transformer for piezo vs buffer

[Guitarist]

It sure would be nice if there was a small, quality tranformer for some of these piezo pickups most of us have littering our parts box. Building a preamp to put within an acoustic is a pain sometimes. I did find a couple designs, however, that look to be the best I've glanced at.
But as it's going into my tube preamps or effects unit anyways the impedance issue is most important right now.

Does running a few piezo disks in parallel lower their output Z?

[David Schwab]

The problem with piezo is because of their high capacitance they form a high pass filter, and if feeding a load with too low impedance, you lose low end.

You want about 5 to 10 M input impedance. Of course some people use piezos with no preamp. There was a guy who made basses like that.. the piezo and magnetic pickup were wired together passively, and he felt the loading from the mag pickup made a more natural sounding piezo.

Since piezo's don't have resistance, wiring two in parallel wont do anything, as far as lowering the impedance needs. Many acoustic guitars use multiple piezos in parallel.

Here's an even simpler preamp/buffer:

[Rob Mercure]

David mostly nailed it - any tranny would have to have a 3-10M ohm input winding and crafting a decent frequency response with this much distributed capacitance along with the inductance of this much wire would wind up costing lotsa bucks. But for experimentation purposes you could find the highest impedance OT - I've seen them around 20 K or so - and then up the secondary impedance to reflect a primary impedance in the mega ohm range. The result would probably be very noisy and subject to all sorts of stray electromagnetic fields as well as UFO communication signals but you never know. I've got an octal base Triad input tranny in front of me with two 50 ohm input windings and an 80K secondary. You could reverse the windings on something like this and a 2,500 ohm load on one of the 50 ohm windings should reflect a 4M impedance.
As I noted when responding to the Bogen CHB35 post the grid leak type bias has a really high input impedance and couples well to piezo pickups - if you're into cobbling up tube circuits - as does the "boot strapped" input which utilizes a bit of NFB to up the input impedance.
Unfortunately, nothing that I've experienced will make a piezo pickup sound like anything else so no matter what circuit you use you've got to deal with the "honk" and "quack."

Rob

[Guitarist]

Alright thanks gentlemen,

I guess the transformer is out then. Last night playing my X-Bridge equiped Godin, I wondered how it might sound feeding the preamped piezos in series with the mag pickups as pups in series really darken the tone.
That circuit is really basic, which is good. I guess it really address both the Z issue and cranks up some gain too.
If there is an odd frequency curve that's fairly common then it would be nice to have a simple tone shaping circuit. Maybe just send a 270 pF to ground after the Fet.

I may get some of that piezo wire you were talking about from that other post.
http://www.meas-spec.com/myMeas/down...xial_cable.pdf

...go fot the spiral and lose the outer jacket?

[Artie]

I don't know if this will be of any help, but its a piezo buffer that uses a transformer. However, the transformer is used more for isolation than for the buffering itself.

http://www.jensen-transformers.com/apps_sc.html

Scroll down to "Musical Instrument Application Schematics" and look at document AS098. (Its a PDF.)

Here's the direct link: as098.pdf

[David Schwab]

I may get some of that piezo wire you were talking about from that other post.
http://www.meas-spec.com/myMeas/down...xial_cable.pdf

...go fot the spiral and lose the outer jacket?

I'm planning on doing this for a fretless bass I made. From what I have read, yes, lose the outter jacket and use the braided kind, not the spiral. One person said they remove the braid (or spiral) and paint the insulator with conductive silver paint, and make the outer connection to that.

Here's two sources for piezo coax:

http://www.members.tripod.com/~tclutherie/page6.html

http://www.windworld.com/products/msipu.htm

[MBSetzer]

Keep in mind that a piezo transducer IS a capacitor, but it actually has quite small capacitance itself. Any piezo element which is suitable over the full audio range as an instrument pickup will intrinsically have low enough picofarads to keep from rolling off the highest audio frequencies to ground at the source. That is the first high-pass filter problem; if the cable from the piezo to the preamp is too lenghty there is then enough additional picofarads of cable capacitance (in parallel with the internal piezo picofarads) to where it adds up enough to roll off audible highs and then you get a dull sound.

