How to measure performance of piezo contact mics / pickups?

[ZDL]

We all know what we mean by the frequency response of a condenser or dynamic mic, but how would you define and measure frequency response for a contact mic? By its nature, the mic is measuring the resonance and frequency response of the object it is attached to. How can I glean the response of the mic itself?
I recently published this short article on the subject, and described how we test our mics in our lab; but it's just a relative, comparative test. I am stuck on how to establish a more absolute measure of contact mic performance. What are your thoughts?

[Mick Bailey]

With your existing setup how do you resolve the frequency response of the driver, as it's both limited at either end of the frequency range and non-linear. Both of these factors will affect the perceived response of the mic.

[ZDL]

We don't resolve it! That's just it! We considered mounting the piezo directly to the magnet of the driver to eliminate the resonator & coupler, but I am still left with the nonlinear response and peakiness of the driver itself. Maybe I should build a mechanical vibrator, built around a cam, so I can fix the amplitude of displacement and precisely vary the frequency. Hmmm, I may be on to something...

[J M Fahey]

a) you can't, as simple as that.
Why?
Because although you might hang one from thin wire inside an anechoic chamber and sweep a speaker nearby for testing, they are simply not used that way, except, maybe, as a school physics lab experiment, definitely not the way we use them.

b) so they will always be attached to some vibrating object which will define vibrations and modes.

c) if anything, study how glueing or attaching a contact microphone may/will affect said object resonance and properties.

As in: I can imagine/envision a piezo disk epoxied to one arm of a tuning fork deviating it from true A 440 so much as to make it unusable (or needing compensation for the added mass).

Edit: just read your article, I congratulate you on testing (whatever the outcome) , simply add that a paper cone conventional speaker is a way poorer transducer than the Piezo pickup, Piezo can go from a couple Hz or even lower if properly loaded to tens, maybe hundreds of kHz ... try that with *any* dynamic speaker.

**Maybe** (just thinking aloud) , you could glue new microphone under test to an existing "reference" driver , also a Piezo disk, maybe larger and sweep.
In any case, I fear final installation will have the main voice on this.
Once piezo is glued/pressed, it becomes part of the instrument.
While "air" microphones do not.

[km6xz]

I just ran into the need to verify a contact pickup mounted in a ukulele. The acoustic instrument amp was suspected of being the problem by the customer. She had taken it to 3 different techs and they all gave up, saying it was just noisy and they are all like that. The amp had a lot of gain and was maxed out, about 65db and with driven by a 1mv signal, very low for a contact transducer, the volume was high and noise inaudible. The instrument had a preamp mounted in it built by B-Band. Testing that revealed that it was fine also, in fact very good. So focused on the pickup. Output was very low above 30-60 hertz. I figure a contact pickup is a displacement generator so voltage out should mimic the frequency response of conventional pressure gradient devices instead of Velocity devices. First conclusion was probably a bad transducer but thought of no failure mechanism that would result in such a response so imagined something else was afoot. The transducer was sealed into the top wood just behind the soundhole so assumed a luthier would be needed. I never work on this sort of thing so was not sure what the output level would have been normally. I loosened the strings and found the saddle very stiff in its U channel and with effort was able to pry it out.A magnifying glass showed a little dirt or oxide on top of the transducer but otherwise seemed ok, I carefully cleaned the top of the transducer with micro-swabs and IPA. reinstalled the saddle and strings and found the level had increased dramatically but still HF shy. Took the saddle out again and focused on the wood U channel and cleaned it with slightly damp lens cleaning paper. The saddle would not even fit back in so I dried the wooden channel with my hot air desoldering station and the saddle fit better and HF response was "normal". Took it all out again and using very fine abrasive, polished the sides of the ivory saddle. Fit was much better and response was amazingly flat and output was as strong as a single coil guitar pickup. The owner was watching me the whole time and seeing the incremental improvements, picked it up and played a number and was beaming. I learned a lot about contact transducers from that situation. The conclusion is the system performance is greatly impacted by saddle to transducer face to body interface. It also got me thinking about how to measure its performance, discovering it to be displacement type response, the response is complex, with being selective in the direction of travel of the wave across the saddle and cancelling on axis of the transducer, A transducer using point sources would sound very different, with waves reflected from all directions would partially cancel leaving the vertical displacement component as the one driving the transducer. So a test exciter would get different response depending on the axis of the various reflections, if a point contact or strip contact. 3 axis accelerometers would be useful is analyzing the phase and timing of the waves exciting the transducer.

[J M Fahey]

Thanks for the detailed report

So basically this contact microphone was not making proper (physical) contact.
Very reasonable, just don't get why that was not found earlier, from the "luthier" who installed it originally to all others.

