quiet momentary loop insert

[blackieNYC]

I don't know where I got this, but it does work [edit-see below post].  It's a momentary loop insert switch for the purpose of inserting effects briefly, or rhythmically, as quietly as possible. It works well. The FETs are straight out of Boss type switching, but does anyone see anything inefficient here?  I'm going to commit it to a very small perf board and changes will then be messy.

[blackieNYC]

eh, it's not as great as I would like.  If I am using  it to momentarily stutter/burst on, there is a bit of a snap in the audio, in both the on and off switching. It's not a zero-crossing edit type of sound - it can be heard when there is no sound coming from the guitar.  changing any of the caps can take some edge of the brief noise, but the response starts to get slow.  A Boss pedal is indeed quieter than this.  What is my circuit in need of?

reducing the 2 meg resistors to 100k (Boss) helps a lot. still not as perfect as an actual boss pedal.

[R.G.]

First, I think there are some issues, as that circuit doesn't exactly work with the same time constants as the Boss/Ibanez stuff. I'm going to take it that you have read The Tech of Boss at geofex.com. If you haven't, it might be worth your looking.

All switching that does an instant switch, including most especially metal contacts, causes a transient if it happens to catch the audio outside of a zero crossing. The cure for this is to ramp the audio up and down over a few tens of milliseconds. That's what those 1M resistors and caps to ground at the gates do. The trick is getting the resistor-cap time constant right. For a quick fix, you might try increasing the caps right at the 1M/gate diode quite a bit, until the sound fades in/out too slowly. There is also the issue of imbalance and timing. The two-FET side has only a 0.002 slow-down cap, and that might be all of the issue. I'd try making that one 47nF too.

Then there's the drive circuit. The input 47nF isn't doing all that much for you, depending on what impedance is driving that 0-9V input signal. If it's a low impedance, the 47nF may be charging and discharging almost instantly. If it's a high impedance, the actual state change may be oddly staggered between the two-FET and one-FET sections.

If you think about it right, the Boss flipflop is two transistor amplifiers in series, just set up with feedback so each drives the other. You can "unwind" that circuit into a no-feedback sequence of two amplifiers, and drive the base of the first one from your input signal. This makes the outputs from the two collectors be very nearly symmetrical switching, as well as driving each 1M/47nF/diode gate circuit with the same impedance. Sure it costs another transistor, but the one-transistor circuit isn't doing what you want. There are other approaches that are plausibly simpler, too.

[blackieNYC]

Thanks RG, I have been referring back to that Boss and Ibanez article.
I'll see if I can figure out what you're saying about altering the Boss style config.
That 47nf on the input can also slow down the switching.  I've played with all the caps - lower values are fast but the quick little switching sound has more high frequencies to it, higher values result in a duller thump and also become too slow for any kind  of stutter purpose. 
Plausibly simpler? anything plausible is... plausible.

[Mark Hammer]

All of the above said, the circuit - once it settles down to a clickless form - holds promise.  F'rinstance, I'm one of those folks who think that most ring-modulator or bit crusher pedals should have a momentary engage switch, rather than a latching stompswitch.

[blackieNYC]

Yes, Mark, that stuck in my mind. A rhythmic insertion of an effect, even a brief solo moment.

If I wasn't clear, I also upped that .002 cap to .047, and it took some high freq bite out of the switching noise. But it could be much better. And I hear this click with no audio from the guitar, that eliminates zero crossing. The thump can vary in amplitude a little, depending on the charge of the cap I suppose.  I still need to try the exact Boss circuit and make it non-latching. 

[blackieNYC]

Should any NPN transistors work in the boss switching control circuit? I'm using two BC182L s. There's no switching - 6.5 volts on one collector, .1 on the other, but sending a ground blip to the bases doesn't change the two.

[Redvers]

Is all this really that much better than a dpdt momentary switch set up so the fx input is sent to ground when bypassed?

[anotherjim]

Although you have those high value resistors to tie the FET channels at some DC reference, we don't know if the other side of the DC blocking caps is referenced, or floating. I'm not sure you can leave it to chance that whatever is plugged in will provide a DC reference.

[Mark Hammer]

Is all this really that much better than a dpdt momentary switch set up so the fx input is sent to ground when bypassed?

If you can find me a soft touch momentary DPDT, I'd agree.  The challenge is that the only momentary DPDTs that seem to be commercially available have a discernible hard click that is identical to latching DPDTs and 3PDTs.  That click is not heard in the audio path, but you have to pay attention to whether you've pressed hard enough to produce it.  If I have the choice, I'd rather an electronic switching system and soft-touch SPST.

That's one of the reasons why I like the old Craig Anderton CMOS switching circuit from Electronic Projects for Musicians.  It uses a 4016 chips, and while it is presented in the book as using a latching SPST switch, it also lets one use a paralelled non-latching switch (as long as the latching one is open) for momentary bypass/engage.

[blackieNYC]

Another Jim - the caps you see on either side of the FETs are the inputs and outputs to separately housed pedals, which will also have caps so no, there will be no DC on the other side of those caps.  Maybe I need 1meg resistors to ground, as we do on the inputs to most pedal circuits.  U think that's the noise source right there?

