/snoa620/snoa620.pdf

 

[Audioguru]

You do not need 3 Jfets, you need only one Jfet plus a resistor to make an audio switch. Also, most Jfets activate from only a few volts, not 18V. This voltage can be the positive supply voltage, not an additional negative voltage.
What will activate the switch? Rectified program audio has many level peaks that are the same level as the peaks in the compressed commercials so the program speech and music will be full of cuts.
As I said earlier, you could feed the average or RMS rectified voltage to activate the Jfet then have the delay so the first part of the commercial will come blasting at full volume, the Jfet will slowly activate then when the commercial ends the program will return slowly.

 

[Frankchie]

Audioguru, Thanks, I can use all the help I can get.

Again, my FET knowledge is pretty rusty but I think I agree with most of your statements..

Yes, I plan on feeding the rectified audio DC which jumps from under 0.4v to over 1.0-1.5v with loud commercials. And I agree there will be some, hopefully minimal, delay. That DC level seems to stay pretty steady both on normal levels and with loud commercials, maybe varying a couple of tenths of a volt. Hopefully if I set my thresholds carefully there will be minimal "cuts".

BTW, using just one FET as an audio switch or attenuator doesn't sound right to me as I couldn't find any such examples. The only example I found was that three-FET circuit in my last message that I think could work with 2 FETs. But as I try and refresh my FET knowledge I'll keep your single FET idea in-mind.

Incidentally, I'm trying to avoid thinking about MOSFETs as that would be another area where I know nothing. Maybe that's a mistake, but I would have to teach an old dog (me) new tricks. Although I guess I could try learning MOSFETs if that is a clear advantage.

In the meantime, i couldn't find any FETs in my junkbox, but I'm still looking.

Frank

 

[Audioguru]

Three Jfets can modulate a Morse Code signal on and completely off, but one Jfet can pass audio or block it, or a resistor can pass audio then a Jfet to ground can almost completely turn it off.

 

[Audioguru]

Junkbox parts?? You must look at the datasheet of a Jfet to select one that works with the supply voltage you are using. A 2N3819 Jfet might kill your supply when it is turned on and needs a fairly high gate voltage to turn off. A 2N5484 draws a low current when turned on and needs a low gate voltage to turn off.
Mosfets are usually used to switch high currents that you do not have.

 

[Frankchie]

1. I'm still struggling with how a single jfet can attenuate an audio bipolar signal without an external power supply. So glimpsing at articles that seem to say that MOSFETs are bidirectional and Jfets are unidirectional (I think), I thought maybe the Mosfet might apply to my problem. Judging from your reply, I guess not.
2. I found some 2n5459's in my junkbox and their spec's are in the ball park for my needs. Or at-least close enough to prove the concept. Although the lower Vgs pinch-off voltage of your 2n5484 suggestion looks like a better choice.
3. I was able to attenuate a similar audio signal using a 2 jfet circuit similar to the three jfet circuit of the link I posted a little while ago. Although using the junkbox Jfets the -1.0v to -1.5v rectified audio signal voltage that controls the Jfet gate would probably not lower the loud commercials enough. No doubt I could use a JFET like the one you suggested. For the time being I think I can add a small bias battery until I can get a better JFET.

Thanks for all your help,
Frank

 

[BobK]

You got that backwards. MOSFETS have a diode in the reverse direction so they always conduct with reverse bias. JFETs conduct or not equally in either direction. Connect the source to the ground of the signal. A negative voltage on the gate will turn then turn it off.

I will try simulating a circuit to see if I can get a no power single JFET switch to work.

Bob

 

[Audioguru]

I sketched an audio attenuator using a Jfet:

 

[Frankchie]

Wow, Audioguru, thank you...

Seeing your diagrams I lashed up a quick common source Jfet circuit and it does seem to turn on and off with both a positive and negative drain voltage, although not with equal vigor (that may have been mentioned in some prior notes). The turn on with a positive drain voltage is much better than with a negative drain voltage. So I guess JFETs do kinda work with bipolar signals. Although that unequal "vigor" probably means it can't be used to reduce the commercial volume, rather just to shut off the commercial completely. Reducing the commercial volume may cause intolerable distortion due to unequal treatment of the plus and minus portions of the bipolar signal. .

I think your common source JFET circuit is designed to shut off the commercials completely. That might be okay, but some of these commercials can be several minutes long and a lot of dead air could be disconcerting

The two-JFET (back to back, source to source) circuit that i got working doesn't seem to have any obvious distortion problem. Although trying to maintain a constant (and lower) volume level could be a problem since the bias voltage does vary a little bit. I'm thinking of having two series resistors to the amp and using this two-JFET switch to turn on and bypass one of the resistors when the loud commercial comes on.

Btw, the output Z of the radio seems very high, or the Amp input is very low Z. Because simply inserting a 10k resistor in series to the amp dropped the volume dramatically. I actually had to turn the amp to full gain to get a decent speaker volume. A 1k resistor seems to have minimal impact. So R1 in your circuit probably shouldn't be much higher than maybe 2k-4k..I'm not sure if that is a problem.or not. The on resistance of the JFET may be a bigger factor with lower R1 values. Also then the C1 capacitor may have to be adjusted, but I'm not sure if C1 or C2 is really needed.

Let me know if I got something wrong.

Thanks, again, for taking the time to draw your diagrams...It definitely taught me something.
Frank