I bet all thought : "hey, it's an under saddle pickup ... it's under the saddle .... so that's not the problem"

Problem being, of course, that floating even 0.1mm (printer paper thickness) away from soundboard/saddle/both stops it working, period.

User made up for horrible lack of sensitivity by raising preamp gain ... welcome noise and hum

[Leo_Gnardo]

So basically this contact microphone was not making proper (physical) contact.
Very reasonable, just don't get why that was not found earlier, from the "luthier" who installed it originally to all others.

I bet all thought : "hey, it's an under saddle pickup ... it's under the saddle .... so that's not the problem"

For all those "matchstick" piezo pickups, without pressure from being squeezed between the saddle and bottom of the saddle cutout, they will not work at all. True that the piezo turns out voltage as the result of pressure changes from the saddle's vibration, but without the "bias" of a pre existing pressure, no dice.

There's a corollary to this observation: the bottom of the saddle can be carved to eliminate excess pressure/response. Often a B or G string will "holler too loud"; this can be fixed. I was taught to glue a thin sliver of wood - spruce a good choice - onto the bottom of the saddle. Then the piezo element is sandwiched wood to wood - for that woody tone of course - and the wood can be easily carved away in oversensitive areas. Of course the overall height of the saddle increases. The best bet if following this scheme is to cut/sand material from the bottom of the plastic or bone saddle matching the dimension of the wood added so the guitar's action winds up not being changed in the end.

[J M Fahey]

Just yesterday night I repaired my son friend's nylon string guitar with a non working saddle pickup and it was also working poorly, I untrusted the saddle slot so I scratched it with a sharpened screwdriver until saddle floated freely, so much so that if I held guitar face down it would fall to the floor.
And if it didn't work properly, I was ready to Krazy glue a couple piezo buttons inside, just under wooden bridge ends and use them instead, they never fail, have great tone, and being high capacitance , usually around .04uF or so, drive any reasonable amp input even without a preamp.
And they pick up a little more soundboard vibration than under saddle sticks which mainly pick strings and nothing else.

[ZDL]

It got me thinking about how to measure its performance, discovering it to be displacement type response, the response is complex, with being selective in the direction of travel of the wave across the saddle and cancelling on axis of the transducer...

This is a great observation. We find, of course, that piezos behave VERY differently with even small differences of installation. A major factor is whether you bond the entire face of the piezo disc to the soundboard, or if you have the opportunity to cantilever a portion of the disc off the edge of something, out into space. This affects the vibration modes available to the disc. In most cases, it seems to greatly increase low-end response. I had not considered your observation about wave cancellation. I think if I develop a "standard" for testing response, it will have to include families of response curves depicting several application conditions, somewhat like how we develop sets of response curves based on distance and direction for air-mikes.

[ZDL]

Yes, it's a crazy chicken-and-egg predicament, a paradox couched in a dilemma wrapped in an enigma.

[michael krzyzaniak]

There are a few accepted methods:

1) There is the Hsu-Nielson method that involves cracking a pencil graphite on the surface of the transducer. This produces a unit-step function. From this, you can recover the frequency response of the mic independently of any sound transmission medium. The procedure is described in a standard known as ASTM E976. I have a quick example with a little more detail on my blog: https://ambisynth.blogspot.com/2019/...icrophone.html

2) If your mic is reciprocal (e.g. it can be used also as a speaker), you can put two of them back-to-back with double-sided tape, play a unit impulse through one and record the response of the other. If you take the dft of that and divide it by two (because you had two identical 'microphones'), you will have the frequency response. There is a good explanation in a paper called "Frequency response of transducers used in acoustic emission testing of concrete" by Berthelot, Soudat, et al.

3) If your mic is not reciprocal, you can use a triaxial (three-microphone) method, with two reciprocal sensors in addition to your mic. I can't describe it briefly, but there is a clean description in a paper called "Reciprocity calibration method for ultrasonic piezoelectric transducers in air. Comparison of finite element modelling and experimental measurements." by Andersen, Søvik et al.

[J M Fahey]

Thanks.
I guess that even if not specifically "meant" to be a "speaker" , a Piezo disk might be "speaker enough" to drive an identical twin.
So your measurement method could be applied.

FWIW I would not even use double sided tape but plain *epoxy* (or krazy glue) them together.

Even making them useless as pickups might be justified in the name of Science

[ZDL]

Ingenious. That ASTM standard, like most, is hilariously arbitrary, but hey, it's a start. The other two methods are very intriguing. I'll give those papers a look. Thanks for the leads!