Audio compressor

Ivan wrote:

Andrew wrote:The idea was straight of the top of my head and I was unclear, I was wondering if, replacing both JFETS with dual-gate ones, it could be possible to use the link and the second gate to obtain a balance w/o the need to use a selected pair of JFETs

Hi Andrew,
your idea is worth trying. The LINK would be connected to gate 1 in both stereo channels, the DC bias (different in each channel) of gate 2 would shift the characteristics to match at least at one point.

VBR from Ivan

You got exactly the idea, forgive me again for being so unclear anyhow, I think that such an approach may avoid the need for matched pairs of JFETs, it remains to see how well it works... w/o some additional "balance" circuitry, but if it works, the idea may be used for multiple sources, not just two

Andrew wrote:The idea was straight of the top of my head and I was unclear, I was wondering if, replacing both JFETS with dual-gate ones, it could be possible to use the link and the second gate to obtain a balance w/o the need to use a selected pair of JFETs

Hi Andrew,
your idea is worth trying. The LINK would be connected to gate 1 in both stereo channels, the DC bias (different in each channel) of gate 2 would shift the characteristics to match at least at one point.

VBR from Ivan

Ivan wrote:

Andrew wrote:I'm wondering if, willing to build a dual channel version of the unit, it may be possible to use a double gate FET so, avoiding the need for a matched pair...

Hi, the schematic symbol suggests it is a J-FET. All dual gate FETs I have encounered were MOS types. What exact type did you have in mind?

The idea was straight of the top of my head and I was unclear, I was wondering if, replacing both JFETS with dual-gate ones, it could be possible to use the link and the second gate to obtain a balance w/o the need to use a selected pair of JFETs

I was short of fets and it took a while to find some.
In the meantime I discovered in both Chrome (audio compressor) and Edge (audio channel) the possibility to get a nice plugin doing compression of audio signal passing through the browser.
I'll play with those plugins to check if they can be usefull and helpfull for me.

Andrew wrote:I'm wondering if, willing to build a dual channel version of the unit, it may be possible to use a double gate FET so, avoiding the need for a matched pair...

Hi, the schematic symbol suggests it is a J-FET. All dual gate FETs I have encounered were MOS types. What exact type did you have in mind?

VBR from Ivan

I never used dual FETs, but it would be interesting to see how good the tracking would be with these!

By the way: you can 'cheat' a bit, by slightly changing the value of the 18K resistor at the input.

Ruud wrote:The 'link' is the point where you can connect two of these compressors together, for stereo use.
In that case you have to select two FETs that have the same characteristic.
Unfortunately FETs have a very wide tolerance.
If I remember correctly, I managed to find two 'pairs' out of twenty 2N3820 FETs.
The linking can avoid an 'image shift' (the sound appearantly moving to the left or the right of the stereo image), when compression occurs.

I'm wondering if, willing to build a dual channel version of the unit, it may be possible to use a double gate FET so, avoiding the need for a matched pair...

The 'link' is the point where you can connect two of these compressors together, for stereo use.
In that case you have to select two FETs that have the same characteristic.
Unfortunately FETs have a very wide tolerance.
If I remember correctly, I managed to find two 'pairs' out of twenty 2N3820 FETs.
The linking can avoid an 'image shift' (the sound appearantly moving to the left or the right of the stereo image), when compression occurs.

Thanks a lot Ruud, I very much apreciated your explanation and circuit.
In fact I'm not sure about what exactly might help my problem, therefore in principle a circuit with the possibiity to adjust several paramethers for sure might help me to test what is the best setting for my purpouse.
In particulr because my unability is "fluctuant" therefore it change day by day. Sometime is better, sometime is worst, therefore the solution should be settable, in order to be adapted.
I agree it is a very promising solution to have a kind of adaptive release. Thanks a lot for the suggestion.
I'll for sure test your circuit.
Sorry for the question, but at a first look I dont understand what is the "Link" arrow in the scheme.
Thanks a lot for your help.

