LDR Question, LA-Light

[Snodgrass]

I have ordered parts to build this:

I've got trim pots for the bass compensation and LED current limiting resistor. Anyhow, I found rather quickly that tracking down an ideal LDR is not so easy and that interpreting the datasheets was a little squirrely.  I ended up going with a NSL-7910-ND which has
Dark Resistance: 1.5MΩ after 5s in the dark
Light Resistance: 7.5kΩ typical at 2 foot candles, 800Ω typical at 100 foot candles.

The designer specified in a thread on this very forum that an LDR which would go to 2kΩ in the sun would work fine.  I have found that it is difficult to find LDRs which range very low to rather high and the NSL-7910 was the lowest resistance one I could find that still had a suitable high resistance when dark.

I took out a second mortgage and bought two of the aforementioned LDRs (5 bucks a pop!) hoping that one of them might be more suitable than the other.  How many foot candles is an led?  Does it matter? Did I mess up here?, is what I am asking. Normally I can make precise determination of what component values and properties I will need.  This LDR thing seems a bit like throwing blindly at a dartboard.

[Snodgrass]

Returning to bump my own dead thread with a followup question.  I have built this and I mostly like what it does, though it is subtle-to-nigh-undetectable in the comp position or with the Compression knob below about 75%.  My questions are these: do I really need the current limiting resistor in front of the LEDs?  I am using a 24V (+/-12V) supply. My understanding from the datasheet is the TL072 can only put out 26mA.  That said I do not understand the designer's recommendation of 1k for a 9V supply when the TL072 can only put out about 3V peak on such a supply. It seems that if a resistor is required at all it would be about 470 for my use case and much smaller for a 9V supply.  Am I missing something?

[ElectricDruid]

The schematic specifically says "low current red LEDs" which might only need a few mA. Certainly nothing like 26mA - that's old 1980's LED current! You'll get *slightly* more than 3V with a TL072 on a 9V supply. I generally reckon about 7V, so +/-3.5Vpp. Minus the forward voltage of the LED (say 1.8V) gives us about 1.7V, so 1K is only giving us 1.7mA. That *is* pretty low, but the 1K was a starting point, and I read it as "you'll be safe with this, but experiment going lower".

What concerns me more is the 1K/100K around that amp, giving a gain of x100. With a 7Vpp maximum output, that means that any input signal over 70mV is going to start clipping, at which point, the sidechain has run out of dynamic range. And 70mV is pretty small for a guitar signal. My friend's old Strat with fairly "traditional" single-coil pickups puts out around 100mV on the higher strings, and up to 1V on the wound strings.

[Snodgrass]

I just used the LEDs I happened to have which are modern and came in a kit of different colors. Should I have been more selective about this? I rather figured modern LEDs were created more or less equal.
I assumed the 100x gain was so that you could just squash the crap out of signal if you wanted to or you could be more tasteful with the compression knob if you preferred. That said I am using a linear pot there and have it at around 80% turned.  It is very possible that I have a terrible ear for compression and am not using the thing right, though I am using a 24V supply.
Most LEDs have a 20mA maximum rating, is this the max I should be shooting for?
Also I always sort of assumed passive pickup output was always in the sub 100mV range. I am not sure what if anything I based this on.  I probably just made it up from nothing.

[merlinb]

You want an LDR with low resistance in light, the lower it is, the more compression you can get. The dark resistance isn't that critical since the ratio of dark/light resistance is kind the same for most LDRs. GL5506 or GL5516 would probably be ideal.

You want the brightest (i.e. most efficient) LEDs you can get. You can always make them dimmer with the limiting resistor. Max LED current is irrelevant, that circuit can only deliver a few milliamps anyway. You can reduce the limiting resistor all the way to zero ohms and the LEDs will be fine, see how it sounds.

You can also get rid of the 68k shunt resistor at the input, it's not really doing anything useful except loading down your guitar.

[Snodgrass]

that circuit can only deliver a few milliamps anyway.

