(Log in to disable ads.)

I've been experimenting with amplifying musical instruments with piezos.

I did all the normal experimentation & reading and discovered that piezos are very high impedance devices that need a buffer before being input to a guitar amp.
You can drive them straight into an amplifier & they usually work fine - but the sound is better with the impedance matched by a buffer.

The purpose of the buffer is:
- to allow the piezo to have a high impedance while talking to an amp of much lower impedance.
- and practically isolates the piezo input from the amplifier. Meaning the amplifier doesn't "load" or mess with the piezo.

So - I merrily built a little piezo buffer using a j201 jfet and all was well.
I even put it into a MerryPak tin can that looked quite cute.
Which had previously had my wood wax in it - so it even smells quite nice.

Except, in my application, I was needing 13 of these little piezo's - the application being the amplification of a 13 bar marimba.

So I just joyfully connected the piezos in parallel.

Reading my own explanation above a problem starts suggesting itself.
While the buffer circuit does a fabulous job of isolating a single piezo from an amplifier; conencting the piezos in parallel means they are all connected to each OTHER. Without any buffering.
And can mess with each other.
The signal of any one piezo is being sent to the amplifier (through the buffer) ... but also to all of the other parallel piezos WITHOUT going through any buffer.

What effect does that have?
Not really sure in theory - but in practice I can report that it's not good.
The strongest descriptors are "variable" and "unreliable".
Which are not good words to use for the output of musical instruments.
Especially when someone else is going to be playing them & trying to earn a living while doing so.

Some background - a piezo is a device that converts mechanical energy (e.g. vibration) into electrical energy.
= you whack it & it generates a current
And a hard whack can generate very significant voltages.

It also works in reverse - if you apply electrical energy to a piezo it generates movement.
If the signal is AC then the piezo buzzes.
So a piezo buzzer & a piezo pickup is the same thing.
Which is why cheap piezo buzzers make piezo pickups.

In short, the multiple parallel piezos don't work.
It's worse than that - they initially APPEAR to work.
So they tease you.
And you get excited.
And you record it & it sounds good & you send it to your friends.
And they get excited.

But then very strange things start to happen.

The odd effects include:
- a few of the piezos start sounding loud and others get soft
- and with use it seems to change. One piezo will work fine for a while ... then just go dead quiet. As if its soul is being stolen by the others
- all of them have lower outputs than when connected separately
- the piezos start failing after a while - are the energy spikes from some of the piezos frying the others? Are they just bad piezos that would have failed anyway?

It all seems a little of a Black Art with the interweb providing precious little guidance.
It seems that multiple concurrent piezo applications are very rare with very little international knowledge available.

An extreme measure would be to build a separate jfet buffer for each and every piezo (all 13 of them).
I have already done an obscene amount of soldering; and find my energy a little low right now to build 13 buffers on a whim that it may solve the problem.

As with all new stuff, the internet is full of guesswork suggestions - and it's difficult to separate the wild guesses which are presented with professor-like authority of truth, from the actual useful facts.

Some suggestions include using a series resistor to protect them (seems vaguely reasonable). With others suggesting with equally profound confidence that it's a stupid idea.

I even saw one person suggesting using a diode to isolate them. And everyone on this particular forum wildly agreeing with an imagined group nodding of heads.
The author clearly didn't understand that audio signals are AC sine waves and diodes only conduct DC.
There was a quiet comment later in the thread stating that the diode approach didn't work. The author seemed surprised.

I am beginning to understand why kksound has a starting price of $1900 to amplify marimba's with piezos; even though they are talking about many octave instruments. They also speak about very bespoke and special preamplifiers.

Alas I will not be defeated.
Even if I have to build 13 separate jfet buffers each with their own trimpots to balance the uneven piezo outputs ...

But ... I don't think I'll do this in the next week or two ...
    Even the best piezo pickups are so simple- the complicated part seems to be the buffering and (in the case of the kind if applications you're talking about) seperating signals. I know that there are hexaphonic piezo buffers for individual string separation on guitars- a fe wof those might save you having to hand solder 13 buffers? I found some good prices on Fishman stuff direct from China when I electrified my guitalele. the other option is to use a dual or quad opamp IC, it will at least cut down on the number of components and the cost.

    If you're looking to use a resistor to seperate the piezos, it will need to be a high value relative to the pickup impedance (which then might cost you volume)- and those are in the megaohm range to start with, normally, so if it works, it will only work with a really high value.
      Remember that many cheaper piezos are ceramic or crystalline - which makes them very brittle. This makes them prone to failure when connected directly to a high SPL sound source. That's one reason why mirror tape is often used to affix contact piezos - it adds an extra layer of shock absorption. It's also why the embedded piezos of the Ghost system have a lower failure rate than the Fishman, Baggs, etc.

