"Stompbox" Stomp Box

[PBE6]

A friend of mine asked me to look at a small homemade "stompbox" another friend made for him. I can't open it up because it's glued shut, but I assume it's got a passive piezo sensor inside. He's a bit frustrated with it because it doesn't sound full enough. Ideally, it should be making a dull thud (really low, like 30 Hz) when he hits it with his foot. Right now it's a bit too high-pitched and missing the low end.

My first thought was that the box was too small to be putting out significant low end thump, but putting it through Logic Pro's multimeter it looks like the bass frequencies are actually quite predominant. The real problem seems to be that the signal is very quiet, something like -70 dB in the low end. I'd like to try boosting the signal and filtering it heavily to remove almost everything above 50 Hz, but I'm a bit worried that too much of a boost is just going to result is excessive hum and/or noise.

Does anyone have any tips on how to make a very quiet preamp boost stage for a signal like this, subject to piezo constraints? I was thinking an opamp gain stage with very large (10 M) pull down/Vref resistors would keep the impedance high for the piezo, and I could filter the high end out with a double-pole Sallen-Key filter centered at 50 Hz. Hopefully any noise will be filtered out too. But is this asking too much? The gain is going to have to be in the neighbourhood of 1000x = 60 dB, which means a 1 M feedback resistor over a 1 k ground resistor. Would it be better to split the amplification into two stages instead?

[brianq]

I also have one of these stompboxes, with the same problem, with mine the piezo had a huge glob of hot glue between it & the box. This muted & muffled the sound & took away important vibration. If you get it open check if it's snug against the box, it needs to be able to pick up more vibration for more pronounced percussion. Vibration is very important for piezo to work, I bet it's surrounded in glue.

[bluebunny]

You might want to put a buffer close to the piezo to avoid losing that bass thump.

[anotherjim]

I've seen them used with a graphic equalizer pedal. It may be a common need to buffer and boost what you want out of it.

[samhay]

^Does anyone have any tips on how to make a very quiet preamp boost stage for a signal like this...

Your instincts are good. I would go for high input impedance - maybe bootstrapped - and lots of gain spread over multiple stages to keep the op-amps happy and the thermal hiss from large resistors to a minimum. Use a good quality cable between this and the pickup that is as short as reasonably possible.

You might find that you want to keep more of the signal than you thought. A steep 50 Hz filter might remove a few harmonics that make it sound like a wooden box rather than a thump.

[ashcat_lt]

My first thought was also about the impedance that the piezo sees.  Those things are basically capacitors in series with a voltage source, which creates an HPF with the load of whatever it's plugged into as the R to ground.  If that impedance is too low, you just can't get much bass out of it.  I've never really messed with them, but from what I've heard, they want to see like 10M or more for decent response.

[PBE6]

Thanks for the replies all.

How about something like this?


High impedance input into a 100x max variable gain stage with a shelving filter around 160 Hz, into a 56x fixed gain stage with a shelving filter around 130 Hz, into a Sallen-Key filter with a variable frequency that goes as low as 5 Hz (although the centre of my dual 500k log pot will be about 50k which puts the frequency around 56 Hz).

The overall gain will be 5600x max (which is about 75 dB) for the band limited signal, which should be more than enough. It might make more sense from a noise perspective to make the first stage 100x fixed and make the second stage variable with a 50k pot, any thoughts? I may also end up increasing the filter caps in both gain stages, as my friend is looking for more of a tuneless "whump" than a woodblock sound.

[anotherjim]

It's HF that a low load impedance on the piezo will cut most, although it's high source impedance means the entire output would be harmed too.
Just for buffering, the basic JFET guitar buffer circuit will do. A "T" peaking filter to give a hump about 80hz could follow.

A more sophisticated approach is to use the piezo to trigger a synth type drum voice, such as the TR808 kick.

[PBE6]

I will think about using a peak boost. The box itself gives off a lot of scrapey handling noise, so in addition to boosting the thump I want to get rid of the high stuff. A LP filter will be necessary at some stage.

[ashcat_lt]

It's HF that a low load impedance on the piezo will cut most, although it's high source impedance means the entire output would be harmed too.
Just for buffering, the basic JFET guitar buffer circuit will do. A "T" peaking filter to give a hump about 80hz could follow.

A more sophisticated approach is to use the piezo to trigger a synth type drum voice, such as the TR808 kick.

