This is the solution to the tinny contact mic sound problem – at least the electrical matching issue. A FET amplifier will sort out the mismatch, and give a useful amount of gain for a contact mic.
By using a separate box for the battery and load resistor the circuit is automatically powered when you plug in the piezo device. By keeping the FET close to the piezo disk you maximize signal levels and reduce the extent of the high impedance part of the circuit, which has a tendency to pick up hum. You also raise the signal level on the interconnecting cable run, which also helps to improve the system signal to noise ratio.
Diodes D1,D2 stop you destroying the FET with large signals from the piezo device if you drop it. You can leave C2 out if you need less gain in your application. Musical instruments and anything where you bash the sounding structure with an object fall into the leave C2 out territory
The FET can be the typical 2N3819 or a BF245A/B. You may be able to get away with RF FETs like the J309 but you may find that the input noise starts to rise at audio frequencies.
My original version of this (no diodes, C2 and slightly different source resistor) I glue the contact mic using epoxy resin to the back of the actuator magnet of an old hard disk. This is a nice strong magnet, and since many resonant structures are steel you get an instant easy good contact with the object. I tested it out on a footbridge
The original recording
The same thing simulated into a 7k mic input
Ugh. Somewhat tinny…
Now try using a 50k line input
less tinny, but the 200Hz rolloff still takes the guts out of the recording. Many of the things you want to record with a contact mic are whacking great big structures that have BASS so you don’t want to go throwing it out at the off.The bad news, however, is that the manufacturers of FETs don’t control their parameters well, and have somehow conned us into living with the problem. The gate-source voltage needed to bias the transistor into the linear region can vary between 0.25V and 8V, which leaves a good 7.75V down to a hopeless 0.4V for the transistor and load if used with a typical NiCad 8.4V PP3
You’ll have to get more FETs than you need and throw out the dogs. It’s easy enough to test, and this parameter is a given for a particular device – it doesn’t age or change greatly with temperature.
Design manuals get all sniffy about that sort of thing because selecting FETs obviously adds to the cost if you are mass producing something. That’s not the case here, and there’s just no way to cope with a manufacturing tolerance which can throw more than 90% of the battery voltage away in variations in manufacture without screening the bad ‘uns. Ideally you’d run the FET from a higher power supply voltage, like two batteries in series and perhaps double the values of R2 and R3, but it would be a shame to have to use two batteries just because the manufacturers couldn’t be bothered to grade by Vgs.
You can tell if you have a good ‘un by measuring the voltage at the drain and source of the FET in circuit. Ideally you would like Vs to be about 2.5V and Vd to be about 6V (assuming a 8.4V Nicad PP3)
In practice you can live with Vs at 1 to 3.5V which will correspond to a Vd of 7.4 to 4.9. This will run the FET at 0.25mA to 0.9mA
I prototyped this and tested it out with all the spare FETs I had in my junkbox, on a supply voltage of 8.5V
| device | Vs | Vd | usable |
|---|---|---|---|
| 2N3819 #1 | 2.07 | 6.3 | OK |
| 2N3819 #2 | 2.03 | 6.2 | OK |
| 2N3819 #3 | 1.4 | 6.8 | OK |
| J309 #1 | 2.1 | 6.1 | OK |
| J309 #2 | 2.27 | 5.98 | OK |
| BF244B | 2.39 | 5.86 | OK |
| BF244 | 4 | 4.2 | BAD |
If you really can’t live with the tastelessness of the select on test FET, don’t want to use two batteries or are looking for the lowest noise performance, you can do better. The majority of piezo contact mic apps just don’t need that performance, but here it is.
a version of this to work from plug-in-power on field recorders
What if I’ll get a guitar piezo with a preamp that runs on 9v? Will it perform better? However these use contact mics that look like sticks (often with a tiny jack that plugs into the preamp) and I’ve no idea how useful they are for recording.
Guitar piezos also want a high impedance not to sound tinny, so if you use a preamp matched to a guitar piezo that should work OK. I think a guitar piezo is usually mounted between the bridge and therest fo the instrument, so it has a reaction mass and is in compression. As recordists we don’t usually use a reaction mass, we are trying to pick up flexing in a metal surface. I don’t know if a guitar pickup has the ceramic material oriented differently or if that would matter.
The great thing about the ceramic disc piezos are that they are dirt cheap on Ebay, however 😉
Hi Richard,
is it possible to upload a higher resolution image of the circuit (image 1). Its very hard to see what the components are.
Regards,
Chip
I don’t have that any more. It’s a very uncritical circuit – though sadly R2 you may have to select on test as described. I used 3k9 for the drain resistor R3 and for the source resistor R2. C2 bypassing the source resistor R2 is 22uF, it wants to have an impedance of less than half R2 at the lowest frequency you want to handle.
