Delay the pop?

[m_charles]

Ever trying to reinvent the wheel, I had a thought.
Aside from the fact that "it's just a little pop", I am among the sort that still gets a bit bugged in certain situations about the "pop" when hitting the switch to bypass. I know relays, or non true bypass are options, and we all employ the good ol' 1M pull down.
But here is something I wondered.
To totally over-simplify the info provided from various sources, the pop is essentially the sudden change in DC that happens when we "surprise" the circuit by asking it to do its thing without any warning.
I picture waking a friend with a fog horn in order to wake them up.
What if there were a way to delay that voltage change at the input and output and allow it to "slowly" ramp up the effect, thus waking it up a bit more gently to avoid the pop?
As far as I understand, in electronic circuitry, a millisecond is a loooong time, right?
Would there be a way to ease into the on/off change via a small circuit at the input, that ramps things up just slow enough to ease that pop?
Crazy, or something to possibly pursue?
:-)

[Buzz]

Some amplifiers employ a relay to kind of do what you're suggesting.

I haven't built a pedal yet that couldn't be de-popped with a simple pull-down resistor and/or good wiring practise.

I generally believe the easiest solution is the best solution.

(edit... oops just noticed you already mentioned the relay... sorry!)

[R.G.]

Slowing down the transition is a time-honored active switching technique. It's used in every Boss and Ibanez pedal with JFET switching.

As a practical matter, only JFETs are actually used for this, although other components have the capability if you want to design long and hard enough to get it right. It can get complicated for other devices though.

[psychedelicfish]

Rod Elliot put an amp/effects switching scheme on his website recently that used LDRs to switch the signal. They have a relatively slow switching time and low distortion, which makes them good for this sort of thing.

[m_charles]

I do realize that fets can be used for this, that's kind of what I mean.
Would there be way to build this into a circuit so using mechanical switching can be tamed.
Perhaps by taking advantage of the extra lug on a DPDT (leaving out an led for simplicity of discussion).
My thought was that if RG could figure out a way to turn an LED on and off with the extra lug on a DPDT (his millennium bypass), then
I'd bet a way to slow the DC change would be pretty doable.
My understanding of electronics is getting better and better as a self taught guy for a few years now, but unfortunately my knowledge doesn't allow me to make this happen on my own, but it does certainly seem possible?
For some reason the relay switching setup seems kinda clumsy to me. I feel like there has to be an easier way to make a close to silent switch while still using the simplicity of our tried and true, true bypass, 3PDT, DPDT setup.
Just thought I'd throw it out there.
If anyone wants to draw up any theoretical ideas, I'd be happy to breadboard and post results.
Thx

[R.G.]

Let's look at first principles for a moment.  Switching elements of all kinds can be characterized by a few of their features. Some of the features can get esoteric, but for switching audio, the things that matter most are
- on-state resistance (Ron)
- off-state resistance (Roff)
- abruptness of change between the previous two
- distortion
- offset voltage
- control signal feedthrough

Mechanical switches are near-ideal, since Roff is huge, Ron is milliohms, distortion is nil, offset voltages (and thermal noises) are tiny compared to guitar levels, and the "control signal" can't feed through - it's a mechanical movement. However, the abruptness of change for a mechanical switch is both instantaneous - the contacts either touch or not - and it can bounce, making repeated contacts and breaks before the contacts settle. For the purposes of your question, it is what it is; there is no known way to make it vary smoothly from Roff to Ron, as there simply is no middle there. Some multi-contact switches can be "make before break", which means that the next switch state contacts close while the previous contacts are still in contact, then the "breaking" contacts open. This is useful in some audio situations, but difficult to get, and not usually available in a stomp switch. It can also cause oscillation if the things that make and break are not well thought out. So - mechanical switches are not so amenable to slowing down any transitions.

Relays *are* mechanical switches, just activated by an electro-magnet. They have all the issues of mechanical footswitches, and add the possibility of electrical feedthrough of the large control signal unless you are careful about how you apply and remove the control signal.

