I originally wrote this guide as a post on Tagboard FX, where additional useful discussion is kept. 

​

The big blue 3PDT footswitch has been the go-to actuator for true bypass switching for who knows how long, and it has served its purpose well with a high success rate and reliable operation. Nevertheless, a popular upgrade and increasingly frequent inclusion in high-end stompboxes is the so-called “soft latch” true bypass system, which uses an electronic circuit actuated by a momentary switch to toggle a small relay that handles the signal routing. The momentary switch doesn’t produce the characteristic mechanical “thunk” or "clack" that the 3PDT does; rather, all you hear is the tick of the relay, giving rise to the “soft-latch” moniker. Systems like these are well worth the effort it takes to implement them, as they hardly ever pop because there is no substantial mechanical component, and the smooth, detent-free actuation lends a satisfying, professional feel to a pedal. I’ve become a big soft-latch nerd, and have tried most of the circuits commonly available today, so I thought I’d write a little guide to help those who might be interested in getting into it.

Soft-latch systems have historically been a bit expensive, due to the far larger number of components necessary to make it work versus a purely mechanical system or even something like the Millennium Bypass and related systems. My main goal has been to find a circuit cheap enough and reliable enough such that I could put a soft-latch in every pedal I built and do away with the mechanical 3PDTs as a matter of course. A short overview of what’s available follows.

​

AMZ FX DPDT Switch Board: When I began my journey into soft-latch, I began by using Jack Orman’s “True Bypass with Relay PCB"s, which were an inch-square PCB designed and dedicated to true bypass switching. He has since discontinued this product and instead sells a general-purpose DPDT relay switch PCB which can be configured for true bypass switching quite easily (I have used both). The circuit is AVR microcontroller-based, uses subminiature single-coil latching relays like the Takamisawa AL5WN-K, Panasonic TQ2-L-5V, or NEC EA2-5S, and has a low parts count. Advantages to this circuit are that it is very physically small and can fit virtually anywhere, the current draw is low and the switching is reliable and crisp. The circuit flashes the indicator LED on power-up, so you know everything is working as intended, which is a nice feature. Disadvantages are that there is a flywire required for the true-bypass configuration, which doesn’t look great and is a bit difficult to keep out of the way given its location. This becomes worse given that I strongly recommend you wire this system to ground the effect input on bypass, requiring more flywires. The unit cost is also expensive, at least for me, at $7 for the board and AVR plus $7 shipping to Canada (although that will obviously vary with your location). This is not including the other components necessary to make the thing work; Mr. Orman sells this as a complete kit for $14 plus shipping. Other systems like this exist all over the place, which are functionally identical. Though there’s nothing negative to say about the functionality, the sky-high unit price is a bit of a roadblock to being able to include one in every build.

​

Mictester’s CMOS module: Developed iteratively from a circuit by Mictester on FreeStompBoxes, this circuit uses a 40106 Schmitt trigger to actuate a single-coil latching relay like the AL5WN-K. The parts count is moderate, and there are PCB layouts available that are nice and small (1.2” x 1.4”, or 30 x 36 mm), however the nature of the parts involved makes it awkward to lay out on vero and thus leaves the footprint a bit on the large side. Nevertheless, it will fit inside a 1590B so long as low-profile parts are used. The actuation on this circuit is also nice and defect-free, however there are some issues with it.  Firstly, the power-on state is not guaranteed and there isn’t any way to know what it will be unless you extend the circuit to include it (this is possible, and is discussed in the original thread on FSB). The one that I built as a tester defaults to “engaged” when power is applied (seemingly a common problem), which may not be acceptable for some implementations. Secondly, the circuit doesn’t much like rapid engage/bypass cycles; there is a (short, but noticeable) minimum time required between button presses to get it to cycle properly. In most cases it should not be an issue, but still something to note. Cost is low, I estimate $3-4 based on Tayda and eBay prices.