R1 wants to be twice the impedance of the piezo self capacitance at the lowest frequency, for the reasons described here. If you don’t know the self-capacitance, use 1Meg to 10 Meg. 1Megohm is absolutely fine for the average disc piezo.
Hi Richard and Chip,
For what it’s worth, I am about to try my hand at making one of these, and ran the original image through Photoshop to up the saturation. From that, it looks like these are the components shown:
Y1, Piezo
R1, 3.9m (or 1M is typically fine, from comment above)
R2, 3.9k (though, I’m not sure which test you are referring to in your above comment)
R3, 3.9k
D1, 1n4148
D2, 1n4148
C1, 4.7uF
C2, 22uF
Q1 2n3819 OR BF245A/B
BTI 8.4V Nicad PP3
Hopefully that helps someone along the way!
Another thing that might be worth considering is that it looks like the 2n3819s are becoming obsolete, with some component suppliers no longer carrying stock. Richard, do you have any ideas for substitutes? I am just a composer, so the specifics of all of this are beyond me. I’m neck deep in datasheets and making no headway…
Thanks a million for putting this all up. The three blogs you have relating to piezos are by far the most helpful I was able to find anywhere! I’m looking forward to making some sound soon.
Best regards,
David
Leaded FETs are getting rarer these days, Digikey don’t seem to list any standard leaded JFETs any more. The 2N3819 was the standard JFET back in the day. BF244 and MPF102 were also common, the MPF102 was more audio-oriented, the BF244 was a VHF device. High-frequency devices can have more 1/f noise that gives a rumble at low audio frequencies, though for a contact mic you usually have more than enough signal. Note the pinouts vary. I’m almost starting to wonder if I should point people more towards the opamp version, there’s less hurt getting the parts.
The test for VGS is described in the second half of this, in the section where I moan starting
This reddit seems to indicate 2n3819 availability was acceptable 4 years ago and that the J111 was available. Never heard of that one, but the datasheet indicates VGS(off) is between -3 and -10v, which is only slightly more disgraceful than the 2N3819’s -3 to -8V. IDSS has a massive variation on the J111, though not technically worse than the 3819.
If you want to give the circuit a quick blast with minimal parts omit the diodes. Once you have checked it works then think about adding them. You will get away without until you are unlucky enough to drop the disc glued to a magnet onto a hard floor, whereupon you may get to change the FET 😉
I have been a bit obsessed with applications for contact mics since I learned of them. I had thought to do the NE5534 way for 5 of them, to mic drums (bass, floor, snare, and two toms) in order for the impact to not be lost when placing other microphones to pic them up farther away, and powering them from the phantom power on the mixer.
I’ve been told it’d be better to just use FETs to do it (although the quality may not come through as well) because of the voltage requirements for the op-amp.
Just thinking, would definitely need the diodes in to protect the circuit, would not need anything to add gain…do you have any advice about what other modifications would be better in a high impact type of situation?
Don’t bother with the diodes for drums, the problem is a field recordist sometimes drops their piezos, or they snap them sharply onto a hunk of steel. For drums you might also want to get rid of C2 (22uF) – you are going to have stacks of signal so eliminating it will improve overload, plus in theory you’ll get a slightly better really low-frequency response.
However BE WARNED that phantom power from a mixer is P48 power, ie 48V via two 6k8 resistors to XLR 2 & 3, and 0V to the cable sheath on pin 1. Whoever told you to use this with the FET was wrong if you think you are going to phantom power it. It won’t work, and if you try running the FET against the sheath, grounding the cold side (ie connecting 3 to 1) and running the FET drain into pin 2 is likely to end in misery because the 2n3819 has a max VDS of 25V. Same for the J309. You might get lucky and have a FET that pulls enough current to drop that across the 6k8 resistor but you are much more likely to get a fried FET and some loud noises.
You have three choices:
Use the design on this page with the 9V battery and go in single ended to your mixer – run XLR pin 3 to ground and pin 2 to the output
Or if you’re in a music studio try using an active DI box. This one has an input impedance of 220kohm, which is a teeny bit low – on this page I calculate the minimum input impedance wants to be 330k with the sort of piezo discs I used.
Alternatively, you could use something like the schematic for Scott Helmke’s Alice microphone, which is actually designed for running off P48. You don’t need to go nuts with the 1 Gig resistor in that design – slap a 1Meg to 10Meg from gate to ground, because any piezo disc has way more capacitance than an electret mic. But the parts count is a lot higher.
I had forgotten I had commented here! Thank you so much for all the info. It was …maybe Jules from the micbuilders group that recommended using the FET rather than the opamp, and that being with my idea to use one phantom power source’s voltage divided across 5 piezos (4 of them going to 1/4” jacks and just one using the clear).
Funny you brought up Helmke, I believe he is a part of that group, and they’ve remade the Alice many ways.