LDRs offer smooth transition between states and no feed through of the optical control signal, but can be disappointing in terms of how big Roff is and how small Ron is (i.e. the ratio of off to on resistance is too small, and you get signal feedthrough unless you're careful) as well as switching speed, which can be v e r y slooooow for some LDRs, and distortion, which they do have. In the micro-signal realm, you can get light-signal modulation of your audio signal if you use fast ones. Then there's that whole EU making LDRs illegal with RoHS legislation, the expense of good ones, lack of standardization...

Bipolar transistors do signal switching, and were used for this. The offer easy availability, low cost, fast and controllable switching speeds from on to off, a wide range of on to off, but have DC offsets, control feedthrough, and very limited signal voltage handling. Sometimes you can get away with switching guitar level, but line level is tricky, and speaker level is pretty much out of the question.

Then there are FETs. FETs are inherently modulated resistors. JFETs do not have the body diode issues of MOSFETs, and offer both very wide Ron to Roff ratios, both fast and controllable transitions between on and off, fairly easy circuits to slow down and/or eliminate control signal feedthrough, no built-in offsets, low distortion, low (compared to mechanical switches and LDRs) costs. They are the element of choice for smooth audio switching because their switching transition can easily be made to be sub audio in speed.

MOSFETs are like JFETs, but have the problem of the body diode limiting signal levels in the reverse direction. This is solved by their use inside ICs as "CMOS analog switches" by the rearrangement of the internal connections, but most switching ICs have very fast internal logic circuits providing the control signal to the actual switching elements, so they tend to feed a tiny bit of the control signal through to the audio path. This can be managed, as the write-ups on Geofex.com about CMOS switching outline. But control feedthrough can cause tiny ticks if you're not careful in applying them.

As for the actual switch pop itself - eliminate this by making the switch element not switch between two DC levels. Make the two signals the switch chooses between have identical DC levels. When  you do that, there is no DC signal pop. There is only control signal feedthrough, which can be inaudible. I do this option professionally all the time; it's not rocket science, just some attention to circuit design.

In regards to your question:

I'd bet a way to slow the DC change would be pretty doable.

It's even better - and possible - to make the DC change be zero.

And that brings up the Achilles heel of mechanical switches and relays - their inescapable on/off binary nature. They switch instantly, as soon as the contacts touch, and if the signal happens to be non-zero at that instant, the sudden make/break of the contacts makes a transient out of the signal itself, of a size equal to the instantaneous signal voltage. This "keying problem" is - well, was - well known to electric organ makers back in the day. There isn't any way around this except to switch when the signal is zero for the instant it flies through zero or to mute the signal before switching. Mechanical switches have poor control of the exact instant they make or break, so the first option isn't available, only making the "zero instant" very long by muting.

And that leads us into series versus shunt switching, which is as long itself as this is already.

[pappasmurfsharem]

Jack Orman has some stuff about using a transitor and a few other components.

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

EDIT:
Ah wait that's just LED pop. Well leaving it here anway.

[m_charles]

In regards to your question:

I'd bet a way to slow the DC change would be pretty doable.

It's even better - and possible - to make the DC change be zero.

RG. I'm sure you can see this question coming from a mile away, but.
An example of this?
Thx for your thought out insightful answer. It is always appreciated.

[Rob22315]

There's an active anti-pop circuit in the Fulltone OCD v4 and the Joyo clone (Ultimate Drive).  A schematic can be found here:

http://www.freestompboxes.org/viewtopic.php?t=14757&p=167796

See the post from guitarmongot about 1/2 down.

[R.G.]

The biggest thing is to make sure the switching element has no DC change. Here is how I did this with the CD4053:
http://www.geofex.com/article_folders/cd4053/cd4053.htm
This looks a lot more complicated than it really is because the 4053 has nine switching pins. The idea here is that the signals coming in and going out of the switching pins are all dragged to the same Vref, and then the DC from the signals being switched are held at whatever DC level they want by their internal circuits. Once the caps have all charged to their final values, the DC differences are kept entirely across the caps, and the pins of the switching elements do not change DC levels.