Now to consider impedance, the problem here can be simplified without having to know everything there is about impedance itself. In this case the piezo simply puts out such a small millivolt-range signal with so little milliamperage oomph behind it, that unless you have at least 5 to 10 megohms of input impedance at your preamp input, like for instance only 1megohm commonly seen on regular guitar amps, a good portion of the signal will be lost straight to ground through that 1meg resistor since it it too low an impedance load for this source, even though its plenty for a magnetic pickup. So to begin with the volume is too low. A mere 1meg *loads down* a small piezo source like it does not do to a magnetic pickup's output. When you drive a 10meg load from a decent piezo, the output signal is not that much lower in amplitude than when you drive a 1meg load from a nominal magnetic PU, so you may not actually need a dedicated piezo preamp for its extra gain as much as for its impedance matching and buffer ability to allow a longer cable to be used between the preamp and the main guitar amp. A 5meg load is not too bad either, a naked ceramic piezo (Yamaha) in a flat-top acoustic without an internal preamp can be excellent straight into my Ampeg V-3 without any preamp at all. You don't have to set the volume very much higher than a regular electric guitar. But you do need to use a short guitar cable or lose high end. Old Ampegs maintained the 4.7M input impedance for many years since they were orignally primarily for amplifying the early piezos of the '50's when they were installed in upright string basses, before the appearance of the solid-body bass guitar. The *amplified peg*.

At the same time the other high-pass filter that David was mentioning begins to take effect. This is kind of like the way you have to make sure that the coupling cap from a plate of a 12AX7 to the grid of the next tube is not too few uF compared to the megohm or fractional megohm of its grid load resistor which goes to ground. (or you will have less bass frequencies appear at that grid) That cap plus the grid load are in series to ground and they themselves combine to form a *high-pass* filter where a full-range signal comes from the 12AX7 plate, but it only allow frequencies above a certain Hz to fully pass through the cap, depending to a definite extent on the ohm value of the grid load. So its a high pass filter but the intention here is for it to let ALL the desirable audio frequencies pass on to the input of the next gain stage, and only filter out the subsonics that are too low to hear and unwanted.
So the analogy here would be the piezo preamp takes the place of the second tube stage in the above example, and the piezo element itself represents the coupling cap where the incoming signal will be coming to the preamp from. However, the piezo is a physical capacitor where you can not select a different capacitance without changing to a piezo with different properties (or putting additional sensors in parallel), so once you have the sensor(s) selected the capacitance is fixed. Therefore you must vary the ohms of preamp input impedance if you want to adjust this high-pass filter so it does not keep too many audible bass frequencies from appearing at the preamp input. In practice once you get it up into the 5 to 10meg range is usually where it starts to sound good. When you get up to 22meg or above it can have too much noise. Even though your choices are limited, you can still get excellent tone when things are fairly optimized.

When you put more sensors in parallel (or maybe use a physically larger enough piece of the same sensor material), it can lower the impedance requirement of the preamp to where a 1meg input impedance is as adequate as a 5meg or 10meg for a single sensor. As long as the piezo material is truly suitable for audio (and you don't introduce too much cable capacitance when wiring them in parallel), you will be adding more signal-producing capability faster than you add physical capacitance, but you will be adding physical capacitance nonetheless. Therefore the absolute increase in picofarads within the combined sensors then allows them to drive a lower-impedance preamp (or regular guitar amp) input like 1meg without being badly loaded down or losing audible bass frequencies. At least for acoustic guitars.
I guess maybe with solid-body bass guitars there may not be enough acoustic energy for the piezo to make much output to begin with, then the requirement for lower octave(s) to pass through the series high-pass may combine to give you frustratingly diminishing returns when you experiment with multiple sensors.
Plus it seems like it would be kind of a black art not unlike paralleling speakers of differing ohms and db efficiency, so its probably easiest to work with multiple identical sensors rather than trying to parallel widely varying sizes or types in positions with dissimilar levels of acoustic vibration as well.