From what I read, you will need a device that will 'iron out' loud peaks that disturb the average level and also raise the level of softer sounds.
Now the question is: do you need a fast adapting compression, or a much slower reaction to level changes.
Both have their advantages and disadvantages...
A very fast reaction usually becomes soon exhausting, because the dynamics are constantly drastically reduced.
(This can be useful in communications with a lot of background noise, to make yourself heard.)
A slow release time gives a much more pleasant sound to listen to, but the problem is that if there is a very short loud sound, it may take a relative long time (maybe some seconds) before the sound comes back to the original level.
To solve this, a form of an 'adaptive' release time is more ideal.
A fast release after a very short loud sound, and a longer release time if the compression was needed for a longer time.
This is often called 'composite release' or 'auto release'.
I have experimented a lot with FET based compressors, with really good results.
Here is a schematic of one of my designs; it is a very simple circuit that works great.
The main component is a TL074 quad OpAmp. The 'bias' (P7) should be adjusted in such a way that there is just no attenuation without any input signal.
The FET is a P-type, such as 2N3820, 2N5461 or similar.
With S1 closed, the result is the before mentioned 'composite release'.
(Of course not all potentiometers are needed, you could use trimmer potentiometers if you want a fixed configuration.
The circuit uses +15/-15 Volts.

Hello Harry,
thanks a lot for your answer, and congratulations for your web page full of resources for hobbysts.
In fact I already did some tests with a software equalizer on my pc to try to compensate the deficiencies of my hearing, and this has helped.
I was looking to a further possible step of improvment and because of that I started to study the sound compression.
I'll build some of the circuits I found on the web, basically your and another one, and I'll check how this works.
I have also found an ic, NE570, which seems to be designed for this purpose, and I'll try also this.
I'll let you know my findings, as well as I'll keep reading your interesting forum here.
Again thanks a lot.

Hello Fabri00,
Thank you for the detailed message and the explanation of your problems and your needs.

I used to work in Multimedia and and wrote manuscripts, did voice-overs and even presented onscreen in some 750 educational videos. I processed the audio for almost all the films. I also did a lot of research on compression levels with respect to the target audience. Dynamic range can be a real problem. So I always added some form of compression.

No compression can be a problem for people with relatively good hearing, if it is not heard in a clinically silent environment. Great with headphones, studio or in a listening room where there is little or no background noise. I do know what you mean about the dynamic range of some videos on the TV. Many people have a problem with this.

Light compression should be used for best effect when the target group is listening in, for example, a conference room or open office where there may be other people or light air conditioning or fans.

Heavier compression (but not too heavy) should be used when the target group is in a public area where there is a lot of background noise.

Heavier compression dramatically reduces the dynamic range and it sounds as though this is the sort of processing you need. I also have slightly impaired hearing, but at 71 years I cannot complain. I sometimes use a graphic equalizer to compensate for my own hearing deficiencies. My hearing starts to fall off at 6-8kHz and at 15kHz I can still hear, but very much reduced. In addition to this I have about 20dB reduction across the spectrum, and add tinnitus to this I sometimes have problems. So amplification and frequency correction work for me, but do not affect listening to the HF bands.

I understand that you have a really reduced dynamic range, but it could be that your own hearing problems are also frequency dependent? It may be a good idea to try a test using headphones through a stereo amplifier with a graphic equalizer, to see if you can get any benefit from adjusting the frequency range. If this helps then the compressor will reduce the dynamic range for you, but you may also be able to add some frequency correction to further improve your situation. There are loads of graphic equalizers published.

My compressor was intended for processing audio before transmission so that my voice could be heard riding over background noise and poor radio conditions. The compressor will compress very heavily so you may have to back off the compression for a more pleasing audio when you are listening, but reduce the dynamic range as much as you need to hear more clearly: clarity before quality. I presume that you are going to experiment with a bespoke hearing aid? I would be very interested in knowing how you get on with any tests and trials.

Your description of your particular hearing disability is probably not so common but I do hope that the information I have given will help you. It sounds to me as though you may  never appreciate HiFi but must concentrate on getting that audio level within a narrow amplitude range for best possible clarity. With my compressor you may need to reduce the attack and decay times as these can cause loss, or excessive sound in the first few milliseconds of a stretch of audio.

If I can help you in any way then please do not hesitate to write here. Although the board activity has declined a little there are still many people here with far more knowledge or different experiences than myself. I am sure that you can get more ideas. Of the various regular members:

Ruud - Studio audio expert with loads of experience
Ivan - Very innovative, especially with bespoke circuit ideas

Both Ruud and Ivan can take an unusual idea and "run with it", or offer ideas and suggestions.