      Polymer piezos are more durable (Highlander, Fishman Matrix, etc.) but are rarely found aside from custom designs for specific applications.

      Individual buffering is ideal, but yeah, pushes up the cost significantly. The hex guitar stuff is a very specific application and the 13-pin connection is a standard with very specific requirements.

      I have an idea for an electromagnetic pickup system you could try: Embed a ferromagnetic slug in/on your bar and mount a individual pickup element below the bar. Something modular like the Cycfi Research Nu pickups would be ideal, as they are available with integrated preamps. They are designed to work with any string spacing/stagger and any number of pickups can be combined - from a single one to as many as you need. They can mount directly on to a circuit board. They also have some add-ons like filters, active volumes, multipin outputs (both 13-pin and a Lemo connector) etc.
        i would buffer every piezo and then run them all into a simple opamp mixer then use that output to a desk or recording device , much like they do with the ghost pickups



          Alan Ratcliffe wrote: Remember that many cheaper piezos are ceramic or crystalline - which makes them very brittle. This makes them prone to failure when connected directly to a high SPL sound source. That's one reason why mirror tape is often used to affix contact piezos - it adds an extra layer of shock absorption. It's also why the embedded piezos of the Ghost system have a lower failure rate than the Fishman, Baggs, etc.
          I originally encased all of the piezos in epoxy shells to both protect the piezo as well as create screw holes for attachment.
          I abandoned the idea due to low output.
          I incorrectly assumed the primary cause of the low output was the piezos.
          Now realise it was the drain from the parallel wiring.

          Now makes sense that the energy from one piezo pickup is being used to buzz the other piezos - thus less available to be amplified.

          Thank you for the mirror tape idea - I will use it.
          Also thought of putting a layer of epoxy over the top of the piezo to both protect the fragile solder connections, as well as protect the piezo layer.
          Alan Ratcliffe wrote: Individual buffering is ideal, but yeah, pushes up the cost significantly. The hex guitar stuff is a very specific application and the 13-pin connection is a standard with very specific requirements.
          I have already built two buffers based on a simple j201 FET circuit with a couple of caps & some resistors.
          (currently have 6 bars going to one buffer; and 7 to another)

          The parts cost is relatively low.
          Also assisted by inheritting a whole bag of j201's.
          But am going to need to do some board layout & lots more soldering.
          I've tested each bar into a buffer on its own & the sound is good.
          If I'm going to this effort should also have a trimpot per buffer to assist in balancing the sound volume from each bar.

          Am thinking of laying out the buffers side by side on a piece of vero board.
          Need to ponder where exactly to put the board & battery in the instrument.
          Am also pondering which parts of the buffer circuit can be shared across all buffers - to keep parts count down.
          Perhaps the FET's and parts on the piezo side need to be separate; but parts after the FET's can be shared.
          e.g. shared power source & dc isolating capacitor.

          I've been pretty radical with earthing all round - which makes a big difference.
          Using shielded cable. And placing boards into shielded metal boxes
          Alan Ratcliffe wrote: I have an idea for an electromagnetic pickup system you could try: Embed a ferromagnetic slug in/on your bar and mount a individual pickup element below the bar. Something modular like the Cycfi Research Nu pickups would be ideal, as they are available with integrated preamps. They are designed to work with any string spacing/stagger and any number of pickups can be combined - from a single one to as many as you need. They can mount directly on to a circuit board. They also have some add-ons like filters, active volumes, multipin outputs (both 13-pin and a Lemo connector) etc.
          I am liking this idea a lot.
          Especially since there are no wires going anywhere near the bars; which have a nasty habit of buzzing.
          I want to try this next.
            7 days later
            Wizard wrote: Now makes sense that the energy from one piezo pickup is being used to buzz the other piezos - thus less available to be amplified.
            That's why individual buffering is better. You don't have to have a full buffer circuit per element, but rather basically build a summing mixer.
            Also thought of putting a layer of epoxy over the top of the piezo to both protect the fragile solder connections, as well as protect the piezo layer.
            Wouldn't help - too rigid. Piezo ceramics and crystals are prone to cracking from shock because they are hard but brittle. Most epoxies couple and transfer the energy to the element too well. Some isolation is required to insulate and dampen a rigid element from the initial attack of a percussive instrument (which is both high energy and very broad-band).
              Write a Reply...