Now maybe that coaxial oil-well-logging piezo wire could really blow someone's mind. I expect the oilfield service companies discard thousands of feet of this stuff per year. The physical capacitor here is not a crystallized ceramic (acting as dielectric) having opposite sides metallized to form capacitor plates (at the same time being the + & - signal sources). Instead with coaxial the inner conductor acts as one plate of the capacitor, then the shield acts as the opposing plate (just like noisy guitar cord). The dielectric coaxial insulation is a special plastic having piezoelectric properties so the unterminated wire itself produces signal depending on acoustic impact. Hopefully a lot more than just noisy guitar cord.
Maybe you could just carefully clip the open end of the coaxial making sure there was no shorting between the shield and the inner conductor, then glue or otherwise attach enough length of the piezo coaxial underneath the top of the acoustic guitar to where it puts out enough signal to do the job into the selected outboard preamp, when the other end of the coaxial is stripped and soldered to the phone jack just like regular coaxial. When somebody looked inside the guitar, on careful inspection they could see no battery, pickup, or preamp, just the place where it looks there should be a preamp, pickup, or obvious signal source but the wire has been cut clean off
Music comes out anyway. It could happen.
That would be a head-scratcher

Anyway, with something like the coaxial or PVDF film piezos, it would seem like you would not need to put more sensors in parallel (for increased output into lower impedances for example) unless you wanted to amplify specific discrete locations upon the guitar body. Using a single piece of bulk sensor material having additional length instead actually does act like having additional sensors in parallel, even though it seems like having longer coaxial cable would be series rather than parallel.

But something tells me we may not yet have advanced enough materials for the output increase to exceed the capacitance increase as you increase sensor length, except for lengths of sensor placed directly under the saddles where the acoustic impact is greatest.

It would be interesting to see signal data comparing a ceramic piezo under-saddle sensor's output to that of the stripped & silvered coaxial (assumed to be highest output without a shock-absorbing outer layer or two which could serve as an acoustic damper) as well as the PVDF film when all other things such as the guitar, saddle, string gage & tension were identical. Carefully measured raw sensor output amplitude using a very high impedance low-noise ocilloscope, as well as additional readings when loaded as a test rig with various metal-film resistor values such as 500K, 1meg, 2meg, 5meg, 10meg & 20meg should allow the differences in acoustic suitability and impedance drive capability to be quantified with some degree of accuracy.

Mike

[Guitarist]

Ok, If one were to use a raw piezo into an amp or modded stomp box with a modded high Z input, what would be a good cable brand ot type to minimize noise and such?

Just built the fet preamp cable tonight and it's hard to complain not having to somehow install a preamp in the guitar.

I also ordered and pick up a fist full of NTE123ap FETs (I think) but now I can't recall what preamp called for them, dang it.

[Luijo]

123's are not FETs, they are NPN trans, as far as I can remember.

[Steve Conner]

MB: Piezo pickups have huge capacitance, several thousand pF if I remember rightly. They only work because the physical principle by which they generate the signal has a built-in 6dB/octave treble boost that cancels out the 6dB/oct rolloff of the huge capacitor across it.

Too low shunt resistance makes this 6dB/oct rolloff crap out at low frequencies, exposing the underlying treble boost.

Of course, the consequence of this is that cable capacitance doesn't matter for raw piezos. Capacitance shunting a piezo doesn't dull the sound: it just lowers the level. Of course if the cable capacitance were shunting the output of a 4.7Meg volume control you put across the piezo, that would be another story.

So, since capacitance is the only thing that matters about shielded audio cable, the kind of cable doesn't matter. All that matters is shunt resistance of the amplifier that the cable is hooked to. It's easy to get tens of megohms with, say, a FET input op-amp.

I used to use piezo accelerometers for industrial vibration work, that use a special charge amplifier front end to work almost down to DC. I tried a charge amplifier for piezo guitar pickups once, but it sounded like crap for some reason that I never figured out. I also tried making a piezo bass pickup based on an accelerometer design, using one of those piezo beepers configured as a bender, with one side screwed to the guitar body and a large metal nut glued to the other side to act as a seismic mass. I found that I could get massive output voltage with a real low-end thump that sounded like an upright bass, but in order to do that I had to make the seismic mass so big, that the weight broke the piezo

P.S. I once repaired a Takamine acoustic that used what looked like 6 of the piezo cylinders from BBQ lighters, wedged under the bridge.

