A/B/Y/S Switcher and isolation for stereo

[cherler]

back to the transformer though, I found this http://www.amazon.com/PYLE-PRO-PHE300-2 ... 00BARTW3I/

Damn, that's tough to beat. I guess I have to decide if I'm stubborn or practical now haha

I should have ported it to 8 pin attiny45.

Yeah I'm thinking I should do something similar in the future, maybe even just for soft switching which I tend to prefer. Relays for daaaaaaaaaaaaaays

[cherler]

Alright so update!

I tried out a mono version of the pedal just to make sure the buffers, switching, and relays were all working like I wanted them too. After sorting out a few dyslexic mistakes with resistor choices on the buffer, the audio sounds fine but the clicks are SO LOUD. So I decided to revisit the transistor situation throwing the relays again. Turns out I was probably just doing something exceedingly stupid before, and switching off the five volts with one transistor is as easy as I thought it should have been originally. I can't even think of what I must have done wrong.

So now I'm moving everything that doesn't touch the signal path off the 9v and onto the 5v rails and I can more heavily filter the 9v rails without fucking up my ridiculous transistor pair switching nonsense. I do think it would still be a good idea to mute the ins/outs before throwing the relays to keep the clicks out of the audio path, but I realized that maybe just suddenly dropping the audio lines to ground could be contributing to the clicks. Anyone got any thoughts on that? I'm going to give it a go tonight hopefully and find out, I just have to finish rewiring all the relays with a simpler/better transistor configuration haha.

[eatyourguitar]

after you put the relays on a dedicated power rail, the remaining clicks are from changing DC bias on the input and output caps. this is why you need 3M or 1M from tip to sleeve on all inputs and outputs. I start at 3M and go down to 1M as needed.

[multi_s]

I do think it would still be a good idea to mute the ins/outs before throwing the relays to keep the clicks out of the audio path, but I realized that maybe just suddenly dropping the audio lines to ground could be contributing to the clicks. Anyone got any thoughts on that?

you would need to mute them with a fade out, not just a sudden chop to ground, otherwise you are right, you are likely to get pop artifacts if you are playing. an alternative is maybe to measure a through zero crossing and try to switch/mute close to that instant to minimize popping.

most relays have a switch time of around 3ms, you can usually find something about this on the data sheet. if you have break before make (BBM) style that means you have 3ms where nothing is connected, but if you use make before break (MBB) you have 3ms where both throws are connected. neither are ideal, the latter is likely to create a less noticeable artifact if audio is passing while you switch, but for whatever reason latching relays tend to be more available in BBM style.

i think most people just live with the possibility of artifacts with relay switching because it keeps the design simple.

[eatyourguitar]

I think the common solution is a combination of relays and solid state devices like analog mux demux switches, FET shunt to ground, or another relay that has the 1M in it. radial has an amp head switch that has time compensated relays that are a lot closer to 0ms switch time and a lot closer to being neither BBM or MBB. radial says there is no dummy load or anti pop resistor. they simply switch two heads to two cabs very fast. this is no joke and I don't entirely trust it with MY tube amp. but there is food for thought on how you can engineer something to solve a problem even when the parts seem like its not going to work.

[eatyourguitar]

you would need to mute them with a fade out, not just a sudden chop to ground, otherwise you are right, you are likely to get pop artifacts if you are playing.

why not just introduce 10M shunts on all switched signals before switching the relays? this should totally eliminate the pop. then after switching you remove the 10M shunts. you can use 4066 cheap. attiny45 also cheap.

[cherler]

you would need to mute them with a fade out, not just a sudden chop to ground, otherwise you are right, you are likely to get pop artifacts if you are playing. an alternative is maybe to measure a through zero crossing and try to switch/mute close to that instant to minimize popping.

most relays have a switch time of around 3ms, you can usually find something about this on the data sheet. if you have break before make (BBM) style that means you have 3ms where nothing is connected, but if you use make before break (MBB) you have 3ms where both throws are connected. neither are ideal, the latter is likely to create a less noticeable artifact if audio is passing while you switch, but for whatever reason latching relays tend to be more available in BBM style.

i think most people just live with the possibility of artifacts with relay switching because it keeps the design simple.

Yeah my relays are latching form c so I'm guessing they're BBM. I can't find that explicitly stated on the datasheet, but it seems likely. They've also got a switching time of about 2ms according to the data sheet. If I go for anything as complicated as watching for a zero crossing, I'll probably end up with a prototype with the possible artifacts first regardless, so we'll see how it goes!

Although I'm pretty sure with the board I'm using, watching for zero crossing wouldn't actually be the worst thing in the world. It has pretty good A/D converters and I could probably sample pretty damn quick, maybe even get up to 44.1k. If I have a decently wide range of acceptably low voltages, it could probably pull it off. I'm gonna save that idea, thanks man!

I think the common solution is a combination of relays and solid state devices like analog mux demux switches, FET shunt to ground, or another relay that has the 1M in it.

So I could use the optical mosfet pairs I have for this. Connect the center pin of the relay to ground through a big resistor to prevent any DC bias from accumulating during the ~2ms time where there's no connection. This seems to make a lot sense. Although if the resistor is big enough, why not just leave it connected always? At 10M it can't be affecting the impedance of the pedals that much could it?

I'll be in LA for a few days so I won't get to try any of this until next week, but I'll make sure to post back when I do. Thanks guys!

[multi_s]

you would need to mute them with a fade out, not just a sudden chop to ground, otherwise you are right, you are likely to get pop artifacts if you are playing.

why not just introduce 10M shunts on all switched signals before switching the relays? this should totally eliminate the pop. then after switching you remove the 10M shunts. you can use 4066 cheap. attiny45 also cheap.

mmm shunt to what? ground? can you draw a schematic?

[cherler]

Ok so I'm back finally and I've managed to get the popping a hell of a lot quieter. I just threw some big resistors to ground on the inputs and outputs of the buffers and on the pole of the relays to keep any kind of DC bias from building up. It made a HUGE difference. I haven't messed around with switching them in and out, but I'm in good enough for a first prototype kinda range now. Thanks for all the help!

The way the clicks manifest now is kind of interesting though. In the standard A/B/Y parallel mode, they're almost completely inaudible. In the order switching series mode however there's definitely some quiet clicks still happening. I honestly think it may have something to do with the timing of how I'm switching the relays. Each tap of the footswitch means two relays have to throw, and if I make sure there's a slight delay between them I think that could get rid of these clicks but I haven't tried it yet. Do you guys have any thoughts? I want to try and get rid of these clicks eventually, but as of now I think I'm ready to build a quick mono prototype and see how happy I really am with everything. At this point I think there's a good chance I can like with the clicks.

[eatyourguitar]

no time for schematic sorry. I am running a guitar pedal company and two synthesizer companies.

[cherler]

I assumed you meant shunts to ground right?

[eatyourguitar]

yeah

[multi_s]

no time for schematic sorry. I am running a guitar pedal company and two synthesizer companies.

which ones?

[eatyourguitar]

no time for schematic sorry. I am running a guitar pedal company and two synthesizer companies.

which ones?

ETIT, ETIT, and non-disclosure agreement. I will have more info when her module is released. in the past, I did work for RYO, steffcorp, frequency central, black sheep effects and a luthier from philly that uses his real name I think?