Really Cheap Compressor

soggybag · Post by **soggybag** » 08 Apr 2010, 14:33

This reminds me of the John Hollis Flatline compressor. I like the use of two opposing LEDs this is a nice solution to the typical rectifier. It has an inverting signal path.

mictester · Post by **mictester** » 08 Apr 2010, 19:12

soggybag wrote:This reminds me of the John Hollis Flatline compressor. I like the use of two opposing LEDs this is a nice solution to the typical rectifier. It has an inverting signal path.

It's the simplest solution! The original idea was to replace the old fashioned way of getting compression - connecting an LDR across the volume control of your amplifier, then illuminating it with an incandescent lamp driven from a tap on the output transformer. The louder the signal, the brighter the lamp, and the more the audio was shunted to earth at the volume control! It was the very simplest compressor, but along with some nice second harmonic distortion introduced by slightly changing the bias voltage to the screen grids of the output bottles, you ended up with a very "loud", very lovely sounding amplifier that would give fabulous sustain to any guitar.

The Really Cheap Compressor is a minimum parts design that was meant to be a replacement for the simple valve amplifier modification outlined above. I've built several of them (which reminds me, I need to make another one this week), and everyone who's used them, likes them. If you use a good quality dual op-amp (a TL072 is OK, but the LM833 or NE5532 are slightly quieter), you won't hear any added hiss, and no added distortion. This compressor is entirely clean, and doesn't have any part of the LEDs (or any other diodes) in the signal path, so there's no clipping (unlike with the rectifier circuit in the Dynacomp). The LDR is slightly slow to respond, so the attack of the guitar note is retained, but the squeeze is nicely applied to the decay portion of the note - just where you want it. This is one of those "is it on?" effects, but you really miss it when it's off!

culturejam · Post by **culturejam** » 08 Apr 2010, 19:50

mictester wrote:You're going to learn a lot of colloquial English on this board!

Damn skippy!

CRBMoA · Post by **CRBMoA** » 08 Apr 2010, 19:58

culturejam wrote:
mictester wrote:You're going to learn a lot of colloquial English on this board!
Damn skippy!

Word!

chicago_mike · Post by **chicago_mike** » 08 Apr 2010, 20:31

Shake a scene you turkeys!

soggybag · Post by **soggybag** » 08 Apr 2010, 20:36

You should take a look at the Flatline. It's similar. Both are built on similar ideas but different designs. Here's a link to the Flatline:

http://www.hollis.co.uk/john/flatline.jpg

I like your LED setup with the two LEDs. The Flatline uses a more complex diode network with a single LED. Flatline uses a single op-amp but in a non-inverting arrangement.

livingnote · Post by **livingnote** » 08 Apr 2010, 22:15

Hey - cool, let me get back to you on your mail (madness here too), but for the
Vactrol version of this one here - do you think it might be smart to do some kind of
rectification before I hit the LED - the VTL5C only has this single LED, not sure how
that would change response characteristics...and on the subject - hmm...

timing network? Okay back to work here

mictester · Post by **mictester** » 09 Apr 2010, 05:07

livingnote wrote:Hey - cool, let me get back to you on your mail (madness here too), but for the
Vactrol version of this one here - do you think it might be smart to do some kind of
rectification before I hit the LED - the VTL5C only has this single LED, not sure how
that would change response characteristics...and on the subject - hmm...

timing network? Okay back to work here

With the Vactrol, you would want to rectify, but you could just put a reverse-biased diode across the LED side. This would mean that the Vactrol would only light on one half of the applied waveform, but it shouldn't matter too much.

soggybag · Post by **soggybag** » 11 Apr 2010, 07:58

I have been thinking about your compressor the last few days. Humor my attempt at analysis. Please feel free to let me know where I get it wrong or if there's something I missed.

The basic circuit is made of two a non-inverting buffer or pre-amp and a second buffer that drives the LEDs.

The first buffer is made up of the first op-amp stage. Ignoring the LDR for the moment. This is an inverting stage with a gain of 10 (2M2/220K). The 22pf cap reduces gain for high frequency stuff.

Considering the LDR in the first stage the gain is variable. At the high end the LDR has a resistance of 1M in parallel with the 2M2 the total resistance in the feedback loop is about 670K. Which makes the gain about 3.

As the two green LEDs light up the resistance of the LDR goes down reducing gain. We could pick a number and do the calculations, but figure this whole thing acts as limiter. The louder the signal at the input the brighter the two green LEDs will get reducing gain. As the signal fades away, LEDs dim and the gain rises, we have the compression effect.

It's hard to guess what the range of resistance will be for the LDR. Since it will vary with the device and how bright the LEDs get. Let's imagine the gain ranges from something less than 1 to about 3.