Very best regards from Harry - sm0vpo

Hi Harry and Hi to the other peoples having posted in this thread.

I'm new of this forum, and I came here looking on the internet for audio compressors/expanders circuits.
I've grew up with the hobby of electronic, starting in the end 70's at the time of the switch from germanium transistor to silicium ones, when tubes were at the end, and I've been there for probably about 20 years, using BC109 and 2N1711 and 2N3055 to make a lot of things. I was there when the first IC started to appear in the market, as well as leds, and many other things now old fashioned, but rather modern and new at that time. Then also because of lack of time, taking care of career, building a family and so on, I went more in the direction of informatic and I left electronic aside. Few years ago I started back to play with electronic, and I'm more and more doing that now.

By the way, the reason I'm here is because I have a problem in my hearing due to a specific disease called Meniere Syndrom, which push my internal hear to accumulate more liquid than necessary.

As consequence, I have some deficiency in hearing, but more than that a malfunctioning of my hearing, and also I hear differently week by week, depending from how my internal hears are doing.

My main problem is the lack of dynamic in my hearing: either I dont hear well, but then with a small increase in volume then i hear too loudly, and my range of comfort is much reduced than in a normal person.

This is a big problem in web-meetings like Teams or Google Meet, very much used since the pandemic started, and very much used in my business to contact customers and supplyiers.
If the participants to a web meeting use good hardware, I can hear them well. But when the hardware is poor or the room is a reflecting environment, or peoples talk in a conference room far away from the mic, then I'm continuosly moving the volume up and down to lissen what people say, and the lissening become difficult and hard.

To give an example in a different situation, watching TV I hear very well the news or any other situation where peoples talk in a studio where the audio is perfect, but I have more problem to follow a movie where the audio is more dynamic.

I was looking to test an audio compressor to be used in between the audio output of my pc and the amplified speakers or headphones, and see if this might be of help for me to make my lissening more confortable.

Looking in the web I was impressed from your "Speech Processor" as you went deeply in to the dynamic of the speech reproduction and I was very interested in this.

I allowed myself to re-up this thread and to ask if you or any other partecipant of this thread and forum maybe have some suggestion for me than just testing your Speech Processor, which I'll do soon.

Thanks a lot for any help I can get from you or from others attending to this forum.

Fabri00.

Yes , i believe i need something like that . If i connect via Bluetooth i have to turn up the volume . Later if i forget the volume, when i connect a pc via jack input , the speakers jump out of the box  because of the high level. Even between 2 radio stations there is sound level difference

hi , the circuit is interesting . One time i will intend to do something like this for an amplifier i use with various inputs. I want to solve the volume issue , if i connect a phone i have one volume level , if i connect a radio i have an other volume level , if i connect a pc its on whatever level was last used . It would be nice to have such a circuit that keeps volume at fixed level. I want to make a nice background sound system outside in the garden .

Hi Zsolt,

If it is of any interest, I once used a silicon solar cell and taped that to a 6v 0.5W tungsten lamp.
this I used as a cheap wattmeter. Works from audio up to a few 100 MHz.

The same technique can be used to control an audio compressor, but the disadvantage is that you need a relatively high audio gain to make the lamp light. You need about 20mW or so to get it started, and it has to be kept in the dark. As a wattmeter it has a non-linear scale, but as the controlling element in a compressor this is not a problem. For audio, the lamp will vary intensity in sympathy with the audio, up to about 5kHz or so.

My latest retro audio generator uses a 6V lamp as a voltage-variable resistor.

But as regards a compressor, the main reason for it was to increase the output from a cheap 5W HF radio. "Barefoot" the modulation was so weak that the RF output was almost nothing. No compression. Nada. I had only one contact on it (in Spain) and the other guy had to work hard for the contact. I had to cup my hands around the mic and talk really loudly.

The compressor amplifiers, so the RF peaks at a full 5 Watts, but the average level was only 100mW or so. With more amplification and compression/limiting my voice came out of the noise.

If you listen to the four audio examples given in the project you can hear the difference between different levels of compression. That will give you a really good idea as to why a compressor is needed. All four example peak at exactly the same level - 95% of the max peak.

I already have a PCB laid out for the project, I only need to test it, and possibly make a few corrections.