If'n there's cable capacitance between the piezo and the buffer, then you might lose treble too, but the piezo is a capacitor, and capacitors in series make high pass filters, and when R gets small, the cutoff frequency gets big, so you lose bass.

[stallik]

I made a number of these using piezo sounders and got the best result by mounting the piezo not flat against a surface but by one edge only with the remainder floating horizontally. I'm now using a unit I bought from peterman in Australia which employs a small speaker stiffened with a bit of plastic across the cone and a signal transformer. It sounds huge and deep BUT, like my own units, the shoes I'm wearing make a really big difference to the sound. I know that just sounds like mojo carp but it's true.
I've mounted the peterman in a cast box and now a routed mahogany block and while the enclosure made a bit of a difference, the footwear made more.

[PBE6]

Very interesting. I've posted this before, but there's an interesting article from the American Journal of Botany about the acoustic properties of different woods:

http://m.amjbot.org/content/93/10/1439.full

One part that was interesting to me was that the higher the mass of the object, the lower the resonant frequency, and the less stiff the material the lower the frequency.  I got to thinking, it's difficult to find a heavy softwood for a nice low tone, but what about a composite object? My friend says another friend has a homemade stomper made out of a license plate wrapped in leather and it apparently sounds good - heavy middle with a nice soft exterior! Same could be said of the shoes, softer soles would probably encourage the whump as opposed to the woodblock sound. Just speculation of course, but it seems to make sense at first glance.

[PBE6]

Ended up making a few changes to my plan, here's an as-built schematic:



It turns out that 40 dB is more than enough for good amplification, not sure why my readings were so different from a few days ago to last night. One problem though - when I turn the gain all the way up I'm getting a crunchy pop spike (along with the normal sound) whenever I stomp on the box. Neither the preamp nor the Logic channel are showing an overload, so I assume it's the input opamp? That's weird too, as I ended up using TLC2272s which are supposed to be rail-to-rail. Does anyone see anything obviously wrong in the input section? I've never used 20M as the Vref resistor before, I wonder if that could be causing problems.

[stallik]

http://m.amjbot.org/content/93/10/1439.full

Thanks for the link, really interesting

[PBE6]

Quick thought..if the (assumed) piezo works like a capacitor, could my tiny series input capacitor being lowering the overall capacitance of the sensor system? And if so, would that cause any pop/spike issues?

[anotherjim]

You probably don't really need 20M input, try 1M and go up from there. You'd need to change the input cap up to 10nF-ish while trying this. 20M is going to be very sensitive to little noises. The pcb is going to have to be, and remain, scrupulously clean if you keep 20M. A thorough degreasing and a conformal coating (pcb lacquer) called for? I've seen (manufacturers) advice to keep such high impedance connections up in the air - bend the amp + pin up and out and solder the bias resistor and input cap to it - will keep it clean & dry.

The 220pF input could well be too small, I don't know the value of an average piezo transducer's capacitance. The 220pF isn't a ceramic is it? Unless a decent COG or NPO, that could be a factor - but you say the noise is only at high gain?

Have you checked the DC level out of the first amp? Having an electro to ground for the inverting circuit has a risk of offset error if the cap's a bit leaky.

The cracking noise might be defects in the piezo that you simply haven't heard with previous amplification - they are very brittle and this one might have some little fractures.

[PBE6]

Found some useful info on piezo signal conditioning at the Texas Instruments website:

http://www.ti.com/lit/an/sloa033a/sloa033a.pdf

[PBE6]

I tried adding some more filtering stages to my circuit and now it's oscillating at high gain levels. I'm thinking the crunchy spike was the start of an oscillation, I'm going to try rearranging the circuit so that the band limiting comes first then add the gain stage at the end. The noise is minimal anyway because the hiss is cut out by the filtering.

[Beo]

I tried replicating the porchboard using some cheap proximity sensors, but I couldn't get it to work. I'd love to figure out the right kind of sensor that would work. This seems like a better solution than piezos. I saw an irish duo in Dublin using one of these and it was a great accompaniment.

http://porchboard.com/

[brianq]

Another thing to think about is to replace Piezo with a quality one. 9 times outta 10 there's a cheap Radio Shack buzzer element in it. Those suck! Most likely your the bulk of your problem. Good luck on that.