The second stage is made of the second op-amp also inverting. It is driven by the output of the first stage. This second stage has a gain 1 to 11 ( 10K to 110K / 10K). The output of this stage lights the LEDs based on how strong the signal is coming off the first stage.

Here's where I'm a little fuzzy. It seems the first stage would bias the second at about half V. If so that means, with sustain at minimum, we have 4.5v at the positive side of the 1uF cap. How does light up the LED?

I'm guess one LED is being lit while the electrons flow up through the 1K from ground until the cap reaches equilibrium. At which point the LED turns off. When the charge on the cap switches these electrons flow out again and light the other LED.

I'm guessing, if this is correct, that this is why the LEDs don't stay lit when the second op-amp is biased at half V+. The charge at the 1uF cap needs to cycle positive and and negative to keep electrons flowing in and out to alternately light each LED.

These are the kind of things that keep me up at night. Now that I've got this out I can go to sleep.

mictester · Post by **mictester** » 11 Apr 2010, 11:08

soggybag wrote:I have been thinking about your compressor the last few days. Humor my attempt at analysis. Please feel free to let me know where I get it wrong or if there's something I missed. [/quote}
OK.
soggybag wrote: The basic circuit is made of two stages - a non-inverting buffer or pre-amp and a second buffer that drives the LEDs.
Yes, but they're not really "buffers" - they're both amplifiers.
soggybag wrote: The first buffer is made up of the first op-amp stage. Ignoring the LDR for the moment. This is an inverting stage with a gain of 10 (2M2/220K). The 22pf cap reduces gain for high frequency stuff.
Correct. The 22pF gets rid of any stability problems. It's not a critical value - I've used as high as 100 pF, and you can't hear any change to the signal through the box.
soggybag wrote: Considering the LDR in the first stage the gain is variable. At the high end the LDR has a resistance of 1M in parallel with the 2M2 the total resistance in the feedback loop is about 670K. Which makes the gain about 3.
That's almost right, but usually the initial gain is a little higher because the dark resistance of most LEDs I tried was higher than 1M. I found I usually got an initial gain of about 4.
soggybag wrote:
As the two green LEDs light up the resistance of the LDR goes down reducing gain. We could pick a number and do the calculations, but figure this whole thing acts as limiter. The louder the signal at the input the brighter the two green LEDs will get reducing gain. As the signal fades away, LEDs dim and the gain rises, we have the compression effect.
It's hard to guess what the range of resistance will be for the LDR. Since it will vary with the device and how bright the LEDs get. Let's imagine the gain ranges from something less than 1 to about 3.
The gain actually varies from around 4 down to 0.7, and acts as a compressor rather than a limiter - the input vs output graph shows a compression ratio of about 8:1 (which, I suppose, is close to limiting).
soggybag wrote: The second stage is made of the second op-amp also inverting. It is driven by the output of the first stage. This second stage has a gain 1 to 11 ( 10K to 110K / 10K). The output of this stage lights the LEDs based on how strong the signal is coming off the first stage.
Correct. Obviously there's a switch-on threshold for the LEDs which means that there's low level range of signals that aren't affected by the compression. It also has a slightly strange effect in conjunction with the response speed of the LDR in that it doesn't affect the initial attack of each note or chord, which makes the effect less obvious, and leaves the instrument sounding more "natural". This is why some people have asked "Is it on?".
soggybag wrote: Here's where I'm a little fuzzy. It seems the first stage would bias the second at about half V. If so that means, with sustain at minimum, we have 4.5v at the positive side of the 1uF cap. How does light up the LED?
The LEDs are just seeing an audio signal. Forget the DC conditions at either side of the capacitor feeding the LEDs - it's not significant. The LEDs just "see" the audio and if it exceeds the threshold of conduction of the LEDs, they will light up . The amount they light will be (within limits) proportional to the audio level applied. One will light on positive half cycles, and the other on negative half cycles.
soggybag wrote: I'm guess one LED is being lit while the electrons flow up through the 1K from ground until the cap reaches equilibrium. At which point the LED turns off. When the charge on the cap switches these electrons flow out again and light the other LED.
You don't need to analyse so closely, but you're basically correct.
soggybag wrote: I'm guessing, if this is correct, that this is why the LEDs don't stay lit when the second op-amp is biased at half V+. The charge at the 1uF cap needs to cycle positive and and negative to keep electrons flowing in and out to alternately light each LED.
These are the kind of things that keep me up at night. Now that I've got this out I can go to sleep.

You can now get your sleep! You're right.
Just try the amplifiers with fixed resistors and LEDs on your breadboard, and you'll see that the LEDs light up with each note you play, and fade out as the note decays. When the LEDs are put into the complete circuit with the LDR inside a dark box, the range of sustain is quite remarkable, but the circuit doesn't suffer from the problems of excessive amplification of very quiet signals, which causes hiss and hum issues in compressors like the Black Finger, Dynacomp and many others.

jrjoe21 · Post by **jrjoe21** » 11 Apr 2010, 14:57

Built this yesterday with dunes2k layout... works perfectly.

