Buffered Electrical Bypass Module

[MattG]

Build Rating

5.00 star(s)

It works! See here for the background: Plug-n-Play Buffered Electrical Bypass Using CD4053 and Momentary SPST.

In short, it's a dedicated circuit and PCB that does two things: (1) always-on buffer, and (2) electrical bypass via CD4053. This is essentially how Boss pedals do buffering and bypass, but with integrated circuits (instead of discrete).

The actual effect is yet another revision on @Chuck D. Bones's great modified Animals Diamond Peak circuit. This is the third revision, which was mostly purpose-built for testing this bypass module. (First board build report, Second board revision build report) In this latest (third) revision, besides dropping the integrated relay bypass circuitry, I also opted for using a BJT instead of a JFET for Q1. This was done mostly out of curiosity.

Eagle-eyed viewers will note my nasty solder job on transistors Q2 and Q3. These are the "money" transistors you can roll to tweak the tone. I initially used the same transistors I used in a previous build, just to reduce the number of variables. Once I verified it was working, I swapped transistors because I wanted to try something different. That turned into a huge debacle - I nearly destroyed the solder pads removing the original transistors. Normally I'm pretty good at cleanly replacing components, so I don't know what happened here. It took a couple touch-ups with the soldering iron to actually make the circuit work with the new transistors.

I actually stole the enclosure from the most recent build (second board revision).

But the excitement here is the bypass board. I had help from the folks at DIYstompboxes on the circuit.

Repeating part of my previous post, my goals here are:

• Transparent buffer
• No mechanical parts to wear out (this doesn't include the momentary SPST, which is only two wires and easily replaced)
• Ultra-low power consumption
• Low cost
• Use modern, readily-available commodity parts.

I haven't measured power consumption yet, but unless I made a mistake, it should be very battery-friendly.

The only real downside here is the physical size, but the next step is to have SMD versions prebuilt; the SMD version design is almost done, and it's dramatically smaller (currently 22x45mm). I wanted the prototype to be through-hole in case I had to debug. The other obvious downside to a design like this (as with Boss pedals), even bypass requires power, so the circuit is completely worthless without power. But the same goes for my amp.

I have plenty of extra bypass boards, happy to give them away for free, just PM me. I also have a few extra Diamond Peak boards available (both this and previous versions), also free, just drop me a PM.

 

[MattG]

I had an miniaturized surface mount version of this design ready to go as soon as the through-hole prototype was validated. I submitted this design to JLCPCB not just for PCB fabrication, but also assembly. They were delivered on Monday!

I had a The Fingers PCB all soldered up, just waiting to be put into an enclosure.

So I wired up my SMD version of the buffered electrical bypass module, put it in an enclosure, and wired it up.

When it comes to the first live test of a new build, I always do a effect-bypass cycle, just to see the LED turn off and on. And my heart sank when I did that with this build: the LED didn't turn on. With cynicism, I went ahead to see if it even passed signal - and it did. And in fact, the core functionality - toggling between effect and bypass - worked just fine. The only problem was that the LED didn't light up.

I poked around with my DMM, and quickly realized my mistake. Changing the design from through-hole to SMD obviously requires changing all the footprints accordingly. And when I changed the 2n7000 that acts as the LED switch, I forgot to take the different pinout of the TO-92 versus SOT-23 into consideration. The SMD SOT-23 version of the 2n7000 is actually a 2n7002, and had I changed the part in my schematic, then the software (KiCad) would have automatically caught the mistake.

Tonight I managed to remove the factory-soldered 2n7002 and replace it with a TO-92 2n7000. And now the LED works!

I'm ashamed to admit it, but I have 20 of these SMD electrical bypass boards. I'm happy to give them away to whoever wants some. They work fine for their core function. But you either have to give up having an LED, or painstakingly replace the mosfet. I'll be honest - soldering a TO-92 transistor to SOT-23 pads is no fun. If no one wants these, they will go to electronics recycling.

 

[JTEX]

^ I would think it's easier to cut and reroute some traces on the PCB to get the pins where they need to be, rather than replace the SMD transistor with a through hole, no?

