Passive compressor for taming a piezo pickup

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

gyraf

Well-known member
Staff member
GDIY Supporter
Moderator
Joined
Jun 4, 2004
Messages
11,856
Location
Aarhus, Denmark
Warning: theorizing only post!

Hi Guys,

I recently had a discussion with some musical-instrument guys about pickup systems - specially the piezo type.

It seems that there was (at least some) consensus that the harsh sound you'd most often get from these systems isn't because of the pickups acoustic/mechanic properties themselves, but because of the very-very-large peaks in signal level at the attack of e.g. picking a string, which in turn overloads the preamplifier stage, if this is set to a gain that is anywhere usable for "the rest" of the signal. And this initial overload causes the annoying sonic signature of the piezo systems.

So basically a problem in too-large a dynamic range. And compressing later on doesn't help much, as it tends to clip already in impedance-converter stage.

Which makes me think: Why not make a full-passive fast compressor to act on the signal already before it leaves the instrument?

Couldn't it be done this way:

A simple FET voltage divider, with relatively high series-resistance (in order not to load the signal too much during attenuation) - and a separate "sidechain" piezo device, that directly drives the gate of the FET through a simple passive rectifier-and-capacitor timing. Yes, small capacitor and large resistors, but FET should have no problem with that. Probably some sort of a detector diode needed, germanium?

Should be possible entirely "passive", no battery needed - as the power needed to turn on the FET is easily generated by the second piezo (that can be placed either close to the "audio" pickup, or where the "thump" is louder than what you'd want in your audio).

As this is theory-and-speculation only - my question is: What did I miss, why (if not) hasn't this been done?

Jakob E.
 
Fun. Last week I was setting the monitors for a performer on stage, who had an awful ac. guitar sound, exactly what you describe. Horrible picking attacks. I was wondering if the DI input wasn't overloading. That might well be why so many live tech love passive DI boxes for accoustic guitars. Despite the reflected impedance probleme, they have ample headroom and the overload characteristics might be softer. Your solution seems new to me, and very interesting, of course... I have nothing to add to feed the discution, but I'll follow it with interest.
Laurent.
 
Jakob,

Part of the problem is that a piezo is essentially a small value capacitor is series with the voltage output. This means that you have to drive it into a quite large impedance on the next stage or you will have loss of bass response, and a harsh sounding signal. 

Also, since the signal can be quite large, you probably need to power the high Z stage with 18v to get more range. This type of buffer can produce good sounds from a piezo pickup.

Even so, you idea is interesting and merits some additional development. 

You would need to rectify the signal from the sidechain piezo with a Ge diode, but if you don't put a capacitor after that, the control signal will have extremely fast attack and decay.  If you use it only as a peak limiter it might be okay, but my thoughts are that the quick decay will sound odd.

regards, Jack
 
I'd be very interested in helping with this, both my electric and acoustic guitars have piezo pickups, and I hardly use either due to the very brittle attack when using a pick.  If, for example, a small device -- even one powered by a 9V battery -- could be implemented to resolve some of this I'd be on board to help prototype, test, etc.  Since both my guitars use 9V batteries for their piezo systems anyway, I see no reason we couldn't tap some of that power if necessary.

 
A common solution has been to use s-limiters to kill the 3-4 k peak that sounds so ugly in the attack, but thats often to late in the singnal chain to avoid dist at some point earlier on.  Very interestig idea with a second piezo to drive the sidechain. If this works, you have a potensial sale of at least a million units, if it will make a piezo sound nice :)
 
This is a wonderful idea!  I have a cajon that I use at least 2x a week on gigs that I have installed piezo pickups in.  The sound is only really manageable with compression or by using a mic as well.  I would be more than happy to help in any way I can...  prototyping, providing some before/after sound samples, etc.  I think in many applications (and at least in mine) there is gain to give up in exchange for more manageable response...


-Jason


 
Just simmed an active idea for this.  I am looking for the capacitance of piezos that are used.  To adjust it correctly one would need to know the desired output level.

