I have been fascinated by compressors for a while now and thought it was time to design a simple , LA2A-ish, optocompressor from scratch.
The gain stage was going to be all tube of course. The sidechain circuit... I first tried BC547 transistors and similar ones, but no consistent results, and I didn't like the sound of the way it made the fotoresistor react.
The light element is an LED, so unlike the original LA-2A, I do not need a couple of hundred volt at some decent amps to drive light emitting foil.
The anwser is simple, very simple. I added a simple 12ax7 gain stage, like you'd see at the input of many preamps. But instead of adding a next stage, I put the LED from the coupling cap to ground and voila! She lit up like a shining star in the night!
I'm a fan of how the compressor sounds, it has some hints to the 2a, but that wasn't the goal for me.
Improvements: Adding attack and release time. I'm thinking of placing a transformer where the led is now, then rectifying the sec winding and place caps to ground from the positive signal. Still need to test this, should work though, no? Switchable caps means varying attack/rel times.
And of course , the schematic: Imgur: The most awesome images on the Internet
Sidechain circuit tube is 12ax7, makeupgain 12a*7.
The gain stage was going to be all tube of course. The sidechain circuit... I first tried BC547 transistors and similar ones, but no consistent results, and I didn't like the sound of the way it made the fotoresistor react.
The light element is an LED, so unlike the original LA-2A, I do not need a couple of hundred volt at some decent amps to drive light emitting foil.
The anwser is simple, very simple. I added a simple 12ax7 gain stage, like you'd see at the input of many preamps. But instead of adding a next stage, I put the LED from the coupling cap to ground and voila! She lit up like a shining star in the night!
I'm a fan of how the compressor sounds, it has some hints to the 2a, but that wasn't the goal for me.
Improvements: Adding attack and release time. I'm thinking of placing a transformer where the led is now, then rectifying the sec winding and place caps to ground from the positive signal. Still need to test this, should work though, no? Switchable caps means varying attack/rel times.
And of course , the schematic: Imgur: The most awesome images on the Internet
Sidechain circuit tube is 12ax7, makeupgain 12a*7.
If you put the emitter section in the anode load of the 1st stage, the compression will be more linear, also, the output cathode follower take-off point should be on the bias point not directly on the cathode.
You will get significant distortion with this circuit, as the LED feed is unsmoothed and will only light on one half-cycle. Essentially you are compressing one side of the waveform only. Not a compressor, but an effects box. Fine if you like the sound.
For the follower I used this design: https://wtfamps.wordpress.com/muchedumbre-3/
I Know the LED is only lighting up on one half of the cycle, I might add another LED in opposite polarity, or go the transformer/rectifier way.
I Know the LED is only lighting up on one half of the cycle, I might add another LED in opposite polarity, or go the transformer/rectifier way.
> LED feed is unsmoothed and will only light on one half-cycle. Essentially you are compressing one side of the waveform
The photo-resistor has significant release time, much more than a half-cycle in midrange audio. Unfiltered full-wave worked smooth in LA2a. Half-wave may get a bit dirty in bass, true; I would throw a bridge-rectifier in there. Also a panel-LED so's I know when it is working.
LA2a, and most rude-gaincell compressors, take lamp/sidechain feed from after the gaincell, to get some NFB happening and a smoother limiting curve. Feedforward is a thing, but either does not limit flat or needs compensation of different effects to get semi-flat over a useful range.
The photo-resistor has significant release time, much more than a half-cycle in midrange audio. Unfiltered full-wave worked smooth in LA2a. Half-wave may get a bit dirty in bass, true; I would throw a bridge-rectifier in there. Also a panel-LED so's I know when it is working.
LA2a, and most rude-gaincell compressors, take lamp/sidechain feed from after the gaincell, to get some NFB happening and a smoother limiting curve. Feedforward is a thing, but either does not limit flat or needs compensation of different effects to get semi-flat over a useful range.
Yup, measurements with scope confirm full cycle function for higher frequencies, basses do clip. Sonically useful in my opinion (fuzzy bass effect). I'll add a transfo, rectifier and caps for attack/release control.
Frontpanel LED indicator was already present, didn't draw it in the schematic because some people might like a VU meter instead. The LA-2A circuit shows how this can be done. I also made the indicator led at the frontpanel switchable, so you don't rely on your eyes but your ears to hear the amount of compression.
Frontpanel LED indicator was already present, didn't draw it in the schematic because some people might like a VU meter instead. The LA-2A circuit shows how this can be done. I also made the indicator led at the frontpanel switchable, so you don't rely on your eyes but your ears to hear the amount of compression.
Some are slower than others, and most are not symmetrical, their resistance decreases faster than it increases.
I have tried them as the active element in a Moog style ladder type Voltage Controlled Filter. As expected they produce a rather unique, but useful effect. LED / mosfet style optocouplers are useful too.
> add a transfo, rectifier and caps for attack/release control.
That becomes a VERY different animal.
The LA2a "works" because some early LDRs had fairly short attack and a long tail on the release. This can be adjusted in chemistry and baking. It appears the LDRs that were used at first were outliers, also not something any other industry needed. I think UREI spent decades trying to source and test-out "similar" LDRs from the ones in production. Recently all long-time production stopped (teeny amounts of toxic elements). Chinese shops are baking LDRs but at this point the quality is erratic, and the future is murky because of EU bans on these elements in new gear.
