signal routing of pedal effects, using MAX 394 (analog switches)

[PeterPan]

I'm basically building an MCU driven patch controller for pedal effects. There's much more too it but that's a simplified explanation. The idea is nothing new, its simply to allow a one single patch selection to trigger various pedals into either "effect" or "bypass" mode, to gain the benefits of a multi effects processor. Anyway in a previous post (which has gotten buried), I spoke my disappointment routing signals using analog Opto isolators (like vactrols). The principal worked, but due to the slow operation of the optical parts, there would be a short time when switching between bypass and effect modes that the two would get momentarily shorted together, and that would cause an unwanted "chirp". So I've decided to forgo this approach. That MAX394 chip contains 4 analog SPDT switches which are rated for audio use, and guaranteed to have "break before make" behavior. They are a bit pricey (around $7), but Intersil has them for about 1/2 that. It works out cheaper than all those OPTOs anyway. 2 chips will let me control 8 pedals, which would have taken 16 Opto isolators.

So I'm kind of sold on using these chips for my signal routing. But since I'm no longer using truly isolated controls (as with the opto isolators), I have to pay a little more attention to biasing. These chips will operate fine from a 5V supply, and its easy to establish a 2.5V midpoint analog ground using op amps. So 2.5V becomes analog "ground" for my circuit, which is fine because in this case there is no chance that any of the connecting pedals will be run off the same power supply as my control circuit. So the issue is that even though some pedal outputs will have a true DC reference to their own ground, there is a very good chance that a lot of analog pedals may have their own internal DC blocking capacitors on both input and output. In those cases, any DC ground referencing of their signals will be gone. Again this wouldn't matter with the purely resistive opto-isolators, but it will with the MAX394 chips. So I propose to compensate for this by putting a fixed resistor at the output of each effect pedal, as it enters a channel of my MAX394 chip. These resistors will simply tie the outputs to analog ground (2.5V), and I'm thinking 10K would be low enough to ensure fast settling of the bias voltage (on initial power up), while presenting minimal load to the output of any pedal. I could probably use 100K instead.

So I'm attaching a simple hand drawn schematic of just the analog / pedal part of this scheme, in this case showing only 4 pedal channels. It works on the bench, and it seems to work fine with combinations of different pedals, running from either batteries or power adapters, and doesn't seem to mind whether the pedals have their own adapters or are powered by a common supply. The OP amps are all TI LME49721s, very good quality rail to rail low voltage op amps, that have always performed well for me with little fuss. But still, Id like some opinions, especially on any "gotchas". Too many time, especially in the analog world, everything will be fine until I go through the trouble of cutting PC boards, and THAT is when i realize I should have added some stupid capacitors.

I'm just putting a link to the circuit here so tyou can look it it in your browser, because The IMG tags on this forum don't display it properly.

http://pixyland.org/temp/max394RoutingScheme.JPG

[R.G.]

If you're working on switching pedal on/off/routing, you might take a look at a video crosspoint switch.

Here's one: http://www.microsemi.com/products/switches/analog-cross-point-switches/mt8808 and it's big brother the MT8816.

I used this in a one-off board I built a few years back. Works well, but you have to worry about keeping the signals down to 2.6V pk-pk or it distorts. Otherwise it's a single-chip solution to routing up to four pedals in any order. Two of them or the 8816 will do eight pedals, any order. They run under software control and can literally patch any input to any output in a non-blocking fashion.

The problem with this kind of thing is not the signal switching - it's the user interface. Getting a comprehensible way for a guitarist to specify and control on the fly what connects to what without using a whole computer to do it is not simple. Of course, using an entire laptop to do the control might be reasonable as prices fall.

[PeterPan]

If you're working on switching pedal on/off/routing, you might take a look at a video crosspoint switch.

Here's one: http://www.microsemi.com/products/switches/analog-cross-point-switches/mt8808 and it's big brother the MT8816.

I used this in a one-off board I built a few years back. Works well, but you have to worry about keeping the signals down to 2.6V pk-pk or it distorts. Otherwise it's a single-chip solution to routing up to four pedals in any order. Two of them or the 8816 will do eight pedals, any order. They run under software control and can literally patch any input to any output in a non-blocking fashion.

The problem with this kind of thing is not the signal switching - it's the user interface. Getting a comprehensible way for a guitarist to specify and control on the fly what connects to what without using a whole computer to do it is not simple. Of course, using an entire laptop to do the control might be reasonable as prices fall.

Well I appreciate that. I think I'm satisfied with the part I chose, as I don't really intend to go the distance of letting the user select the effect chain order programmatically. This is actually an add on to a very big project that has been the victim of so much product creep (LOL!) that i have to draw a line at some point. There's a whole other analog section I didn't mention to allow "per patch" level settings, not to mention the wireless controller. (I won't bore you  ). But I THINK that even with the part you're suggesting, the issue of biasing (or at least stabilizing the "zero" center point to a reference) can be a problem if not addressed. So for now, I'm just looking to see what obvious issues I might be forgetting.

[mth5044]

I think, with the big post, you are asking if you need to bias the signal? The answer is yes, in the middlish so you don't clip the supply rails. If you could post a link to the datasheet of the part you are talking about, it would be helpful.

[PeterPan]

I think, with the big post, you are asking if you need to bias the signal? The answer is yes, in the middlish so you don't clip the supply rails. If you could post a link to the datasheet of the part you are talking about, it would be helpful.

OK thanks... Here's the data sheet for the Max 394...

http://datasheets.maximintegrated.com/en/ds/MAX394.pdf

I guess I know the signal has to be biased, but I'm always glad when I let others see if I'm missing anything. Thanks for any advise or comments!

[PeterPan]

Anyway, for anyone interested, 100K biasing resistors to analog ground have worked fine so far.  I've also found at least one drop in substitute for this IC, The intercil ISL8394IB, available at Rochester.com for $4.37. Rochester has a $50 minimum policy, but that's a good price, several dollars less then the Maxim chip. Frankly, to do the same switching with my original scheme (analog opto-isolators) would have required 8 optos to replace the functionality of just one of these ICs, so even though its a pricy chip its still a more cost effective solution. I also should report that I have sampled both the MAX and Intercil parts and have been bench testing them in my circuit, and I have to say they both work equally well. No clicks or pops apparent during switching, nor the telltale feedback "chirp" caused by momentary overlap during switching (they are guaranteed "break before make").

Both maxim and Intercil were very good about samples. In fact the more expensive Maxim had no problem when i requested 6 samples! For experimentation on a white board, I've also created a simple "breakout adapter" PCB layout. I'll include an image here containing 3 IC patterns. You can scale and use this to etch some temp PC boards (if you can't scale it, let me know and I'll make you a Gerber or a PDF