[anotherjim]

I don't believe you should have wire between 2 caps that is floating. It's a noise trap. Also, if any coupling cap is electrolytic, then there is no polarising DC across it.
You could argue that they will be referenced to ground provided whatever is plugged in has anti-pop resistors or an output volume pot that's doing the same job, but it can't be assumed every possible thing will have them. 

[blackieNYC]

I will look closer at that. Meanwhile, I've tried the complete Boss circuit but my transistors are not switching.

[PRR]

> dpdt momentary switch

As MH hints, it is hard to ensure BOTH poles of a DP switch engage at the same time, unless you build a SNAP.

The work-around would be a SP switch working a DP relay.

If the acoustic clack is offensive, play louder! Or use flex-wire and wrap the relay in foam. (A long-run relay could overheat but a momentary foot-action relay won't.) Or some FET or Photo-R technique to make the one pole work two audio switches.

[blackieNYC]

I need to clarify and I think I have this right. It's been said in this forum and Geofex, that FET switching of audio allows you to get around the fact that any hard and fast mechanical switching will not happen at a waveform zero-crossing and will therefore result in an electronic noise. The FETs can be turned on slowly, diffusing or diluting this "pop". It's a great technique if you want silent switching, and in a boss pedal - it works. I have put a low frequency sine wave (most revealing method) thru a boss pedal and the switching really has no discernible noise at all. Even CMOS switching can't give you that. LDRs can get there but there is a lot of residual resistance in On mode.   
My goal is to achieve a momentary looper that can insert an effect in the manner of a very clean stutter/kill switch.
(Large copper thingy optional.  )

[R.G.]

I read through all this again, and in the spirit of "just sayin'," I have to repeat - I have put literally tens of thousands of pedals into the commercial market using a CD4053 for all the bypass switching. I have driven this chip from hard-contact switches and pullup resistors, CMOS logic, and microprocessors to make it switch. I can find the switch "click" on a scope, mostly by using the control logic signal as a trigger, and I can turn amps full up and hear it as a faint little ((... pip ...)) but I can't hear it with a guitar's excess noise going it.

Again - just sayin'.

[blackieNYC]

RG - I thought you were a fan of the jfet for silent switching. I stand corrected. Anyway, I've got a boss in front of me and it works I just have to get there with a momentary circuit. 

[R.G.]

Not much of a correction. It's a matter of picking a suitable tool.

There are no perfect switches, only switches better or worse suited for a task. JFETs have some things that CMOS switches can't do, just as CMOS does some things well that JFETs can't at the same level of external complexity.

There are also issues with considerations other than just switching that wind up creeping in and changing things. These are often cost, size (i.e. number of parts or pins) or general dislike for ICs.

If I were doing your task in the absence of other issues, I'd use a CD4053 plus some logic to make the thing be either momentary or latching depending on some other setting. But I'm not doing it, and this thread started with you having already adopted a Boss style switch. OK, that works if you deal with the number of parts and the unit-to-unit variations in JFETs.

Here's a blow by blow on how to do it. In the article "The Technology of JFET Switching in Boss and Ibanez Pedals" (http://www.geofex.com/Article_Folders/bosstech.pdf)
at the top of the third page, I have some illustrations of the flipflop in these pedals. In the leftmost illustration, imagine taking C1 and R2 loose from the collector of Q1, and connecting them instead to a 10K resistor to +9 and an SPST switch to ground. If the switch to ground is open, current flows into R1 and Q2's base, turning it on, and making its collector go low. That then pulls R3 low, and Q1's base low, turning it off. So with the switch open, the collector of Q2 is low, collector of Q1 is high.

Now close the switch to ground. R2 is held to ground, turning Q2 off, so its collector goes high. This high collector turns Q1 on through R3, and its collector goes low.

Notice that the JFETs driven by this setup cannot tell any difference in how they're driven from the true, Boss flipflop setup. They only see driving collector low or high.

I personally would ditch the transistors and put in a CMOS hex inverter flipflop. One chip, three resistors and one cap, and a floating SPST switch. In this version, the 1M driving resistor to each JFET gate cap/diode is simply tied to the flipflop output pins on the CMOS chip. I like the hex inverter version because it's more reliably triggered by switches than something like the CD4013. YMMV.

The ugly part of the JFET switch is that you have to get JFETs that have a low enough Vgsoff to switch with a 4V or so bias voltage and a pull down to 0V to turn them off. Even then, there are timing issues, as each JFET will have a smaller or larger Vgsoff, and so turn off sooner or later in the voltage ramp fed to its gate. Matched JFETs are not mandatory, but useful if you want smooth transitions. Where the JFET gate turns on or off in the ramp fed to its gate is also an "abruptness" issue, as the gate ramps move fastest at the start of their change, and more slowly as they approach end state, as the "ramp" is actually an exponential curve.

But JFETs are easier to understand at a high level, and they were very cheap for the Japanese companies using them. It's another mixed legacy, just like 9Vdc, center negative DC jacks, and single ended guitar signals with high impedance cables.  Of course, if the job were easier, they'd have hired lesser people than us to cope with it.