 

[MattG]

^ I would think it's easier to cut and reroute some traces on the PCB to get the pins where they need to be, rather than replace the SMD transistor with a through hole, no?

I don't know, maybe? I think it's a "pick your poison" kind of thing.

This is the wrong routing around that transistor:


And this is the corrected routing:


I think it would be tricky, since one of those SOT-32 pads needs to be GND, which isn't just a trace, but a whole copper pour. So pad one needs to be removed from the ground pour and re-routed to that relay control signal line. And the other pad has to be removed from the signal control line, and tied to ground. I didn't use thermal vias on the SMD pads because a robot was doing the soldering.

 

[JTEX]

Hmmm... Yeah, not fun. I imagine I'd desolder that transistor, flip it and solder it upside down (assuming I don't break the pins in the process). I've done worse things I'd actually use a fresh transistor if I had a stash, rather than try to reuse the desoldered one.

 

[MattG]

I never thought about trying to solder it upside down. I might just give that a try!

 

[szukalski]

I love to see the evolution, great stuff.

What happened to the "IN GND SW OUT" standard? Then it becomes a drop-in replacement for most PCBs out there. Plus, you can use pin headers to attach which is 10x simpler (and 10x more annoying to de-solder).


If you're looking for further optimisation, you could use 0403 resistors and use a MLCC for that 100uF cap. Should be able to get it pretty small, though a 125B is like riding around in a Cadillac!

 

[MattG]

What happened to the "IN GND SW OUT" standard? Then it becomes a drop-in replacement for most PCBs out there. Plus, you can use pin headers to attach which is 10x simpler (and 10x more annoying to de-solder).

It becomes a drop-in replacement for most PedalPCB PCBs out there. But not for e.g. EffectsLayouts, Aion, Madbean, etc.

I did think about that a bit. But, I decided to deviate from the PPCB standard because...

• I added "BPW", ("board power"), basically just the power input (typically 9v) post reverse-polarity protection diode with some basic filtering. This is to accommodate (some) AionFX boards, as he seems to always put polarity protection on the switch breakout board (and +Vin is usually at the bottom of the actual effect PCB).
• I also added an extra couple ground connections. The purpose of this is twofold: one, it deliberately adds space between input and output, hopefully lowering the chances of crosstalk; and two, again in a nod to AionFX, will (at least in some cases) facilitate hacking in the actual effect circuit components that sometimes end up his breakout boards.
• I also wanted to keep the status LED net away from the inputs and outputs.
• And this arrangement makes actual trace routing slightly easier.

It's always a game of trade-offs. Although I've never personally used it, I do see the appeal (and neatness) of the pin-header attachment method. That would be really useful too if, for example, I wanted to do a quick swap of bypass boards to be buffered vs true bypass.

If you're looking for further optimisation, you could use 0403 resistors and use a MLCC for that 100uF cap. Should be able to get it pretty small, though a 125B is like riding around in a Cadillac!

I might indeed go with smaller SMD components for a future revision. But as for the 100uF MLCC - I'm having these assembled by JLCPCB, and I don't see a 25v 100uF MLCC in their parts library(*). I want this board to be 18v safe. But they do have 25v rated 47uF MLCC. I might switch to a pair of those.

About half of the components in this design are coupling caps and VREF bias resistors. I'm flirting with the idea of adding something like an LT1054 to generate a -9v rail. With a bipolar power supply, I could do away with most of the coupling caps and all the VREF bias resistors. Also, the buffer would have double the headroom. But the tradeoff is of course adding a charge pump to the design, which comes with its own set of issues.

(*) The JLC parts library web interface is terrible. I've never been able to get the parametric search to work reliably, e.g. like Mouser or Digi-Key. The search feature seems the only way to find stuff, and even that is pretty flaky.

 

[JTEX]

The JLC parts library web interface is terrible. I've never been able to get the parametric search to work reliably, e.g. like Mouser or Digi-Key. The search feature seems the only way to find stuff, and even that is pretty flaky.

I usually go straight to LCSC to search for parts to use with JLC's assembly. Much better interface, but I often end up using non-basic parts that incur the extra fees for loading into the pick-and-place machine.