solid-state la2a-inspired opto compressor

Besides the question "why", does anyone have comments/criticism about this rough draft of an optical compressor?  My goal is to build something with a similar topology to the la2a, but solid state - I like the relatively gentle compression characteristics, but I have a bit of an aversion to tubes.

Thanks in advance for any advice.

EDIT:  there's a revised version below.

Sorry, but without even looking at your design the phrase "LA-3A" comes to mind ;-)  Perhaps you meant "Improved LA-3A Opto Compressor"?

OK, now I've looked at the schematic.  At first glance everything seems to be in order.  But seriously, if you haven't looked at the LA-3A circuit then you are ignoring a lot of existing research in this area.

Well, it's actually more of a la4, except with the autoformer EL panel driver for the t4.

Nothing wrong with the concept, except that a lot of the la2a's sonic character comes from tube distortion.

Also, I think U2B output should be connected at the junction between D1 and D2 and C5 should be bypassing RV1.

You could take D1, D2, R7 and R8 out and get a class B amp without a problem here, instead of 100Ω change R10 for a 100µF cap.

You should look for both schems, LA3A and LA4A. LA4A don't use T4B but it has a couple of intresting things for it, LA3A is complete discrete and need 4 TX (PS, IP, OP and autoTX) so not cheap to build as is but you can get something out of it.

JS

+50dB gain in U2A seems a lot, coming from a -10.5dB decreased line feed (-6dB from U1 and -4.5...-inf. from voltage divider R9/RV2). RV3 will be rev.log taper. Just my 2ct.

I'll look at the suggestions more tomorrow, but re: the la3a, the autotransformer driving the EL panel and the meter section are similar to the la3a, but I've tried to keep the gain/compression structure similar to the la2a, thus the 50dB in the sidechain (+the 20dB from the autoformer, given that I've omitted the step-up input transformer from the la2a).  The 100ohms in front of the autoformer is also a (misguided?) attempt to keep roughly the 10kohm impedance feeding the EL panel from the la2a.

More tomorrow, thanks again for the suggestions!

Meter network looks switched-around. Is that really how LA2 does it?

Unless I'm misreading (which is quite possible), I think that's how the la3a does it.

looks right except the zero adjust is 250K, not 100k.

Yes - if I calculated right, the 100k should give a finer adjustment than 250k (the meter needs 200uA when the photocell is dark).  But, I've got a sifam vu meter & kenetek t4b coming in the mail, and so I'll hook it up and make sure about that.

Ok, here's my second draft.

Changes:
made the EL driver simple class b (thanks joaquins!)
changed the capacitor on u2a to 330pF to start roll off the upper audio frequencies - realized that I probably don't want those going to the EL panel
added resistors on U4's output
specified log & rev log on the pots (thanks harpo!)

Re:  C5, isn't the current placement safer as far as stability than putting it around the trimpot?

> that's how the la3a does it.

You are correct.

I still have doubts.

What we really want is 1.23V DC (0VU) through a 68K (same as audio-path series resistor).

While you can get an equivalent in many different ways, for 26V supply I get numbers over 1Meg.

Maybe just so hot here my brain melted.

Yeah, I'm not terribly confident about the meter section.  Once the T4b comes in the mail I plan to try it out and see.

One other thing I don't know about:  right now, there's the potential for hitting the EL panel with some rather large voltages.  Would it be a good idea to clamp the signal somewhere before the autoformer to keep from blasting the optical cell?  I don't really know how durable the EL panel is and what sort of signal is necessary to damage it.

I've always wondered why they do that in the LA2A and 3A? In the 2A, I think the meter circuit is up at around 60V Or whatever that neon bulb is regulating to) Was it a simple matter of whatever stable voltage reference was cheap and available?

> potential for hitting the EL panel with some rather large voltages.  Would it be a good idea to clamp the signal

Large voltage to EL = bright EL.
Bright EL = light on LDR.
Light on LDR = less signal to out and sidechain.
Less signal in sidechain = less voltage on EL.

It's self-"clamping", isn't it?

I used to design for over 40mA in the LEDs. On tone or signal I could never prove I had more than 2mA-5mA happening. Reverse deduction suggests it should have gone to max current while the LDR was reacting, but when the light is bright the LDR reacts pretty darn quick. (My metering was not super-speedy.)

Or forget my logic. LA2a has been around for 50 years; LA3 for 40 years. Do we have a pile of poofed EL panels? They may die, but the older ELs also died when plugged into steady 117V as nightlights: they are electro-chemical, they are made in bulk cheap, they have defects and impurities and won't live forever. OTOH the fact they did night-light 24/7 for a decade suggests that limiting 10% of time 4 hours a day they will last 600 years, and most of us won't care.

PRR... always bring us back to reality.

PRR, you're too smart!

As another juicy little tidbit, Bill at Kenetek estimates that the EL panel is gravy up to 130Vrms;  I'm planning on 17V rails, so it'll still be copacetic when some gorilla cranks the threshold knob and blasts the input with +24dbu (the sidechain amp will still be slamming the rails even with the photocell fully on, no?).

Just tried out the metering, and it does work - RV4, however, has to be close to 250kohms (as in the LA3A) at +/- 17V rails.  Gain reduction on the meter is within a dB or two of reality, which seems good enough to me.

So, I got a pcb made, wired everything up, and it works, kind of!  It's compressing, the meter's working, controls all seem to work right,

BUT

if I crank up the makeup gain, I get rail to rail hf oscillation.  Interestingly (to me), it starts to oscillate at a lower gain setting when there's more compression.  If I disconnect the autoformer in the sidechain, it doesn't seem to oscillate at all.  There are rather large voltages coming out of the autoformer;  perhaps this is screwing up my grounding?

Before I really get into it, does anyone have a suggestion as to where I might start?  Schematic is attached.

xoxo

You might be getting feedback from your el-driver signal into an early gain stage. Check your routing in that regard.