[hewo]

I have a single ended el84 guitar amp that uses tl072 input buffering fed into single ecc83, varying volume and tone via filtered energy bleeds. I beefed it up pretty good so the power supply could be more prompt, the ot wouldn't cause hinderance, and the power tube could breathe as much dc as it wanted to on signal demand. anyway now the sound engineer wants the paper speaker's voicing signal. so i got this extremely puny dynamic (dynamic? 1.2kohm dc resistance) microphone element extracted out of an aircraft pilot's headset mouthpiece that i want to send signal of the paper speaker. do i need a buffer preamp? i can sound isolate mount the element on the inside of the speaker's steel perforated screen facing the paper speaker centroid. then i can check what ac rms is generated from the element playing the guitar. but yet, how do i know whether the element's output really deserves (is it justified?) a buffer (active 2SK170-GR toshiba jfet). i can't ask the aircraft's manufacturer what kind of receiving impedance their headset amplifier utilizes. church stage has a master input (xlr) floor box for a dozen inputs feeding the sound engineer in the auditorium rear. this means a trunkline extends from stage to that distant engineer. how can signal even be transmitted that distance without 1st bringing it up to at least one volt? thanks in advance

Keep in mind that a piezo transducer IS a capacitor, but it actually has quite small capacitance itself. Any piezo element which is suitable over the full audio range as an instrument pickup will intrinsically have low enough picofarads to keep from rolling off the highest audio frequencies to ground at the source. That is the first high-pass filter problem; if the cable from the piezo to the preamp is too lenghty there is then enough additional picofarads of cable capacitance (in parallel with the internal piezo picofarads) to where it adds up enough to roll off audible highs and then you get a dull sound.

Now to consider impedance, the problem here can be simplified without having to know everything there is about impedance itself. In this case the piezo simply puts out such a small millivolt-range signal with so little milliamperage oomph behind it, that unless you have at least 5 to 10 megohms of input impedance at your preamp input, like for instance only 1megohm commonly seen on regular guitar amps, a good portion of the signal will be lost straight to ground through that 1meg resistor since it it too low an impedance load for this source, even though its plenty for a magnetic pickup. So to begin with the volume is too low. A mere 1meg *loads down* a small piezo source like it does not do to a magnetic pickup's output. When you drive a 10meg load from a decent piezo, the output signal is not that much lower in amplitude than when you drive a 1meg load from a nominal magnetic PU, so you may not actually need a dedicated piezo preamp for its extra gain as much as for its impedance matching and buffer ability to allow a longer cable to be used between the preamp and the main guitar amp. A 5meg load is not too bad either, a naked ceramic piezo (Yamaha) in a flat-top acoustic without an internal preamp can be excellent straight into my Ampeg V-3 without any preamp at all. You don't have to set the volume very much higher than a regular electric guitar. But you do need to use a short guitar cable or lose high end. Old Ampegs maintained the 4.7M input impedance for many years since they were orignally primarily for amplifying the early piezos of the '50's when they were installed in upright string basses, before the appearance of the solid-body bass guitar. The *amplified peg*.

At the same time the other high-pass filter that David was mentioning begins to take effect. This is kind of like the way you have to make sure that the coupling cap from a plate of a 12AX7 to the grid of the next tube is not too few uF compared to the megohm or fractional megohm of its grid load resistor which goes to ground. (or you will have less bass frequencies appear at that grid) That cap plus the grid load are in series to ground and they themselves combine to form a *high-pass* filter where a full-range signal comes from the 12AX7 plate, but it only allow frequencies above a certain Hz to fully pass through the cap, depending to a definite extent on the ohm value of the grid load. So its a high pass filter but the intention here is for it to let ALL the desirable audio frequencies pass on to the input of the next gain stage, and only filter out the subsonics that are too low to hear and unwanted.
So the analogy here would be the piezo preamp takes the place of the second tube stage in the above example, and the piezo element itself represents the coupling cap where the incoming signal will be coming to the preamp from. However, the piezo is a physical capacitor where you can not select a different capacitance without changing to a piezo with different properties (or putting additional sensors in parallel), so once you have the sensor(s) selected the capacitance is fixed. Therefore you must vary the ohms of preamp input impedance if you want to adjust this high-pass filter so it does not keep too many audible bass frequencies from appearing at the preamp input. In practice once you get it up into the 5 to 10meg range is usually where it starts to sound good. When you get up to 22meg or above it can have too much noise. Even though your choices are limited, you can still get excellent tone when things are fairly optimized.