It's nice and "transparent" (compared to a dynacomp or orangecomp, which are also nice in their own way). mictester is right when he says "it's the is it on" type of comp...it does not pump at all. I really like it and recommend it.

A little bit of advise, if you use futulec's LDR measure them... I initially used LDR #10 which is supposed to be 1M dark but measured 500k... which reduced the gain of the signal... and then switched to #11 which measured close to 1M instead of 2M as specified. This was perfect.

Also, make sure you put the led/ldr in a dark space when testing or no signal will come through (gain will be close to 0).

Thanks mictester and dunes2k!

mictester · Post by **mictester** » 11 Apr 2010, 15:23

jrjoe21 wrote:Built this yesterday with dunes2k layout... works perfectly.

It's nice and "transparent" (compared to a dynacomp or orangecomp, which are also nice in their own way). mictester is right when he says "it's the is it on" type of comp...it does not pump at all. I really like it and recommend it.

A little bit of advise, if you use futulec's LDR measure them... I initially used LDR #10 which is supposed to be 1M dark but measured 500k... which reduced the gain of the signal... and then switched to #11 which measured close to 1M instead of 2M as specified. This was perfect.

Also, make sure you put the led/ldr in a dark space when testing or no signal will come through (gain will be close to 0).

Thanks mictester and dunes2k!

Thank you for your kind comments!

dune2k · Post by **dune2k** » 11 Apr 2010, 15:34

Fine.

So the layout is now verified.

soggybag · Post by **soggybag** » 11 Apr 2010, 16:22

Thanks for giving me the review on my analysis. I'm still learning about this electronics stuff.

Here's my take on this. As an exercise I rearranged things so it had a non-inverting input. Basically this is just a mash up between the Really and the Flatline compressors.

mictester · Post by **mictester** » 11 Apr 2010, 18:07

soggybag wrote:Thanks for giving me the review on my analysis. I'm still learning about this electronics stuff.

Here's my take on this. As an exercise I rearranged things so it had a non-inverting input. Basically this is just a mash up between the Really and the Flatline compressors.

Your version might work, but some of the values I calculated were deliberately chosen to shape the response (for example) of the sidechain (the second amplifier stage driving the LEDs). I wanted to minimise "handling" noise - to get rid of the thumps that most compressors accentuate when a guitar is moved around. I also wanted to keep the maximum gain down (to minimise noise) and give slightly more compression to upper frequencies, because the lower strings on a guitar naturally have more sustain than the upper strings. I also wanted the circuit to be able to go from almost no compression to completely squashed over the range of the "sustain" control, which I achieved with the component values that I calculated. Finally, I wanted to isolate the LED drive and rectification functions completely from the audio path - I wanted the results to be entirely clean.

As described, the simple compressor works really well. We've used them in studios, and every guitarist that tried one has wanted one!

soggybag · Post by **soggybag** » 11 Apr 2010, 19:05

My version was more of an exercise.

I brought up the LED thing, because without the cap the the arrangement becomes a DC thing. Where only one of the LEDs is lit. With the cap the situation becomes AC. The idea of a cap being a virtual ground still throws me at times. Just trying to wrap my head around this idea.

Post by **azrael** » 12 Apr 2010, 19:38

Anyone tried this with bass?

dune2k · Post by **dune2k** » 12 Apr 2010, 20:02

I'm going to build it some time soon...probably next month.
I just need a LDR and the 2 LEDs and I don't want to order stuff just for that...
The layout itself should be verified since jrjoe21 built the guitar version (I guess!) and it works. Only difference between the 2 layouts are the changed capacitors mentioned by mictester.

CRBMoA · Post by **CRBMoA** » 12 Apr 2010, 21:29

Anyone tried this with bass?[/quote]

Not yet. But hopefully the question will prompt the designer to recommend part substitutions where appropriate

, as it is obvious (to me, anyway) that any ham fisted alterations I might make to the circuit would lack the reasoning or analysis of the originator.

mictester · Post by **mictester** » 13 Apr 2010, 15:39

dune2k wrote:I'm going to build it some time soon...probably next month.
I just need a LDR and the 2 LEDs and I don't want to order stuff just for that...
The layout itself should be verified since jrjoe21 built the guitar version (I guess!) and it works. Only difference between the 2 layouts are the changed capacitors mentioned by mictester.

Increase the values of the capacitors as described, and (possibly) increase the gain of the first stage slightly by reducing the 220k to 180k. That works really well with my Precision.

Enjoy

Really Cheap Compressor [documentation]