led brightness controlled by guitar signal "strength"

[candidate]

From my understanding, I'm most familiar with this design as implemented in the Mu Tron III, which has a half wave rectifier to help determine the envelope of the signal.  Are there better ways of controlling the brightness of an led with respect to the guitar signal?  Not looking for supa fancy Leds that can see through time.  Just wondering what other design implementations are available out there.

Thanks

[frequencycentral]

The Midfi Clari(not) does something similar: http://www.diystompboxes.com/smfforum/index.php?topic=74294.0

[Processaurus]

The front end of Tim Escobedo's Uglyface does it with an LM386 and a 1K pot pretty well, and simple.

[merlinb]

A peak level detector is a classic arrangement, fast response and very high input impedance

(ignore the second opamp and transistor, it was just the first diagram I found via google). However, you would need to amplify the signal first, quite a bit, to get it large enough to drive a LED.

[candidate]

Well I suppose the next question is how many of these qualify as "peak level detectors" and which ones have the 'smoothest' most 'responsive' ramp.  Eliminating 'ripple' seems like a worthwhile endeavor if led accuracy is a priority, but at the same time ldrs are pretty forgiving. 

[R.G.]

Well I suppose the next question is how many of these qualify as "peak level detectors" and which ones have the 'smoothest' most 'responsive' ramp.  Eliminating 'ripple' seems like a worthwhile endeavor if led accuracy is a priority, but at the same time ldrs are pretty forgiving. 

Your eye is way too slow to see the response of even an un-filtered peak detector working on audio. Your eye can't see, for instance, the pulsing of an LED run directly at power line frequency. Smoothing is not an issue.

Converting signal size to current in the LED is. An LED doesn't conduct at all until you feed it the 1.4 to 3V it needs to turn on (exact voltage depends on the LED) and thereafter it emits light in rough proportion to the current through it, while the voltage doesn't much change at all. Feeding an LED a voltage through a resistor doesn't do much good for your purposes unless the LED voltage is trivial compared to the size of the signal fed to the resistor plus LED.

A simple way to get this done is to take an LED and put it in the feedback path of an opamp. Feed your signal into the (-) input through a resistor, having suitably buffered and amplified it first. The signal voltage now causes a current to flow in the input resistor. The opamp forces an equal but opposite current to flow in its feedback path, which is the LED. You'll need to put a reverse-paralleled diode in parallel with the LED to keep the opamp and LED from fighting it out when the signal is the "wrong" polarity. This way, the LED is fed a current, which makes its response much more linear, and it puts out pulses of light at signal frequency, which is way too high for you to see as anything but smooth. No additional filtering is needed.

The only difficulties are the practical ones. You have to feed enough signal current into the opamp's input to run the LED, and the opamp is feeding the LED an equal amount. So the current from the power supply is double what you'd otherwise need. And it's not as bright as an LED fed an amplified and smoothed signal because it's on for less than half the time. Still, with one of today's super-bright LEDs, the problem isn't all that bad.

And it's very quick and simple.

[Gurner]

Here's one (ugly) simple option...

http://www.youtube.com/watch?v=JtR5kkf7ipw

personally, I'd have gone for a Peak detect cct followed on by $1 PIC to tackle a lot of the issues.

(the PIC to the assess DC level & then adjust the duty cycle of a PWM stream to an LED. If multiple LEDs wanted, Use a tranny on the PIC output)

[Processaurus]

A simple way to get this done is to take an LED and put it in the feedback path of an opamp. Feed your signal into the (-) input through a resistor, having suitably buffered and amplified it first. The signal voltage now causes a current to flow in the input resistor. The opamp forces an equal but opposite current to flow in its feedback path, which is the LED. You'll need to put a reverse-paralleled diode in parallel with the LED to keep the opamp and LED from fighting it out when the signal is the "wrong" polarity. This way, the LED is fed a current, which makes its response much more linear, and it puts out pulses of light at signal frequency, which is way too high for you to see as anything but smooth. No additional filtering is needed.

The only difficulties are the practical ones. You have to feed enough signal current into the opamp's input to run the LED, and the opamp is feeding the LED an equal amount. So the current from the power supply is double what you'd otherwise need. And it's not as bright as an LED fed an amplified and smoothed signal because it's on for less than half the time. Still, with one of today's super-bright LEDs, the problem isn't all that bad.

Excellent, substituting an LED into the classic "Super Diode" opamp circuit (which uses an opamp to create diode with no forward voltage drop).  The LM358 would be a good candidate for that, as it can supply enough current to light an LED nice and bright, and can operate down to the negative rail (so could probably get the right output without having to make a bias supply), and is a dual package, so the first one could be a buffer for the signal, and with its output be able to sink all of the LED current from the series resistor to the next opamp(the super diode circuit)'s inverting input, to light it up.

[candidate]

All very excellent responses.  I can't thank you enough.

Now I suppose the next question is whether 'filament bulbs' could provide a more sensitive representation of the nuances of the signal envelope?  at the expense of more power considerations, of course.

[Processaurus]

Neater looking maybe, but if needing to know more than merely if sound is coming through or not, a VU meter is much more useful (look up LM3915 if you're interested in a low parts count digital VU meter, it is a great chip).

[R.G.]

Now I suppose the next question is whether 'filament bulbs' could provide a more sensitive representation of the nuances of the signal envelope?  at the expense of more power considerations, of course.

Depends on what you mean by "more sensitive". They're far, far slower in response than LEDs. The LEDs would be following every instant of the signal. Any perception of an average light level would be done in your eyes, not the LED. So the representation would be much more time-accurate with the LED. The incandescent bulb's light output is a very nonlinear function of the applied voltage and current. But its thermal mass does average the applied signal in a visually appealing way - if not accurate. Basically incandescent equal no nuances at all - but that may be pleasing, instead of sensitive as I think you mean it.

[SpencerPedals]

I am currently able to recognize patterns and tweak fuzz circuits as a result of the addiction; my degree is not in electronics and I am not a numbers guy yet.  Disclaimer in place, I was thinking about the original question posed here a month or so ago in regards to it's use to functionally "tweak a knob" by the striking of notes, etc.  Is this something that's used in more circuits than I thought?  At a glance, it seems like you could get some good fading/blooming fuzz using this sort of method.  Really any of the fuzz increasing or decreasing tricks seem like they would be well-suited by this. 

[R.G.]

Envelope control is fairly common. Not hard to figure out - generate a voltage that's an analog of the average loudness of the music, then use that voltage to control stuff. Synths are fairly loaded with examples, guitar effects less so. But it's not that hard.

The envelope controlled resonant filter is the commonest example, in the form of the Mutron III and other envelope controlled wah/filter circuits. Some LFO controlled circuits use the envelope to control speed. You could certainly do an envelope controlled fuzz intensity, although I think that an envelope controlled fadeover to a fuzz channel would be more useful.

So sure - go for it. It's not all that common, but not unknown.

[candidate]

You could certainly do an envelope controlled fuzz intensity,

by altering the resistance to something like clipping diodes?

although I think that an envelope controlled fadeover to a fuzz channel would be more useful.

used like a compressor or actually re-routing the signal?

[SpencerPedals]

At first I thought clipping diodes when I was pondering it, but you're going to get volume drop by doing that.  I was thinking altering resistance between stages, emitters to ground, or like R.G. mentioned, a blend feature.  I was thinking a re-routing and I'm assuming he was as well.  Using it to control blend would really open the door to a lot of interesting fading.