LED indicator causing clicks

[Mark Hammer]

Just to add some realism and context here, the audibility of the clicking is a function of the things one does to minimize it, but it is also a function of the amount of current sucking the LED requires in the first place.  If I've understood it correctly, siphoning 0.2ma off to the LED probably won't make a big difference in what comes out of the audio path if the devices in that audio path are properly decoupled.  Start shifting over 2-3ma to an LED in rapid bursts, and you WILL start to hear it.

This is all the more reason to use ultra-bright LEDs in high-contrast, high-visibility circumstances (e.g., against a dark surround) where the current requirements can be reduced significantly enough that whatever power-line disruption might be created bybringing the LED in and out of circuit do not result in audible or annying clicks.

While we're talking about it, though, just what exactly is the audibility of LED-derived clicking in solid-state switching systems (e.g., Boss, DOD), and if it is significantly less than the 3PDT scenario being discussed here, how exactly is that produced?  I ask because while I own a number of such pedals, I never play at the sort of volumes where one would hear the clicking over the ambient noise.

[R.G.]

While we're talking about it, though, just what exactly is the audibility of LED-derived clicking in solid-state switching systems (e.g., Boss, DOD),

The SS switching systems have a similar problem, as you know. They inject charge into the signal path. The cure that has been adopted by Boss and Ibanez, and probably DOD as well is to use edge slowing.

What they do is to put enough slow-down on the gate signal to the switch elements to make the switch take units of milliseconds to turn on or off. That lets the same amount of charge injection happen, but it happens over a long enough period that the audio path doesn't have audible results.

According to the National Semiconductor Audio Handbook, solid state switching spikes are sometimes not audible in themselves because the spikes that get injected are so short - micro- or nano-seconds - that they overload things like opamp input stages and what we hear is the recovery time. JFETs and MOSFET analog switches both fall into this category a lot. Slowing the gating signal in this case reduces charge injection both by making the injected charge lower because the gate signal cannot be well coupled by the few pF of input-output capacitance on the FETs and by slowing down the signal path resistance change.

On this one, the answer to your question is - however little you want it to be. If you get switch clicking from these kinds of circuits, you can slow down the signal to the switching elements more and the coupled transient goes down.

Transient edge problems is also an issue in CMOS logic setups. The fast edges of logic signals are coupled capacitively to other conductors and can add/subtract to the signals already there and cause timing and glitch issues. Slowing down edges is not a possibility that can be used very often there, because fast edges are what makes your logic fast. The amelioration in such setups is to slow down the edges as much as possible within the timing constraints with series termination by resistors, and by making all of the PCB traces constant-impedance transmission lines so that line-ringing does not happen. End-termination was used at one time, but it's too costly in terms of both money and PCB real estate to use in modern high speed computer boards. A bluebird that helped the computer industry was that CMOS logic needs input protection. The input protection diodes happened to form snubbers that clamp input transients to V+ and ground, eating transient energy out of the lines.

As to learning from others, I remember the advice a young man posted on the web a long time ago. I think I still have it in my archives. The advice was to read everything, to steal circuits from wherever you could, steal everything possible, in the pursuit of your own learning.

We all stand on the shoulders of giants, yes?

[LyleCaldwell]

Earnestly taking your advice, I am currently using a 2.2K resistor for my LEDs.  Would it be better to use two 1.1K resistors with the 10uF cap between to ground, or a 2K resistor, then the 10uF cap to ground, and then a 200 ohm resistor?

[R.G.]

Earnestly taking your advice, I am currently using a 2.2K resistor for my LEDs.  Would it be better to use two 1.1K resistors with the 10uF cap between to ground, or a 2K resistor, then the 10uF cap to ground, and then a 200 ohm resistor?

The more stages of slow-down, the better of course. However, generally just a split resistor and one cap is good enough. If you split it multiple times, equalize the resistors. So if you used two slow down caps, make your 2.2K into three 750 ohm resistors. But one split is enough.

