>
from mercenary audio:"An optical compressor performs gain reduction control via a light source into a photo sensitive cell... as the light source gets brighter, the photo sensitive cell tells the amplifiers to turn down the volume..."
> How do the photo sensitive cells "tell" the amplifiers to turn down ...?
Uh, that "explanation" is for people who just want to buy, not really understand.
>
2 years ago i did not know what a resistor was.
So now you probably know how to use an ohmmeter. Go to Radio Shack for "CdS Photoresistors (5-Pack) # 276-1657 $2.99". Hook one to your ohmmeter. Put it in total dark, it goes up past 1 Meg. Put it near a 100W lamp, it goes down toward 1 K.(*)
That's all it is.
If you want a phyzical explanation: in dark, the electrons are bound to their atoms. But light "loosens" the electrons so they may go with the flow. There is more elaborate theory available, bands and holes and groundstates. However commercial LDRs are developed by trial and error guided by experience, mixes of impurities to bridge the bandgap.
How would you use this? Most audio outputs can't be loaded-down easily. You build an L-pad, with a fixed resistor and a variable (photo) resistor. Since we will probably represent "too loud" with "more light", and the Photo-R goes low-ohms in light, we put the 50K in series and the photo-R in shunt to ground.
For proof: wire a 1.5V battery to the input of the L-pad. Measure the voltage on the photo resistor. In total darkness it will be over 1.4V, 90% of signal goes through. In bright light it will be near 0.1V, less than 10% of signal gets through.
Now wire the L-pad between a source and a beefy power amp's input. Put a car tail-lamp on the speaker output, and shine it at the photo-R. Obviously the photo-R and lamp should be in a light-tight box. For small output, the lamp stays dark, full signal passes. Around 7V-8V the lamp will glow enuff to reduce photo-R resistance, reduce signal into the amp. Large increase of signal into the L-pad will make the lamp slightly brighter, photo-R resistance goes down and down, you probably can't find enough signal to bring the lamp past 11V.
I have done exactly this scheme, except with LEDs, as a concert-hall PA amp protector. It is pretty rude, but effective.
The trick is doing a good signal-sensitive lamp without a power amplifier. Also incandescents are very slow. In 1960 the most efficient fast lamp was the electroluminescent cell, still occasionally sold as nightlights. That's what the LA2a used, but it still takes a 6V6-size driver.
The photo-R is slow. That is good and bad. If it were perfectly fast, it would clip the signal. We want it to drop quickly and hold for many audio cycles. The LA2a used specified and selected LDRs, which give a semi-fast attack and a lingering decay. They got lucky. It is hard to find a happy time-curve, and you can not change it.
The next step is to do what tube and FET limiters do. Peak-detect and decay with R-C networks. It is possible to over-drive the LDR and shorten a too-slow attack time, and your other time constants are simple R-C nets. I have built good dual-curve small-fault recording limiters this way, using the RS 5/$3 LDR pack. I spent most of a month in a dark room getting it to work well.
(*) These are made-up typical numbers. What RS really sells for 5/$3 is "Various styles and ratings".... whatever they can get cheap. In the past it was often assorted scrap: some low-R some hi-R some fast some slow. However about 9 out of 10 packs (I bought a lot of these!) had the same assortment year after year. I saw a pack recently and it looked like 5 of the same part. Even so, LDRs vary a LOT and you won't find two-the-same even if they have the same part#. Unless you can buy dozens, you will need to trim your L-pad resistor to get any sort of "match"... between channels or even just to get a mirror-voltage for a GR meter.