LED popping/clicking, a solution?

[Toney]

 This is intended for situations where you have a popping LED that just wont quit!
Its a current balanced diff amp so whether off or on, the current between the rails remains the same. Obviously this wont help with popping passing through the pedal circuit itself but in situations where you have isolated the LED as the culprit but just can't beat it, this may work. Some single transistor designs, crappy el-cheapo switches with bouncing/grubby contacts etc. I have been there and been frustrated!
Its sudden current rushes that can upset things as an LED is switched in if the design is sensitive to it, so this is intended to fix that.
Heres the schem:



 When one side is on, the other is off. Current is flowing either through the 2n3906 or the 2n3904 with LED. The 2n3906 is biased to equal the current draw of the 2n3904 with attached LED. If you want to, you can replace the 1k going to the 2n3906's emmiter with a second LED. Perhaps something along the lines of "power on" and "bypass"  but the main point is that both sides draw equal current.

And a Vero:



The actual board is only 1.8 x 1.6cm so it should be small enough to tack-on unobtrusively in most builds.

Please note:
I haven't tried it in a build yet but it seems to work on the breadboard.

[JustinFun]

This is great, thanks!

I've got some flashing LEDs that I'd like to use in some projects (just 'cause, that's why) but I haven't been able to get round the ticking they make as they turn on and off. I'm going to give this a try and see if it works.

[Toney]

Well this relies on keeping the current flow even, that's the cause of a lot of LED popping issues. The flashing LEDs will still cause current rises and falls with this, unfortunately.
You will have to try some of the regular approaches. AMZ has some methods you can try.

[amptramp]

This seems to be a lot of complication to switch a LED.  In fact, you could put in a voltage regulator and run from a higher input voltage and use fewer parts and have the benefits of a regulated supply.

Or you could have a resistor in series with the LED and have a switch across the LED so when the switch turns on, the LED is shorted out.  If the voltage across a blue LED still creates too much of a difference in current, put a green LED in series with the switch so the green LED is in parallel with (and shorts out) the blue LED.  The green LED need not be visible.  I have a pair of green LED's used as circuit elements in a project I am debugging now and they never see the light of day (or the outide of the box) because they are not used as indicators.

You could also use a power MOSFET with a long R-C time constant in the gate circuitry to turn a LED on or off gradually.  This has the benefit of drawing no current when the LED is off.  Use a resistor in series with the LED, of course.

[Seljer]

Or you could setup a constant source with a transistor, [zener] diode and two resistors, then short out the LED when you want to turn it off. The current would remain the same all the time.

[R.G.]

Yeah, it works. I've done this before - and I thought I'd posted about it here. You're right, with a diffamp, the current stays constant, but is steered into different paths. This is the basis of the once-great  () current steering logic families. Done properly, these families had nearly zero power supply noise because the current never changed, only went to different paths.

The constant-current source with a clamp on the LED to steer the current around the LED works the same way.

Another way that works is to wire the current paths so that current changes in the LED wires do not flow through the wires supplying power or ground to your pedal circuit. If you do that, no change in ground or power wires is possible, and no click occurs unless you have a high impedance power supply itself, and decoupling on your pedal fixes that. Mostly. 

[Toney]

 All of this was motivated by an exquisitely sensitive Rangemaster.
That thing popped and clicked no matter what I threw at it, even after trying all the regular tricks. The circuit was normal in every way, voltages all spot on, tested everything from every angle. It sounded damn good, but it just popped!. The 3pdts I bought may be part of the issue. I tried some of the cheapest Chinese ones, just a small batch, but I think they really added to the problem. They test fine for continuity but I suspect the contacts of uneven switching and possibly bouncing at the point of contact. So I figured, OK, I will try to solve it for the worst case scenario. Kind of a challenge to defeat the pop!
It actually had two issues. The LED and some almost microphone type of switch chatter that became apparent at louder volumes,  I didn't at first notice it with my practice amp but it was annoying enough to motivate me to look for solutions.
The first was to slow down the switching of the actual circuit using a Jfet to slow-transition the audio pathway. The delay was enough for the switch to settle before releasing the signal a split second after the contacts.
The thread is here:
http://www.diystompboxes.com/smfforum/index.php?topic=97299.0

That worked great but I swore like blue monkey when I reconnected the LED and yep, pop.
So, I looked at ways to switch the LED with the minimum current disturbance
A ridiculous amount of time on one circuit, yes, but I kind of took it as a puzzle to be solved and hopefully learn a few new tricks along the way.

@ Amptramp
I went with switching the bases of BJT's to minimize the current flowing through the switch contacts. I like the idea of you mentioned of a "power MOSFET with a long R-C time constant in the gate circuitry to turn a LED on or off gradually". I was looking for a simple method of slow transitioning the LED at first but couldn't think of away to do it.
Do you have an example?
@R.G.
 I saw you suggest this when reading up on the subject somewhere. I searched but couldn't see any examples, so just messed about on the breadboard until I found a set of values that seemed to work reliably. I am not sure about how to implement separate current pathways? Do you have any reading or examples you can point to?

[R.G.]

@R.G.
 I saw you suggest this when reading up on the subject somewhere. I searched but couldn't see any examples, so just messed about on the breadboard until I found a set of values that seemed to work reliably. I am not sure about how to implement separate current pathways? Do you have any reading or examples you can point to?

That is what you implemented. The point of a diffamp, in any of the several styles, is that there is a fixed amount of current going through the pair. What happens on the control nodes affects which of the two (and possibly more) pathways the current goes through. In your version, the two current paths are through either the NPN or the PNP. In a classical two-of-the-same-kind diffamp, it's which of the two active amplifiers gets the current.

The practical issues are how to make the overall current be constant. A classical diffamp with a current source on the two emitters gets very, very close. I just did a bit of sim on an LED driven by an NPN diffamp with either a resistor "current source" in the emitters or a two-transistor current source. The resistor version had a significant change in total current. The two-transistor CCS got the total current variation down to about 20uA. The resistor current source had a variation up around 500uA, both with a 20ma LED current.

Your version uses a PNP to steal the current from the LED path; same thing.