LA4 RMS detector

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Kit

Well-known member
Joined
Mar 13, 2006
Messages
467
Location
Sweden
Im working on a optocomp/limiter not to dissimilar from the la4 sidechain.
In the manual it states the following:

The gain reduction is accomplished with an RMS detector,
which is the analog of human hearing, so gain reduction is smooth
and natural sounding.

I guess they are reffering to the C6/R29 combo in the feedback loop of the led driver opamp.

Ive read somewhere that this also helps to simulate the "afterglow" of the
EL-panel found in the T4 module.

Anyone got any insight in this matter?

http://www.waltzingbear.com/Schematics/Urei/LA_4.htm
 
It certainly has nothing to do with RMS.. That is just marketing gas..

Ah, thats kinda what i suspected.

How about the afterglow thing then?

To me it looks like that cap will charge/discharge through R29 and surrounding resistances wich changes with different ratios.
There is a timeconstant here, but it will be inverted.
I fail to see how this will simulate the "afterglow" thing though.
 
[quote author="gyraf"]It certainly has nothing to do with RMS.. That is just marketing gas..

Jakob E.[/quote]

Agree. In fact I know Brad Plunkett thinks rms-based compressors are in general not the way to go.
 
Good points Wayne. I suspect though that the overall phosphor and photoconductor decay dynamics are governed mostly by the photoconductor characteristics, but maybe not. I suppose if LEDs had existed in that era the photoconductors would have been optimized for the LEDs' essentially zero decay time.

As far as using the incandescent lamp, even if it were fast enough responding, the spectrum is so enormous and variable that the net transfer function, due to that and the spectral response of the photoconductor would likely not be a good fit to r.m.s. except possibly over a narrow range.

When you say LavaLamp are you referring to the tri-color LED?
 
I think one of the problems with the LA-4 is that LEDs aren't really good at that.

I hear you, but the EL-panel of the T4 is pretty useless in cheap low voltage circuits.

One thing i like about the "precision rectifier/virtual earth led driver" setup is that it has the potential for adjusting threshold without affecting ratio and vice verca.
But the problem with these type of designs is that "sound" lies mostly at the mercy of the LDR timeconstant.

One could get one slow and one fast LDR to simulate a "stacked"
release.
Any suggestions on good LDR´s?
The R29/C6 network just appear to provide HF rolloff
.

Im not so sure, such an odd way to do that.
It would be easier just to strap a cap around IC2C.
 
[quote author="mediatechnology"]From the LA-4 schematic: The driver for the LED is a constant current driver. Summed voltages from the absolute value circuit, IC2, CR4/5 et al, are conveted to current to drive the LEDs. This is to linearize the light output vs Vin. [/quote]

Makes sense.

The R29/C6 network just appear to provide HF rolloff.

But it's *after* the rectifier (absolute value circuit). While theoretically this can also be described in the frequency domain ("HF rolloff"), we're dealing with a control voltage here, and not with audio anymore. (As opposed to the LA-2 and -3) So the RC combination really has an effect on the time constants of compressing, not on the frequency response. Modelling some "afterglow" with a faster device and additional time constants is what it looks to me indeed.
(And yes, I also suspect the long time constants were in the LA-2 / -3's LDR, not in the EL panel, but not 100% sure.)


Without knowing the value of R28 it's hard to say if it provides integration or not.

Right. But even the values of C6 and R29 give us some clue about which frequency range (or time constant range) the whole combination will work.
Speaking sloppily, I'd say it's more of a bass boost than a HF rolloff.
Which means, in time domain, it allows *some* percentage of the Vactrol's (relatively) fast action to come thru, and slowing down the rest.
To determine how much of either, we must know the value of R28.

BTW, in all ratio settings except the highest (20:1), the "afterglow network" has little effect, as it's paralleled by a total resistance that is smaller than R29 alone, and the series connection of C6 and R29 is always higher in impedance than R29, at all frequencies.

R35 is probably there to protect the LEDs in a fault condition (250ma) and C7 appears to be for bypass.