Can you explain this statement? I assumed it could conceivably deliver the rated short circuit output current of a TL072, which is I believe the 26mA cited above.
The idea about the 68k is well taken and I am not sure I even included it in my build. I had also considered upping the 68k series resistor so the LDR wouldn't have to go so low if tinkering with the LED/opto had no result. The problem I have is that I find tinkering so much more tedious when PCBs are involved rather than eyelet boards and terminal strips that I am quite averse to doing so.

[merlinb]

I only just noticed you're using a bipolar 12V supply so I assume your LEDs go directly to ground, not limited by the Vref divider. So yeah you might get close to the short circuit current of the TL.

[PRR]

The 68k -series- resistor before the LDR stuff means the LED will never "have" to be driven very hard. A mA or two in a healthy LED will drop an LDR far below 68k, and high attenuation. Since this IS a feedback loop, it won't drive the LED any harder, not on steady-state.

Don't worry about blowing the LED or '072. Satisfaction will need carefully selected LDR to get ear-pleasing time constants. Try-and-see works much better than theory.

A gated signal generator (beep beep beep) is very useful on the bench.

[Snodgrass]

Satisfaction will need carefully selected LDR to get ear-pleasing time constants. Try-and-see works much better than theory.

Most unwelcome news.
A good thing is that I already like what it does, I just wish it did it a little earlier on the controls so I had more range of adjustment. And maybe that I could see some sort of feedback telling me when action is happening because my personal sensitivity to gentle compression seems to be deficient. I will probably drop the current limiting resistor, put some fancy high efficiency LEDs in and see how that suits me. Failing that I gravely fear some actual work and testing is in order.

[Snodgrass]

Say... while I have such worthy gentlemen as these responding to my thread I would like to ask: I realize the TL072 is a celebrated Swiss army knife of a gain element especially among stompbox enjoyers, but if I were able to choose from any possible active component(s) to boost the sidechain signal and drive those LEDs, which would be the most choice? Even say I didn't care about whether it was discrete components or not. Would say a JFET minibooster do better? BJTs? Mosfets? Triodes are probably out due to their impedance. Opamps are simple and useful in their way but just for fun is there something more appropriate?

[PRR]

No.

[Matthew Sanford]

How did you set up the LDR/LED combo? Taped together like a vactrol?

[amptramp]

If you have an oscilloscope available, check the voltage at the 100 µF capacitor in the Vcc/2 divider.  You may need something more substantial there, like an op amp follower for this voltage.

The 68 K resistor from the input to Vcc/2 is unnecessary and loads down the input excessively for no reason.  It should be removed.

[Snodgrass]

How did you set up the LDR/LED combo? Taped together like a vactrol?

I cut a circular piece of lexan about 10mm diameter and hot glued the two LEDs to one side and shrinkwrapped and double shrink wrapped the LED/lexan/LDR suite all together so the LEDs face the LDR dead on through the lexan.  Looks kind of like a bigger version of the Fender tremolo dealy from the outside.

If you have an oscilloscope available, check the voltage at the 100 µF capacitor in the Vcc/2 divider.  You may need something more substantial there, like an op amp follower for this voltage.

I think if/when I redo this thing I will probably do just that.  Not really in the cards for this iteration since I'm sort of stuck with the pcb I had made.

[moosapotamus]

... And maybe that I could see some sort of feedback telling me when action is happening because my personal sensitivity to gentle compression seems to be deficient...

As I recall, the last version of the LA_Light that Johan posted here included a VU meter.

https://moosapotamus.net/wp-content/uploads/2023/11/LA_Light_last_one_001.gif

I imagine it might be possible to replace the VU with some sort of LED arrangement, maybe?

[Snodgrass]

Well I reckon I have a rebuild to do with a new PCB and an '074. I don't see why I couldn't swap that VU for two anti-parallel LEDs, although a VU would be pretty sweet if I could fit it all together. I should have just bitten the bullet to begin with and used that more deluxe schematic.