LED indicator causing clicks

[amz-fx]

LED indicator causing clicks...  a short article on preventing them:

http://www.muzique.com/lab/led.htm

regards, Jack

[Eirik]

Wow that's grat! I've had this problem with several builds, and I allways thought the pulldown resistors didn't do their job. This might be the solution.
Thanks Jack!

[zachary vex]

maybe i'm insane, but that circuit looks like it will create a MUCH greater inrush current when the switch is turned on.

here's the thought experiment:  with the switch off, the capacitor charges up to full battery voltage through the 4.7k resistor.  next, the switch is thrown, and the full battery voltage that was on the cap RUSHES through the 390 ohm resistor at an instantaneous level more than 10 times the inrush current of a 4.7k resistor alone, causing an enormous click, imho, since that massive inrush current passes so close to the audio wiring in the switch.  the 23 mA spike won't destroy the LED, but it will make it flash very brightly for a moment.

the real solution is to put a small resistor (depending on the circuit's needs) in series with the 9V battery to a 47uF cap to ground, and use that for the audio circuit... then connect the led using a much larger resistor, say 15k, and an ultra-bright led, to reduce wasted current (i can power a whole pedal with what an led draws through a 4.7k resistor... 2mA!), and power that directly off the battery, carefully running your control wires away from audio wires inside the enclosure.  the battery may see the inrush current to the LED that way, but the audio circuit will be operating off the isolated 47uF cap as its supply.

[R.G.]

LED indicator causing clicks...  a short article on preventing them:

Aw... did you draw that up for me? Thanks!
That is what I was taking about when I posted on this issue here:
http://www.diystompboxes.com/smfforum/index.php?topic=41059.msg319202#msg319202
Best suppression of pops is when you split the LED resistor in half and put the cap there - it slows down the edge of the current waveform most, because the transient edge is smoothed into and out of the capacitor by the same equivalent resistance.

And it's all about controlling the edge of the current transient. As I noted in my post, the mechanism of the LED switch pop is sudden change in power supply or ground - and usually ground, given the way most pedals are wired - voltage and the reflection of that transient into the active circuit. The semi-standard use of the input jack ground wire to supply ground current for the entire pedal including the LED builds in the LED switch current as a ground transient. It's a poor wiring practice that just perpetuates itself.

What makes this hard to realize is that people think that power filtering on the B+ to ground at the pedal should fix it. Unfortunately, decoupling at the effect circuit can't do anything about the ground lead transient. The ground lead transient is a sudden voltage that appears in series with the effect input signal. Decoupling on the effect board can't touch it.

Use of a PNP transistor in the B+ lead with a 10K resistor from the base which gets pulled to ground by that same input jack ground lead does the same job and is pop-free if you then hard wire the battery ground to the board. This is because the suddenly changing LED current does not share the input ground path. Use of the output jack for grounding the pedal current transfers the problem to the next pedal in the chain - which might help if the next effect has a lower gain, but still better to fix the problem than paper over it. Even this fix can be made to pop if the circuit has too little decoupling on the B+ line and the circuit is sensitive.

It's all  about controlling the LED current edge. If  you slow the edge down enough, the "pop" is converted to a slow wiggle and gets filtered out of the audio because the rest of the audio chain and the human ear can't hear it.

maybe i'm insane, but that circuit looks like it will create a MUCH greater inrush current when the switch is turned on.

It depends on the relative sizes of the split LED resistor. You control the inrush with one of them, the outrush with the other.

the real solution is to put a small resistor (depending on the circuit's needs) in series with the 9V battery to a 47uF cap to ground, and use that for the audio circuit... then connect the led using a much larger resistor, say 15k, and an ultra-bright led, to reduce wasted current (i can power a whole pedal with what an led draws through a 4.7k resistor... 2mA!), and power that directly off the battery, carefully running your control wires away from audio wires inside the enclosure.  the battery may see the inrush current to the LED that way, but the audio circuit will be operating off the isolated 47uF cap as its supply.

That is in fact another version of the split LED resistor circuit, just rearranged.

Consider that for transient suppression purposes, capacitors are equal to batteries - the instantaneous voltage on the caps does not change, only the current out of them. The effect circuit may be assumed to be zero current, only a voltage step sensor. It pops when it detects a voltage step. So the effect circuit and the LED are still separated by a series of resistors and caps. One can do a better or worse job of picking resistors and caps, of course.

carefully running your control wires away from audio wires inside the enclosure.