There's more fault protection from R34 than from R35.
R35 helps to prevent the distorted (rectified) strong signal (LED current) to invade the positive PSU rail, together with the bypass capacitor.

JH.
 
I also suspect the long time constants were in the LA-2 / -3's LDR, not in the EL panel, but not 100% sure

You could be right. In any opto comp the LDR is gonna have a major impact on sound. The exception being the really fast devices, driven with "real" RC-time constants.

I have no experiense with LA2/3 but many people do rave about the "magic" being in the EL-panel. Then again i hear some people are substituting the EL for a anti-parallel led combo with no percived difference in sound.

Ill have to read through the LA2A meta to see if i can find some additional info.
 
. I'm sure a lot of the cost in the T4 are the cells that can't be used and the labor in finding the ones that can.

Indeed, and the sidechain also becomes more "expensive" in that we need high voltage to drive the ELP. Theres no way around it.
BUT, if it turns out that the time constants of the ELP is not important, then why bother.
 
[quote author="mediatechnology"]My point was more that R35's not an afterglow thing.[/quote]

Oh yes right. Perfect current source to drive the LEDs - no influence of that resistor on the time constants.

I'm not sure who originally coined "afterglow," was it UREI? It seems to suggest that it was an attempt to emulate the T4. Which makes me wonder if there is indeed an afterglow associated with the EL at high ratios. Now I think everyone agrees that afterglow linguistically sounds better than "hangover" - which is what the CdS cells do. If the Vactrol cells were similar to the T4B cells in being the dominate time constant would we need afterglow? Maybe. Or, maybe we call it afterglow because it sounds better that "Cadmiun Sulfide Hangover."

Ha ha, I like this!

Just like RMS sounds better than mean value detector.

JH.
 
I think the EL element is not such a light load at high frequencies - this has been subject of earlier threads.

Other than that, I think it's highly interesting what you write. Especially the second order integration idea.

JH.
 
[quote author="mediatechnology"]The EL is quite capacitive isn't it? A "70V" transformer might work then. [/quote]

Capacitive, yes. What is a "70V" transformer?

My guess is that the EL, maybe by virtue of phosphor, smoothes small step changes better than the CdS at higher light levels. Will advise.

Interesting.
Somewhat related, I can confirm that there are "vactrols" that are too fast for compressor/limiter functions without some extra time constants.

I've built a (very crude) compression function into a gain stage that will be used on a patchable synthesizer (actually, for my Matrix FX project). You can see it here:

http://www.oldcrows.net/%7Ejhaible/matrix_fx/jh_matrix_fx_wavefolder_invamp_sch.pdf

Amp output driving the LED via bridge rectifier, LDR attenuating the amp input voltage. I'm using a NSL32SR3 (very low impedance device), and this is way too fast all by itself. Connecting a 100uF cap across the LED gives smooth limiting. Without that cap, there's ugly distortion.

JH.
 
[quote author="mediatechnology"]A "70V" transformer is used in the so-called constant voltage distributed speaker systems common to paging and background music systems. They are step-up sometimes auto-formers. There are 25V systems too. These may not be used in Europe.[/quote]

Ah, I see. We have 100V systems in Germany.

JH.
 
You can find some informations about EL panel here.

Image1.gif
 
Some time ago i built something similiar to the LA4 sidechain using an NSL32 optpcoupler from silonex . If i remember it correctly it sounded allright.

Im gonna try it again, but this time i think ill order a bunch of LDR´s and sort them out to find a "sweet" one.

Pretty good prices on these CDS ones:

http://www.electrokit.se/item_list.php?maingroup=Motst%E5nd&secondgroup=Fotomotst%E5nd+LDR

Dark R=1M
On R @ 10 lux= 2-5K
Rise and fall time= 30-40ms
 
For a more natural sounding compressor, you can try this:

http://www.physorg.com/news151.html


There are some decay curves here:

http://www.photek.com/characteristics/Phosphor/P20decay.pdf
 
Back
Top