> simply reading the {CdS}manufacturer's datasheet...
> I don't know much about the EL part.
Insufficient information. It has been observed the EL panel "color" changes with excitation. But we have no hard data, not even semi-firm data, nor any home-shop way to measure it. If someone did plot the EL panel response, then there is the tricky business of multiplying EL response against CdS response, both fairly steep hills and prone to graph-reading errors.
_I_ think it is sensible to treat the EL+CdS as a "black box" (which it is), and not fool around with what happens between.
> I get a full-voltage bandwidth of 18 kHz which is probably fine.
What small-signal response do you get? I suspect it too is in our top octave.
Without overall NFB, or deliberate trickery, a resistance-coupled amplifier's small-signal and large-signal bandwidths are "the same" for practical purpose. Here the small-signal bandwidth has been selected to run up into the top-octave. And there are few musical signals where top-octave signals dominate. If the sidechain "works" well past 6KHz, it will limit music "correctly", within the inherent slop of an LDR limiter (it will never brick-wall).
But as later noted, the LA-2a sidechain has NFB, and could slew-limit. So? What is the RMS of a triangle versus a sine or a saxophone? Until you get _far_ into slew, the RMS or average is not very different even when the wave looks bad to the eye.
> what the hell is that big notch for?
This is an older speech/music radio limiter. You want full midband power, so you slug the midband through the sidechain. You don't want essss power; de-essers are limiters with highs boosted in the sidechain, this gives a modest amount of de-essing. The bass lift sorta-corrects for the apparent poor light for bass drive. None of this appears to be deeply thought-out or sorted-out; it all interacts. You set your midrange for strong modulation, you trim your HF cap to reduce esss-splatter, done. It was more than another decade before radio got so nasty that many-band phase-twisting computer limiters became hot chit.
> vari-mu side chains to have plenty of peak
As you thought-through: not really the same thing. The time-constant is 99% in the CdS. Also we could drive with an LED or a Neon, which have negligible capacitance and fairly flat response. The older EL was "handy", "new" (for 1959), neons have problems and LEDs were hardly a glimmer in a lab. The modest capacitance is a small drag, but 6V6/6AQ5 is a small price.
> using a servo-loop
I built such a thing nearly 30 years back. It used cheap unselected (but semi-matched) LDRs and gave predictable response. It is still in use in my old joint. It was an awful lot of trouble to tweak. If I had understood the LA-2a better, I wudda sorted LDRs for speed and gone simple.
> I don't know much about the EL part.
Insufficient information. It has been observed the EL panel "color" changes with excitation. But we have no hard data, not even semi-firm data, nor any home-shop way to measure it. If someone did plot the EL panel response, then there is the tricky business of multiplying EL response against CdS response, both fairly steep hills and prone to graph-reading errors.
_I_ think it is sensible to treat the EL+CdS as a "black box" (which it is), and not fool around with what happens between.
> I get a full-voltage bandwidth of 18 kHz which is probably fine.
What small-signal response do you get? I suspect it too is in our top octave.
Without overall NFB, or deliberate trickery, a resistance-coupled amplifier's small-signal and large-signal bandwidths are "the same" for practical purpose. Here the small-signal bandwidth has been selected to run up into the top-octave. And there are few musical signals where top-octave signals dominate. If the sidechain "works" well past 6KHz, it will limit music "correctly", within the inherent slop of an LDR limiter (it will never brick-wall).
But as later noted, the LA-2a sidechain has NFB, and could slew-limit. So? What is the RMS of a triangle versus a sine or a saxophone? Until you get _far_ into slew, the RMS or average is not very different even when the wave looks bad to the eye.
> what the hell is that big notch for?
This is an older speech/music radio limiter. You want full midband power, so you slug the midband through the sidechain. You don't want essss power; de-essers are limiters with highs boosted in the sidechain, this gives a modest amount of de-essing. The bass lift sorta-corrects for the apparent poor light for bass drive. None of this appears to be deeply thought-out or sorted-out; it all interacts. You set your midrange for strong modulation, you trim your HF cap to reduce esss-splatter, done. It was more than another decade before radio got so nasty that many-band phase-twisting computer limiters became hot chit.
> vari-mu side chains to have plenty of peak
As you thought-through: not really the same thing. The time-constant is 99% in the CdS. Also we could drive with an LED or a Neon, which have negligible capacitance and fairly flat response. The older EL was "handy", "new" (for 1959), neons have problems and LEDs were hardly a glimmer in a lab. The modest capacitance is a small drag, but 6V6/6AQ5 is a small price.
> using a servo-loop
I built such a thing nearly 30 years back. It used cheap unselected (but semi-matched) LDRs and gave predictable response. It is still in use in my old joint. It was an awful lot of trouble to tweak. If I had understood the LA-2a better, I wudda sorted LDRs for speed and gone simple.