And we do have a winner here.
The current and voltage transients can still cause pops if you run them too close to sensitive inputs, either by capacitive couping or by inductive coupling. It is almost always possible to route your wiring to cause yourself grief.

It's almost as if Mother Nature sets down some rules and then expects us to follow all of them, isn't it?

[zachary vex]

It depends on the relative sizes of the split LED resistor. You control the inrush with one of them, the outrush with the other.

that's funny, R.G.  thanks.  there is a specific value in the circuit.  i'll post it here in case you overlooked that fact.



the inrush current is determined by E=IR, or 9/390= 23 mA, which is ten times higher than the 4.7k resistor by itself.  if you "split" the 4.7 in two, the inrush current would still be double the value of the 4.7k by itself, doubling the EMP at the switch by the audio wiring.

That is in fact another version of the split LED resistor circuit, just rearranged.

rearranged into a completely different circuit, yes.

[R.G.]

Hey, thanks, zv.

Now I gotta get back to tracing out these pedals on the bench.

[zachary vex]

how original!  8^)

[R.G.]

Like they used to say in the Ozarks, sometimes even a blind hog finds an acorn.

[TELEFUNKON]

far east example here:
http://www.geocities.com/george_giblet/effects/Flanger_FL-01_FLG-1.png
in the lower right corner

[Processaurus]

One brute force solution to avoiding ground transients on the input jack is to use a different jack, instead of the old stereo input jack trick.  The switchcraft #13 jack for instance is a good alternative for pedal effects because it has an isolated SPST switch (a "make" circuit) that can be used to turn the battery on when you plug in.  You then avoid any LED transients or also any other noise that might come from the circuit drawing power suddenly along the ground line that the input is sharing ( charge pumps that operate in the audio frequency for example).

 

far east example here:
http://www.geocities.com/george_giblet/effects/Flanger_FL-01_FLG-1.png
in the lower right corner

Whats the zener diode for?  Note they implemented the idea like RG suggested, with equal resistors before and after the 10uF LED decoupling cap.

[TELEFUNKON]

Note they implemented the idea like RG suggested, with equal resistors before and after the 10uF LED decoupling cap.

should that read: "R.G. suggested what they had implemented" ?

[zachary vex]

Whats the zener diode for? 

most likely to cause the LED to extinguish when the battery gets low, warning the user to replace it.  with just current-limiting resistors LEDs will continue to glow even when the battery is too dead to operate a circuit like that properly.

[zachary vex]

Like they used to say in the Ozarks, sometimes even a blind hog finds an acorn.

in the land of the blind, the one-eyed hog is king.

[Processaurus]

Note they implemented the idea like RG suggested, with equal resistors before and after the 10uF LED decoupling cap.

should that read: "R.G. suggested what they had implemented" ?

Close, that should have read "Suggested R.G. what implemented they had."

Another dreadfully low tech solution: Use the stereo jack trick on the output, rather than the input.  Then you could do the other trick where the input uses the mono shunt switch jack (like the switchcraft 12) to ground the input when there isn't a plug in it, that way no more pops when you want to repatch it.

Whats the zener diode for?

most likely to cause the LED to extinguish when the battery gets low, warning the user to replace it.  with just current-limiting resistors LEDs will continue to glow even when the battery is too dead to operate a circuit like that properly.

Thats pretty cool, in as far as zener diodes can be "cool".  I'm guessing from the part number that it is a 5.6v zener, so would that make the LED turn off when the battery got lower than 5.6v? or would it be the zener's voltage of 5.6v plus the 1.5v voltage drop of the LED = 7.1v?

[zachary vex]

i'm not sure the exact threshold where the LED would extinguish... that entirely depends on the LED's forward voltage and the required current to keep it lit.  what i do know is that if you put a 5.6V Zener in series with the supply, you use a much lower current-limiting resistor value and 5.6V is subtracted from the 9V supply to begin with, changing the lit/unlit threshold dramatically.  i imagine it's something very close to what you said.

