using capacitors to reduce piezo output - polarized OK?

[chconnor]

Hi - recently installed a couple piezo pickups in my cajon (box drum). The very high signal level was clipping my audio interface, and i learned online that you can wire capacitors in parallel with the element as a means of reducing the output level.

I used a non-polarized 0.1 uF up and it works fine. If I used a polarized capacitor, will it also work? If so, does the orientation of the cap matter? My loose understanding is that capacitors pass AC and block DC, which would seem to imply that polarization/orientation isn't relevant, but my grasp on this stuff is tenuous.

Also, my understanding (and my listening tests seemed to confirm this) is that such an arrangement has "no" effect on the frequency response (within the general parameters likely to be in play, here.) Sound right?

Thanks for any ideas!
-c

[MarkusBass]

I wonder why would you use polarized capacitors if non-polarized work fine. In general the capacity is much smaller than 0.1uF - rather values closer to 0.01uF are used. Maybe this is the reason why polarized capacitors are not used in this case. If you use 0.1uF and still need higher value, I'd say that there is something wrong with the interface.

Mark

[Steve Conner]

A cajon is a wooden box that you bash like a drum, I think extremely high output is possible.

Yes, in theory connecting a capacitor across a piezo will just decrease the output without changing the frequency response. In practice, when the input resistance of your audio interface is taken into account, there will be an increase in bass, as the extra capacitance gives the piezo a lower output impedance at low frequencies. With a 1M "instrument" input this is probably negligible, but it might be noticeable with a 10k "line" input.

Polarised electrolytics should work, but a plastic film capacitor would theoretically be better. Mylar film caps are easily available larger than 1uf.

[J M Fahey]

0.1uF looks reasonable , you needn't larger, so you are not *pushed* onto using polarized ones.
Agree that you may have an interface problem, but I attribute to poor matching.
What input are you using, which interface is it (link), how is it setup?

[MarkusBass]

As Steve mentioned with cajon you may get extremely high signal. I remember that I could easily get signal > 1V so there was no need to amplify it. The transducer needed only high impedance buffer. I was using capacitors between 1 and 3.3n to decrease the signal. If you need 100n or more, it means that either the piezo pickup has very high capaticance (can you tell what pickup you use?), or there is something wrong with the interface (it should be set to minimal gain), or the signal is really huge and you need larger capacitor. But as Steve mentioned, you can easily get capacitors like 1uF-4.7uF non-polarized.

Mark

[chconnor]

Wow, thanks everyone for the very helpful replies!

The pickup is under the dead center of the striking area (for various reasons), so yeah, the signal is probably about as strong as you could get from a piezo short of hitting it with a hammer. :-)

I have no reason or desire to use polarized except that I happened to have a couple handy (and was curious if it mattered).

For now the piezo is going into a 1M-Ohm "instrument" jack of a Presonus Firebox. (My undertsanding is that I want 3-10M for best results, but I didn't want to shell out for the DI box just yet.)

The signal was way too hot for that input (gain all the way down, no boosts engaged, no pads available, etc.) With the 0.1 uF it came in just under the wire... regular peaks at something like -3 dBFS now. My undertsanding is that this kind of consumer-level interface is typically calibrated for optimal average peaks around -10 to -20, with percussive sources peaking at maybe -5 to -10, so I guess ideally I'd bring it down a hair more... I have a couple (polarized) 0.33 uF's I was going to try... think they'll hurt the frequency response much into >= 1M Ohm?

Steve C - what's the theoretically-better part about plastic film vs. mylar plastic film vs. electrolytics? Less noise / better frequency response, etc?

This is the piezo element: http://www.mouser.com/ProductDetail/...8QnKHkhQ%3d%3d

This is the Firebox: http://www.presonus.com/products/FireBox/techspecs

Thanks again!
-c

[MarkusBass]

The piezo element that you used has 20 nF capaticance. That's why you need 100nF capacitor. Just use 2x0.1uF and it will be fine. Typical capacitance of simmilar transducers is 3-4nF. This requires much smaller capacitors (e.g. manufacturers of piezo preamps use 1n capacitors). You can also buy a smaller piezo element but check first that it has a smaller capacitance.
What is wrong with your solution is low impedance of the preamp. This should be at least 3.3MOhm. Otherwise it influences frequency response of the signal. In other words the sound is very poor. You can solve the problem with a simple J-FET buffer (one transistor), which was published either on this forum, or on diyaudio.com.

Mark

[Steve Conner]

Markus, what do you suppose is the -3dB rolloff frequency with 1M and 0.1uF? The frequency response will be fine, because the piezo is so big and then he added even more capacitance. The 3-10M rule of thumb applies to tiny piezos like the ones used in acoustic guitar bridges, that might only have 1000pF of capacitance.

Case in point, I just designed a piezo preamp with an input impedance of 22k, for an industrial ultrasound application.

Peaks to -3dBFS should be fine. I've always run my audio interfaces at about this level. Never heard of this recommendation to keep the peak level down at -10. A digital audio interface should work fine with no distortion as long as the signal is less than 0dBFS, and the hotter the signal you go in with, the better the SNR will be.

Electrolytic capacitors generate more distortion than film caps, and they have a loose tolerance, so the capacitance could change over time. You're not supposed to use them in applications where you need a well-defined capacitance.

[chconnor]

Thanks again, all.

Peaks to -3dBFS should be fine. I've always run my audio interfaces at about this level. Never heard of this recommendation to keep the peak level down at -10. A digital audio interface should work fine with no distortion as long as the signal is less than 0dBFS, and the hotter the signal you go in with, the better the SNR will be.

I'm obiously not the expert, but received wisdom around the web is that many interfaces (especially cheaper ones like the Firebox, I presume) are less tolerant of being pushed. Since digital a/d gear is apparently usually calibrated so that -18 or -20 dbfs ~= "0 VU", the range from -18 to 0 dbfs is considered the headroom and supposedly a lot of interfaces don't hold up as well in that range. I've not noticed a grand difference on my equipment, sonically, but it does make level setting more sane once in-the-box, and from the plethora of posts from experts on the subject, it seems like a wise habit.

http://messageboard.tapeop.com/viewt...=339045#339045
http://www.harmonycentral.com/docs/DOC-2717 ("tracking levels" section)

...and many more. Thus i figured my cajon was probably still peaking too high...

-c

[Steve Conner]

Well, look at it this way. If you never peak above -10dB, you never use the two most significant bits. You paid for a 16-bit ADC but you are only getting 14 bits.

Granted, this is not so important now that even cheap interfaces have 24-bit converters. But still, in the digital world, "headroom" is nothing but a buffer to make sure that unforeseen peaks don't exceed 0dBFS. With 24 bits you can set levels conservatively and be confident of never getting any "overs". -10dB or even less might be appropriate in some situations.

[chconnor]

Yeah, i'm not qualified to argue one way or the other, but many related posts on the subject do make the argument that it's not just about clipping... i.e. that the "headroom" idea still applies in the digital realm for some nebulous reasons having to do with the way the input circuitry was designed to be "optimal" in the -18 dbFS area. Maybe it's all internet legend, or advice that used to be true but just hangs around, but there are an awful lot of qualified-seeming people out there swearing that it's true. I couldn't find a central, well-sourced, authoritative page on the subject, which is a little suspicious. :-) (Also, yeah, the Firebox is 24bit).

Thanks,
-c

[eschertron]

To add to the urban legends, IIRC some of the consumer gear has built-in limiting just to make life easier for the user/recordist. So it's possible that staying below -10dB or -18dB is suggested to prevent the intentional, protective distortion of the signal (limiting) that occurs before the unintentional distortion (clipping).