You can *instead* semi-ignore the LDR's time constants and do it all as Analog Computing, rectifier and asymmetric filtering. Pick a faster LDR (these are more available than slow ones). Double-up and wrap feedback around the driver so when you want resistance to go down fast, an opamp WHOMPs the LED with current, and tapers-off when the LDR resistance comes down. I built such a thing in the 1980s, with two-stage time constants, and used it for live recording protection for a couple decades.
Remember that LDRs distort. Below 0.1V, hardly at all. When I built speaker protectors limiting at 2.8V, the over-threshold distortion was audible (but not unhappy for that application).
That becomes a VERY different animal.
The LA2a "works" because some early LDRs had fairly short attack and a long tail on the release. This can be adjusted in chemistry and baking. It appears the LDRs that were used at first were outliers, also not something any other industry needed. I think UREI spent decades trying to source and test-out "similar" LDRs from the ones in production. Recently all long-time production stopped (teeny amounts of toxic elements). Chinese shops are baking LDRs but at this point the quality is erratic, and the future is murky because of EU bans on these elements in new gear.
You can *instead* semi-ignore the LDR's time constants and do it all as Analog Computing, rectifier and asymmetric filtering. Pick a faster LDR (these are more available than slow ones). Double-up and wrap feedback around the driver so when you want resistance to go down fast, an opamp WHOMPs the LED with current, and tapers-off when the LDR resistance comes down. I built such a thing in the 1980s, with two-stage time constants, and used it for live recording protection for a couple decades.
Remember that LDRs distort. Below 0.1V, hardly at all. When I built speaker protectors limiting at 2.8V, the over-threshold distortion was audible (but not unhappy for that application).
I have decided to just bite the bullet and skip over all the middle steps and go directly to the end game. Input audio gets split two ways, one path goes to the Voltage Controlled Amplifier stage. The other path goes into the A/D port of a microcontroller with DSP capabilities which drives the control port of the VCA via its D/A port. The VCA output is also sampled by the A/D for Gates Sta-Level emulation.
The VCA can be built several ways. I am currently exploring a few, with two polar opposite benchmark designs for reference. Simple, cheap and excellent, a THAT 2180 chip. Complex, expensive, but traditional, the 6386 tube based VCA cell made popular in the Fairchild 670 compressor....I have several vintage GE 6386 tubes.
For the digital control stuff, I am using a Teensy 3.6 module. It has an excellent drag and drop GUI audio development tool, and is programmed using the standard Arduino tool set.
The VCA can be built several ways. I am currently exploring a few, with two polar opposite benchmark designs for reference. Simple, cheap and excellent, a THAT 2180 chip. Complex, expensive, but traditional, the 6386 tube based VCA cell made popular in the Fairchild 670 compressor....I have several vintage GE 6386 tubes.
For the digital control stuff, I am using a Teensy 3.6 module. It has an excellent drag and drop GUI audio development tool, and is programmed using the standard Arduino tool set.
Here are some sound samples: https://soundcloud.com/mk5-studios/opto-comp-mk5-light-rider
First clean signal, then : No compression, just tube and transformer sound ======== Light, fast compression======Heavy fast compression======Heavy slower compression=====Heavy slowest compression
As you can see I added response time controls.
Enjoy!
I believe it was selenium in the LDR's that made some people nervous...
First clean signal, then : No compression, just tube and transformer sound ======== Light, fast compression======Heavy fast compression======Heavy slower compression=====Heavy slowest compression
As you can see I added response time controls.
Enjoy!
I believe it was selenium in the LDR's that made some people nervous...
Last edited:
Some LDR's use Cadmium Selenide, but most use Cadmium Sulphide. I know that Cadmium is the toxic metal that caused the plant where I worked to shut down its battery manufacturing operations and undertake an expensive clean up.
From Wikipedia:
The use of CdS and CdSe[3] photoresistors is severely restricted in Europe due to the RoHS ban on cadmium.
There are some newer LDR chemistries that do not require toxic metals, but they are expensive and hard to find. Google searching seems to indicate that most of this new technology is directed toward military, medical and thermal imaging applications.
It should be possible to use an ordinary LED / phototransistor optocoupler (probably the mosfet version) in a compressor or VCA's and VCF's for music synthesizer applications, but some means of linearization will be needed. There are some devices with one LED driving two matched phototransistors. The second transistor is used in a feedback loop for linearization.
From Wikipedia:
The use of CdS and CdSe[3] photoresistors is severely restricted in Europe due to the RoHS ban on cadmium.
There are some newer LDR chemistries that do not require toxic metals, but they are expensive and hard to find. Google searching seems to indicate that most of this new technology is directed toward military, medical and thermal imaging applications.
It should be possible to use an ordinary LED / phototransistor optocoupler (probably the mosfet version) in a compressor or VCA's and VCF's for music synthesizer applications, but some means of linearization will be needed. There are some devices with one LED driving two matched phototransistors. The second transistor is used in a feedback loop for linearization.
- Status
- Not open for further replies.