Cut the LED current all you can. Would a 4.7K make it bright enough?

[puretube]

how about a PWM-ed indicator-LED (low consumption),
switched by help of a through-zero-detector?

(runs towards the P.O.... )

[zachary vex]

Earnestly taking your advice, I am currently using a 2.2K resistor for my LEDs.  Would it be better to use two 1.1K resistors with the 10uF cap between to ground, or a 2K resistor, then the 10uF cap to ground, and then a 200 ohm resistor?

there are many disadvantages to using high-current for LEDs.  my suggestion remains to use at least a 15k to 22k resistor and an ultra-bright LED such as the ones in this series from Mouser:  351-3232.  likely, that will solve any problem you have without modifying the circuit any further.

[Mark Hammer]

That's what I was figuring too.

Isn't it nice when something you do to prolong battery life also ends up improving the sonic acceptability of something too?

[LyleCaldwell]

Zach,

While I use the ultrabrights a lot, in this particular effect I have to have blue, yellow, and red LEDs.  The ultrabrights make getting matching (or near enough) angles very difficult.  The best matching I have been able to do is with the Kingbrights here: http://www.mouser.com/catalog/625/53.pdf and this blue one: http://www.mouser.com/index.cfm?&handler=data.listcategory&D=604-WP7113MBD&terms=604-WP7113MBD&Ntt=*604WP7113MBD*&N=0&crc=true. 

And if I use a 4.7K the yellow gets too dim.  2.2K is a pretty good match for all of them.  I have plenty of current available (non battery circuit), but would like to further reduce the clicking.

If you have a better way to get matching angles and intensities with red, yellow, and blue LEDs, please let me know.

This is the part I hate - no calculator ever tells me what looks good together - only what is safe for the LED.

[RDV]

Coming from a blind hog: I kind of like a little pop cause then I know I found the button and something's gonna happen.

IKIAI

RDV

[hairyandy]

IKIAI?  I've never seen that one before!  If that means "I Know I'm An Idiot" then I'm gonna have to start using that A LOT...

[The Tone God]

how about a PWM-ed indicator-LED (low consumption),
switched by help of a through-zero-detector?

(runs towards the P.O.... )

Hmm...yet another patent that can be invalidated with prior art.

Andrew

[R.G.]

Yes, we could use up more circuitry that way!

A simpler way to do a no-current change LED on/off is to hook up two transistors in a diffamp, and limit current with something in the  common emitter.

The LED is in the collector lead of one side and either a resistor or just a wire in the collector of the other side. One base is hooked to a reference voltage, the other to a switch signal (which could also be a real switch through a resistor network) and the bases are tied together with a slow-down cap.

When you flip the switch(ing signal) the current from the diffamp pair is rerouted to/from the LED. A constant current flows outside the diffamp signal so you can't possibly have power line or ground line clicks generated. The slowdown cap across the bases would slow the transition from transistor to transistor so the leads to the LED couldn't even radiate. This would make both power/ground line and poor wire routing clicks go away. We'd be left with signal step clicks, which might be neglegible in some circumstances.

[brett]

Hi.
Isn't the main problem with the "split resistor" idea that inrush current of the the "damping" cap is very large ?
? What about putting a third, equal-sized resistor in series (upstream) with the cap ?
cheers

[R.G.]

Isn't the main problem with the "split resistor" idea that inrush current of the the "damping" cap is very large ?

It is. Well, "very large" is a matter of how you split the resistor. The smallest inrush is when the resistor is split in half. In that case, the inrush is twice what it would otherwise be, because the cap is charged up to the full 9V of the supply when the switch makes. However, the current drawn from the main power supply at the effect, isolated by the first half of the resistance, is initially zero, and only ramps up with the time constant of half the series resistor and the cap. When the switch is opened, the cap is sitting at the LED's 2V or so, and when the switch opens, the current in the LED stops instantly, (or as instantly as the inductance of the leads will let it) and the current from the power supply ramps down with the same first resistor/cap time constant. The first resistor effectively controls the ramp up/down time that the effect power supply sees, the second half determines how big the transient inside the LED/switch loop is.