​

The Andy Relay: Developed by the famous R.G. Keen, this circuit uses a 4069 hex inverter to actuate a dual-coil latching relay. One of the best circuits of this type I have used; the switching is very reliable and there is no extraneous noise to speak of. The PCB is 1.8” x 0.94” or 46 x 24 mm, and so is easy to fit in 1590B and larger boxes at various locations. It also includes wire-routing holes for strain relief, which is very welcome. It uses very common components with a moderate parts count. The only disadvantage to speak of is that it uses dual-coil latching relays rather than single-coil ones, and the relay “of choice” is the Panasonic TQ2-L2-5V. This part is more expensive than the single-coil relays even if you get them from eBay and the like, and the big-box retailers (e.g. Mouser) want to charge $8 apiece for it, which would double the overall cost of the system. Other similar relays could be used, although most are equally difficult to find, not to mention at a good price. The NEC EA2-5TNFG work brilliantly in this circuit, and they can be found for a very reasonable ~$2.50 USD apiece from eBay (not quite as cheap as the AL5WN-K and its congeners, but manageable nonetheless), and are what I employ in my boards. I have used many of these to date and continue to use them; the system cost is slightly higher than something like the Mictester module, but the quality of switching and of construction is well worthwhile. Do note the wiring of the LED if you are to build this; it has caused much confusion. Additionally, and very importantly, the silkscreen on the OSH Park boards (see below) have an error in them; the 10K resistor immediately to the right of the top-right corner of the 4069 must be 100K. If a 10K is placed in this position, the relay will not toggle (although the LED will still work correctly).

UPDATE September 2022 - OSH Park used to sell PCBs for this circuit, but the shared project has disappeared from the site a few years ago and not returned. If you have previously ordered one of these, you can order them again, but I'm not sure at the moment what is to be done if someone wants to obtain these boards for the first time. Currently the only recourse seems to be to get them from someone who has previously ordered them, or etch them/have them fabricated yourself.

​

Pruttelherrie’s “Generic I/O Board”: Introduced to me by Pruttelherrie on FreeStompBoxes, this very simple system uses an ATTiny85 microcontroller to drive a single-coil latching relay. The code is available freely on the original FSB thread. The parts count is very low, and so even on vero the board can be made small enough to fit just about anywhere. The switching quality is very good. The default power-up state can be toggled, which I think is a cool feature. After having used this circuit a few times, I like it a lot as well. The major disadvantage is the up-front entry cost and relative complexity; you’ll need something to program the microcontrollers and a computer of course, and there are a number of solutions of varying price to accomplish this. I chose to use an Arduino Uno as an in-system programmer, which cost me all of $7 to set up. From there you can load the code onto the ATTiny85s, and you’re good to go. System cost is similar to the Andy Relay. Also recommended.

UPDATE September 2022 - I had been getting the ATTiny85-20PU microcontrollers for this project from Tayda, but due to the supply chain disruptions as a result of COVID, they have been out of stock of them for what may be years now. Fingers crossed they come back soon. 

​

Celestial Bypass: I've been spending a lot of time on the PedalPCB Community Forum recently, and in that forum there is a lovely repository of expertise run by a gentleman that goes by Chuck D. Bones called, appropriately, "Chuck's Boneyard". One of the many circuit design and modification articles and discussions in his subforum contains his take on the PedalPCB Basic Relay Bypass module, a low-parts-count inexpensive bypass module which uses a 555 timer to toggle and hold a non-latching relay. There are many attractive advantages to this design. Firstly, no microcontroller, transistors or large CMOS chips are required, which significantly cuts down on board size and setup complexity. Secondly, it has a low count (14) of extremely common components, which means that the total system cost is as low as $4 (based on Tayda and eBay prices). There are some clever design choices in this circuit that mean that it can drive any relay coil up to 9V without the need of a voltage regulator, and it allows the use of the 555's more modern CMOS cousin, the 7555 timer, for lower current draw. This turns out to be a useful feature, since the module's only disadvantage is the non-latching relay. Any such relay should work fine; I have been using Takamisawa A5W-K (the non-latching version of my favourite AL5WN-K) or NEC EA2-5. The circuit works fantastically well; so well, in fact, that I am including it on board all Deep Harbour Devices PCBs moving forward. This is truly a silver bullet for the "put a soft-latch in every pedal" dream. 

​

COMING SOON: 

​

Paulinthelab Soft Latch Daughterboard: One of the first DIY soft-latch systems to surface if I remember correctly, but not one without its issues. This one uses a transistor network to actuate a single-coil non-latching DPCO relay, and people have reported significant popping and poor isolation issues with this circuit. Although it is not included, this circuit needs a flyback diode across the relay coil, but even when this is added issues are still reported. Another major disadvantage is that since the relay it uses doesn’t latch, the coil must be energized as long as the pedal is engaged, making this unsuitable for a battery-powered build. Unit cost looks to be quite low. Previously I had dismissed this project as an inferior option compared to the others discussed above, but I think that I was being unfair in doing so given that I hadn't actually tried it myself, so I plan to do so and report back about it at some point.

​

Demedash Incandenza: A super simple, low-parts-count and dirt-cheap relay bypass module based on a 555 flip-flop circuit. I'm working on trying this one out and will report back about it at some point.