When you put more sensors in parallel (or maybe use a physically larger enough piece of the same sensor material), it can lower the impedance requirement of the preamp to where a 1meg input impedance is as adequate as a 5meg or 10meg for a single sensor. As long as the piezo material is truly suitable for audio (and you don't introduce too much cable capacitance when wiring them in parallel), you will be adding more signal-producing capability faster than you add physical capacitance, but you will be adding physical capacitance nonetheless. Therefore the absolute increase in picofarads within the combined sensors then allows them to drive a lower-impedance preamp (or regular guitar amp) input like 1meg without being badly loaded down or losing audible bass frequencies. At least for acoustic guitars.
I guess maybe with solid-body bass guitars there may not be enough acoustic energy for the piezo to make much output to begin with, then the requirement for lower octave(s) to pass through the series high-pass may combine to give you frustratingly diminishing returns when you experiment with multiple sensors.
Plus it seems like it would be kind of a black art not unlike paralleling speakers of differing ohms and db efficiency, so its probably easiest to work with multiple identical sensors rather than trying to parallel widely varying sizes or types in positions with dissimilar levels of acoustic vibration as well.

Now maybe that coaxial oil-well-logging piezo wire could really blow someone's mind. I expect the oilfield service companies discard thousands of feet of this stuff per year. The physical capacitor here is not a crystallized ceramic (acting as dielectric) having opposite sides metallized to form capacitor plates (at the same time being the + & - signal sources). Instead with coaxial the inner conductor acts as one plate of the capacitor, then the shield acts as the opposing plate (just like noisy guitar cord). The dielectric coaxial insulation is a special plastic having piezoelectric properties so the unterminated wire itself produces signal depending on acoustic impact. Hopefully a lot more than just noisy guitar cord.
Maybe you could just carefully clip the open end of the coaxial making sure there was no shorting between the shield and the inner conductor, then glue or otherwise attach enough length of the piezo coaxial underneath the top of the acoustic guitar to where it puts out enough signal to do the job into the selected outboard preamp, when the other end of the coaxial is stripped and soldered to the phone jack just like regular coaxial. When somebody looked inside the guitar, on careful inspection they could see no battery, pickup, or preamp, just the place where it looks there should be a preamp, pickup, or obvious signal source but the wire has been cut clean off
Music comes out anyway. It could happen.
That would be a head-scratcher

Anyway, with something like the coaxial or PVDF film piezos, it would seem like you would not need to put more sensors in parallel (for increased output into lower impedances for example) unless you wanted to amplify specific discrete locations upon the guitar body. Using a single piece of bulk sensor material having additional length instead actually does act like having additional sensors in parallel, even though it seems like having longer coaxial cable would be series rather than parallel.

But something tells me we may not yet have advanced enough materials for the output increase to exceed the capacitance increase as you increase sensor length, except for lengths of sensor placed directly under the saddles where the acoustic impact is greatest.

It would be interesting to see signal data comparing a ceramic piezo under-saddle sensor's output to that of the stripped & silvered coaxial (assumed to be highest output without a shock-absorbing outer layer or two which could serve as an acoustic damper) as well as the PVDF film when all other things such as the guitar, saddle, string gage & tension were identical. Carefully measured raw sensor output amplitude using a very high impedance low-noise ocilloscope, as well as additional readings when loaded as a test rig with various metal-film resistor values such as 500K, 1meg, 2meg, 5meg, 10meg & 20meg should allow the differences in acoustic suitability and impedance drive capability to be quantified with some degree of accuracy.

Mike