However, note that the split resistor setup localizes the inrush current transient to only the loop containing the LED and switch. This transient can only cause problems if the wires that carry it are close to other sensitive wires. Slowing down the transient to the rest of the circuit prevents the circuit from amplifying the change in ground/power lines if it's slow enough. National Semi recommended units of milliseconds, as I mentioned.

What about putting a third, equal-sized resistor in series (upstream) with the cap ?

Sure! Or three, or four, or as many as you like. However, at some point, the cost of those addtional steps is as much as you'd spend on something like three bipolar transistors and some resistors in the case of the constant current driver. How far to go depends on how much effort you want to spend. A split resistor plus one cap, and some care in routing wires seems like an OK compromise.

I suppose I should go write all this up for folks who aren't used to thinking in terms of what happens in one circuit loop as opposed to another. It's all pretty simple stuff, really.

[Transmogrifox]

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

did you accidentally reply to the wrong thread? 

[amz-fx]

Thinking Outside the Box
by Ed Bernacki

Thinking outside the box requires different attributes that include:

* Willingness to take new perspectives to day-to-day work.
* Openness to do different things and to do things differently.
* Focusing on the value of finding new ideas and acting on them.
* Striving to create value in new ways.
* Listening to others.
* Supporting and respecting others when they come up with new ideas.

Out-of-the box thinking requires an openness to new ways of seeing the world
and a willingness to explore. Out-of-the box thinkers know that new ideas need
nurturing and support. They also know that having an idea is good but acting on
it is more important. Results are what count.
-----------------------------------------------------


-Jack

[analogmike]

outside the box.... that's the ticket.

I have tried all the split resistor etc methods above years ago, and they almost always don't work. Same with the problem of using the input jack ring for battery ground - if you plug into the power jack that connection is not used, and the pedal still pops, so that idea won't help. I have split grounds, even used different power supplies for LEDs and it's never easy. Keep on plugging and you will find something that works with ONE PEDAL, but it won't work for your next circuit though

Here is something helpful...

We started getting POPPING problems on a pedal that we have sold hundreds of with no problems.
Bad switches? I was able to mearure a voltage on the switch, bleeding though from the LED switch contact.
I studied and found the only change was solder... I had ordered
water soluble flux type solder by mistake. Cleaning off the flux stopped the popping.
Threw the solder in the garbage. Don't buy that solder!

Good luck!!!

[gez]

I bought a cardboard box home with me from the supermarket this morning.  I've tried thinking both inside and outside it, but have yet to have a single original thought.  Where am I going wrong??

Confused of Wimbledon...

PS   Nice thread (good info)

[B Tremblay]

did you accidentally reply to the wrong thread? 

No, he was replying to a comment that has since found its way into the memory hole...

[zachary vex]

he was replying to my edited post.  i removed my comment about empirical work... but i'll say this about that... Analog Mike's observations come from building thousands of pedals and having years of experience with production.  my obervations come from building tens of thousands of pedals and having over a decade of experience with production.  trust the builders.   you will always get the solid dope from the production-level builders.  remember, my suggestion with respect to LEDs is... to reduce problems overall, use ultra-bright LEDs and hi (>10k) limiting resistance values. 

and rosin is, as Mike said, a serious problem sometimes.  recently i started seeing a 30% failure in my iMPAMPs after i improved the metal in the output transformers.  turned out, the new magnetic material caused an arc to jump through the rosin between the plate and grid pins on the tubes during power-up!  this in turn burned the board and formed carbon tracks that either created noise, robbed power, or killed that whole channel.  we had to use a dremel tool to etch away the carbon between the plate and grid traces to restore operation, and then protect them from future arcing using Q-Dope (a mixture of MEK and polystyrene.)

by removing the rosin, we stopped all cases of this happening.  the rosin was the weak link, allowing the arc to form in the first place.