BR Harry

Link to project, if you have not already seen it:
http://85.226.187.247/audio/mic_processor.htm

Yes, the idea of an audio compressor is to reduce the dynamic range of the signal.
So: reduce peaks and eventually raise the signals with a low level.
This will give a more constant output and make the audio appear 'louder'.
(Which is an advantage in communication, especially if there is a lot of disturbance or noise.)

Hi ,
what is the compressor circuit for ? does it keep the audio level constant at the output?

Well, we all are going to die - once. We can only wish it comes in far future and painless.

VBR from Ivan

We are lucky that the safety / health regulations are not so strict in every country!
Our 'friends' in China will still produce everything we need.
And nobody is thinking about taking CdS cells apart or touching/eating the contents...
In California the situation is really crazy: on every product you will find a sticker that says: "this product may cause cancer"!

Hi Ruud,
thank you for a lot of valuable info. Production of CdS elements may be stopped in the EU due to toxicity of cadmium, but photoresistors are still freely available. See e.g. this one. The amount of cadmium in one photoresistor is small and it is sealed in resin. LEDs also contain toxic parts, e.g. arsene, and they are produced and selled in huge masses.

VBR from Ivan

It seems the production of cadmium sulphide cells is forbidden now, anyway in Europe.
Yes, the combination of a LED and a light dependent resistor can make an excellent compressor.
They are also used in professional audio, but the biggest problem is that they are slow.
The time before the opto cells react can be in the order of 10 milliseconds.
This may cause 'overshoots' that can't be tolerated in serious audio applications.
A big advantage is that they have a very smooth regulation.
They behave like an 'auto-release' without extra components.
After a short peak (=a short burst of light) they recover pretty quick, but after being lit for a longer time, it takes more time to get back to the original (high) resistance.
As you mentioned, another problem is that no two opto cells are the same...
In professional audio, people want to have a 'gain reduction indicator', meaning that two gain reduction stages should be matched: one for the audio and one for the meter.
A good example of an opto compressor is the Teletronix LA-2A
At the time they used an electroluminescent panel, that worked on a high voltage to light the opto cells.
The 'famous' T4 consisted of an electroluminescent panel together with two (selected) LDR resistors in a metal plug-in enclosure.
Still in use in many studios nowadays!

Hi Ruud,
what about using a LED + photoresistor combination in one case (Vactrol or similar) for audio level control? I did some experiments some years ago and the results were promising. The project was stopped, because each unit required individual setting - bad for mass production, negligible problem for a ham. Note a CdS photoresistor was employed, not a silicon photodiode.
Just another idea.

VBR from Ivan

Hi Ruud,
That is really interesting. I am about to lay out the PCB for this project as I need it for quite a few projects.
I know that for may operators they just use one control - off/on, but I wanted something a bit better.

The technique of feeding a bit of audio into the FET is interesting. I will have to do some experiments when I get the first prototype built. See how it performs in practice and then see if I can improve it.

As I stated in the article, my little X1M needs something. "Barefoot" it is sadly lacking on TX. The Karaoke mic mod improved it a lot. But that was nothing more than increasing the AF level and let the internal clipper do it's job.

The projects I am working on now I hope will all fit together using common modules:

1 - FT-290 + 144>14MHz transverter + 35W Linear
2 - X1M with D-sub connector on the back instead of 3.5mm sockets + 35W linear + compressor
3 - Audio compressor + SSB exciter/RX + 35W linear.

But thank you for the information and the ideas. There are a lot of ideas for improving my design :-)

Best regards Harry

Hi Harry,

I just looked at your audio compressor design.
I have been building many audio compressors, mostly for high quality recording purposes.
So I was curious to see your approach.
FETs have the disadvantage having a very non-linear characteristic.
As an audio level control element, I found that it is important to reduce the level of the audio signal between drain and source to a low level, something in the range of 30 mV max.
Also, it helps to modulate the gate of the FET with a bit of the audio signal, so you feed the DC control voltage to the gate, mixed with a portion of the audio signal. This can reduce distortion significantly!
Not that distortion is the most important thing when using a compressor for communication purposes(!), but it might be interesting to experiment with this principle.
Here is an example of how this is done in the well known 1176 compressor:

Note: C3x should be 200 pF.

There are some very interesting VCA IC's (by for exemple THAT), but a FET is a simple and cost effective way to use as a gain reduction element in a compressor.

B.R.

Ruud.