This is **the same** technique you'd use for a mechanical switch, the only difference being that "vref" can be equal to ground. That makes the resistors to Vref be resistors to ground, and this then is the pull-down resistor solution. The resistors force the caps not to ever be able to leak a bit and generate pops.

For CMOS, there is a secondary popping mechanism where the fast logic level signal on the control pins that tells the switch circuit to actually switch can couple in a small amount of charge to the signal path. This is minimized by biasing the CMOS switches near the middle of their DC power supply, where the coupled-in control signal tends to cancel from the P-type and N-type switch elements and so the coupled control signal pop tends to be lost in the noise it's so small.

So you force the DC change to be zero with capacitors to break the DC path, and resistors to hold the capacitor leads at a constant DC level on both sides of the cap.

I haven't looked at the active anti-pop stuff yet, but the bog-standard way to do that is to mute the switching pop and signal for a few milliseconds with a shunt switch to ground. Shunt switching in general tends to be quieter, but you still have the issue of making sure the shunt switch doesn't insert pop while it's keeping another pop from being heard. Still, it can be a useful technique if you're willing to pay the freight.

Note that
(1) absolutely quiet switching may be impossible
(2) it's good enough to lower the total amount of popping signal below the inescapable circuit noise
(3) if the signal being switched is big enough, whatever pop remains may be so small that it is unnoticeable, which is as good for all practical purposes as not being there
(4) by being able to switch the signal very slowly, the energy in the pop may be moved to frequencies so low that they are not passed by the audio chain; "pops" with frequency components under about 20Hz are in general not a problem to humans.

[Howard]

Hi. I reach out to you for some advice. I have designed a passive a/b box using the 4DPT switch. This allows me so isolate each output and eliminate ground-loop hum. Also I short the output not in use in order to keep that amp silent. Works as intended, but adds a new problem. As most tube-amps have a small (0,X) DC voltage on/from the input, when releasing the short/switching to that channel, the DC raises and makes a sharp peak that gets amplified, thus causing a loud "thump" in the amp/speaker at switchpoint. Now, if you look at the DC being shorted/released i guess this could be evened by (instead) switching a capacitor between ground/live in order to maintain the DC-level, whilst still keeping that amp almost silent? Howard, Norway

[R.G.]

The first thing I'd try is making the path to ground for the "shorted to ground" inputs be a resistor, not a short. There may well be a value that keeps things quiet.

The capacitor to hold the tube amp's DC input offset is a good idea, but it gets tricky to figure out how to make that work for all switching conditions.  This is complicated by the tube amps' high input impedance.

[Howard]

Thanks! I guess it will be a compromice between having a small R enough to redusing hum whilst not pulling down the DC? I found a 400V 22nf cap that I replaced in my F Junior II (Bills mod), and connected it in paralell to the input of my various tube-amps (Champ 600, Excelsior & Jr ). At mid range volum-settings the hum from an open cable was redused to a hiss that is of no real annoyance. Also the pop was evened to an acceptable level.

[amptramp]

Go to page 2, Figure 1 here:

http://www.onsemi.com/pub_link/Collateral/MC14007UB-D.PDF

and break the connections to pins 3 and 10 and add an RC network (R in series, C to ground) between the input and pin 3 and pins 8/13 and pin 10.  Set the RC time constant to be greater than 0.05 second at the inputs to the 4007.  A 1 meg resistor and a 0.05 µF cap will work fine and if you have to go to 0.047 µF, that will probably work too.

Go to www.geofex.com in Tech Tips and look at the latching relay bypass circuit.  The first three inverters on the left form a bistable flip-flop.  Use it to drive the MC14007 / CD4007 circuit above.

If you want to get fancy and do make-before-break, you can break the R into two series elements and bypass one resistor with a diode to ensure faster switching in the "ON" direction than the "OFF" direction.  This will ensure there is no time when you get nothing connected to the output, but it is unlikely you would need to do this - the circuit above will do just fine.