The circuit is simple and if I was to build it I would improve the circuit design from how it is drawn.
 
piezoclampamp.PNG


The main idea is the use of the pad to not clip the JFET or a tube or an opamp.  Then to have a limiter,compressor,EQ etc after the buffer/pad.

C2 and C1 are guesses I do not have a piezo pickup to measure.
 
The control voltage should probably be bigger than the signal voltage.

The control must be a large fraction of FET Vgto, the signal must be a _small_ fraction of Vgto.

Recall what you have in studio FET limiters: 50mV-100mV limit threshold, 2V-7V of control voltage.

Large resistance leads to large hiss. Hiss is not generally a problem on piezos; maybe it could be.

If it is overloading, hang a large cap across the piezo. Simple capacitive voltage divider. This also makes bass-loss less of a problem.

 
A very interesting subject, and clearly a fresh look at this age-old problem.
Acoustic instruments solve the problem by providing an almost infinite number of acoustic paths, which spread the original impulse over a longer time.
One solution (although not a passive one) may be to process the signal through a number of all-pass delays. Just an idea...
 
While I have mixed feelings about this, the schematics posted so far seem unrelated to the initial concept.

Just off the top of my head, the input connected such that the signal flows through the drain source channel, to a fixed resistance to ground forming a pad. The JFET channel is low impedance when Vgs is 0V, but a cap from gate to source is charged up by signal swing when it exceeds some threshold. As the Vgs increases, the channel conducts less and pads down the signal.

This would be crude and add some distortion of it's own, since the non linear load caused by charging the gate voltage capacitor would reflect into the source impedance.

Note this doesn't need a second piezo input but would work off the actual input signal (assuming enough voltage swing).

I am not very optimistic about the fidelity of this, (or even that it will work) and technically it is not actually passive, while it draws it's power from the signal source. This gate cap can be very tiny, only limited by the availability of a large enough resistor to set reasonable release time. There may be some clever way to use leakage currents to make the release, but this doesn't seem very stable or reliable for establishing time constants.

JR
 

 
I wonder about the fidelity of the initial assumptions  ;)

Who has done some analysis of the root cause?

When I designed the Electrolyre, I built a "tester" consisting of a pair of Tynex (nylon) strings, each with it's own Graph-Tech piezo pickup. One string is tuned about middle C, 20.5" long and .050" diameter, and another a couple of octaves up, 7" long and .025" diameter. The two pickups are wired in parallel with a 40M ohm resistor across them.

Here is a one-shot scope trace of the waveform taken with a 10M ohm scope probe using a Tek 2430A storage scope. The lower pitch string was plucked about a 45 degree angle normal to the plane of the bridge and a pretty aggressive pluck. The string was plucked about 1/3 of the total length from the bridge using a finger pad.

Note the peak voltage is about 3 volts and occurs around the time that the string backwave arrives at the bridge. By comparison, I can get 10 volts peak out of my Les Paul Studio and some fat chords.

I designed the Electrolyre with a voltage gain of 10dB from the pickups to it's output jack to normalize the output to electric guitar levels such that it drives amps and effects pedals properly. This hasn't caused any problem with overloading any inputs.

I'm more concerned about the 20mS of waveform offset right after the pluck, which causes an audible thump.
 

Attachments

  • 200hz-5mS.jpg
    200hz-5mS.jpg
    48.3 KB · Views: 23
Thanx... It's always good to test our ASSumptions with real data..

I only helped on one piezo preamp and IIRC that was 9V battery powered and didn't saturate, but I only helped on that project from a distance. I helped a junior engineer in a different engineering group, so I mainly pointed him the in a good direction and he ran with my advice. The preamp was in a special artist instrument (a bass guitar), and this artist, who had pretty good ears was happy with the result. 

JR
 
I use piezos daily on stage:
http://rattletree.com/phpBB3/viewtopic.php?f=19&t=30

For my needs, I just have an in-line tone control and always use a tube amp.  I have always heard the problem is the impedance mismatch of piezos to the input stage of an amp.  You guys all know a lot more about electronics than me.  But I do know that a tube amp makes my instrument sound good.  Maybe there is compression happening in the tube that is evening out those spikes?
 
Back
Top