[Peter Snowberg]

I see two sources of noise being discussed. The first is any surgre or drop in supply voltage which is then coupled into the audio portions, the second is EMP radiated by power flowing through the indicator.  

[brett]

Hi.
I've had clicks ever since I went from the Millenium 2 to a 3PDT. 
Maybe feed a small current (1M resistor) to a BJT base and small cap (0.01uF) ??  A 0.1 second turn-on should fix things.
cheers

[R.G.]

I see two sources of noise being discussed. The first is any surgre or drop in supply voltage which is then coupled into the audio portions, the second is EMP radiated by power flowing through the indicator.

Surely "discuss" is too dignified a word. 

You are correct. There are two versions of noise there. The first is supply/ground coupled, the second is - well, EMP is probably too strong a word - more like capacitively coupled to sensitive parts. Inductive loop coupling is possible, but the number of turns and the area of the loops is pretty small in most effects.

I focus on ground coupling because in most cases on-circuit-board decoupling suppresses supply lead noise. The ground-lead transient is not and cannot be suppressed by on-board decoupling because it happens directly in series with the input signal, right at the jack. Someone else who is heavily focussed on high input impedances and tight construction might instead be worried about capcitive coupling to sensitive wires. The voltage transient in the leads to/from the LED can be picked up if your wire routing is not very careful.

Splitting the LED resistor and using a cap there to on-board ground slows down the edges of the transient. The magnitude is the same, but the edges are slowed to the point that the transient is no longer perceptible as a click or pop. This does cause a higher current transient inside the decoupling cap/LED resistor/LED/switch/ground loop. This transient is not perceptible outside that loop unless you have wire routing that couples that loop to other circuitry by either shared inductive loop area or capacitive couping from one of the wires. Get your wiring wrong and you can get a pop from the LED current loop.

I did some modelling work on this. The split LED resistor works very well outside the LED power loop, cutting the transient in the input ground lead particularly to something so slow that it can't be heard unless you have a massive gain circuit to amplify it. Splitting the resistor in half offers the biggest reduction, largely because both on and off transients are slowed down equally - there's not one fast edge to come through. For followers of history, that circuit comes from vacuum tube days.

The split LED resistor does cause a larger current transient. It's limited to the LED/power supply/switch loop. It *is* possible to route your wires so that the transient can be picked up if you're unlucky.  Isolating the LED to a separate power supply loop still has a switch transient as well, and you can pick that one up if you route your wires poorly on that one as well.

It's always possible to do a poor job on some things. Wire routing is a good example.

Maybe feed a small current (1M resistor) to a BJT base and small cap (0.01uF) ??  A 0.1 second turn-on should fix things.

Good point. When one turns on an LED, the supply current changes unless you are using current steering to do the switching, and that's something I've never seen in a pedal for purely indication. If the current changes, the only options left to nail down are how fast does it change and which wires does that change run through. It's obvious that if it changes by, say, 1ma per week, the change is too slow to notice as audio. Likewise, if the current changes slowly, it pretty much does not matter how you route your wires, because the transient couples poorly either by capacitive or inductive coupling. Once you know that there is a "too slow to be heard" you just pick how slow is slow enough in your equipment. An NPN with a 1K to 10K base resistor and a 0.1uF cap from collector to base is slow enough to suppress relay coil voltage transient edges pretty well.

If you don't slow down the edges of the current or voltage transient, you are left with isolation to get rid of pops. You can isolate by distance, isolate by shielding, and isolate by electronic components. Wire routing away from transients helps cut coupling capacitance; twisted pair to/from LEDs and switches helps both by capacitive cancellation and by inductive loop area minimization. Separate decoupling sections is component isolation. It amounts to separation by a resistor, decoupling by a cap in as many sections as you like.

[zachary vex]

Surely "discuss" is too dignified a word.

it's the perfect word! 

[brett]

Hi.
I am surprised by the (ill-informed?) suggestion that empirical work = original, modelling work = something else?? (copying??)

Empricism is the means by which we test models (mental, logical, physical, computer, etc).  That's it.  Unplanned empriicism is like giving chimpanzees computers and hoping they will write a best seller.  Modelling, however, relies on planned empiricism to ensure that its predictions are accurate and/or useful.  We all model and plan, some better than others, and in different ways, but we all do it every second of every day, about every conscious decision and every plan we make.

cheers