I am so glad for all you’ve done here though! Now I’m thinking separate FET assemblies for each drum, no C2 or diodes, with piezo attached for no long cables to pickup RF or increase capacitance. Maybe coin cell batteries could work? Anyway, then run cables to the mixer from each.
Thanks again for all your brilliance!
The lower diode is inverted !
Remember that this is a depletion mode FET so while the gate is roughly ground, the source is slightly positive due to the current flowing through R2. This is not a NPN BJT where the base is expected to be positive w.r.t. the emitter
What you do not want is gate current by biasing the gate +ve w.r.t. the source, the FET is only good for 10mA of that at Abs max rating
Hi Richard, great article! Did you have problems with audio levels saturating and if so, how would you go about attenuating the signal to the transistor/opamp without affecting the response?
I’ve ever put anything loud enough into this to saturate 😉 What on earth are you putting into it?
If it’s saturating on normal contact mic signals, first check your drain voltage is around 6V, you might have a FET at the edge of spec. Either swap FETs until it is or consider dropping R3, which will reduce output and raise Vd.
If your input is really loud then use the FET as a source follower, take out C2 and D2, short R3 and take the output from the source via a 10uF capacitor. You may need to play with the value of R2 to bias the source to roughly 4.5V. You then need to supply power, signal and ground separately
To begin, I lack knowledge of electrical particulars, but am keen to experiment with piezo contact mics to capture vibration object recordings. I understand a high impedance pre amp is required. Would the high z pre amp on my Focusrite ISA 1 work to boost sufficient level on a standard cheapo piezo mic? Or a Zoom H4n?
Failing that, I’d be interested in either buying a pre amp and piezo mic set, off the shelf in the UK.
Any pointers most welcomed. Apologies for my lack of knowledge to act on the thread above.
Many thanks, Sam
From their specs
Use the focusrite, and set it to whatever you need to do tho make this happen
because this input looks like a classic DI box. So you’re all set. Use what you have 😉
How bout this? It uses a LSK389B . It is a preamp and impedance buffer for input from two piezoelectric elements. It has balanced inputs and outputs, and uses 48 volt phantom power. It could perhaps use voltage regulation to enable 24 to 48 volts phantom power.
https://github.com/Supermagnum/piezo-balanced
Schematic diagram:
https://github.com/Supermagnum/piezo-balanced/blob/main/piezo-jfet.pdf
Look like a riff off the Alex Rice design – nothing wrong with that, though I have no personal experience of it.
I am confused by the two piezo disk layout. Intuitively you would want to ground the brass disc part of each, in which case the wanted signal would appear as a common mode signal that the diff amp would do its very best to cancel out. So you’d have to mount one the other way, since you probably aren’t going to be able to ring up the manufacturer and ask them to make you a few with the ceramic element mounted in opposite polarity, which would be the right way to do that.
That will make your hum pickup profile very unbalanced, to the point that I’d personally start off with one element and connect the diff amp to either side, accepting the unbalanced hum pickup.
But hey, this is armchair theorising – get it on a bench
Nice find of the LSK389, though you may find a cheaper device works fine
I would slightly worry about the 48V exceeding VGS max as you plug it in. You’ll be fine in steady state, with 6k8 P48 series resistors if you are running at say 4mA you’ll drop 10V in the supply resistors and a little bit more in the parts in your design. You can get round that in bench testing, but in thefield, well, you know what roadies are like 😉
It’s tested and works. The crystals produces a signal that is 180 degrees out of phase with the other as they have different polarity.
This is also tested and works, but a strange phenomenon sometimes happens: This circuit is connected to a pair of piezoelectric discs wired in balance, the cable has 3 conductors and shield.
It’s 5 meters long.
https://github.com/Supermagnum/double-gain/blob/main/double-gain.pdf
Sometimes a high pitched whine occurs, it has significant strength. It sounds like a feedback howl.
The output cable has 3 conductors, wired in balance. It’s 1 meter long .
What is the cause and cure for that?
I built this yesterday using a BF245A. It seems to be working very well. Here’s a picture https://instagram.com/p/CaNSAXkJAl0/
Looks good. You might want to flip up the piezo to get the brass disc in contact with what you are listening to, but hey, if it works, go for it. The non brass side can be delicate and crack if flexed too much 😉
Hey Richard,
Thanks for putting in all this hard work.
Was wondering would something like this work, cigar box guitars often use pizeo pickups and frankly I wouldn’t have the time to build the above. So looking for a complete unit I can hack to work with my little piezo mic.
Thanks a bunch!
https://www.cbgitty.com/news/easy-diy-preamp-makes-piezoequipped-cigar-box-guitars-sound-better/
If you’re a musician you probably already have a DI box, try that first for an easy win.
No questions, just thanks. I don’t know why the almighty Google was so unsatisfied with my entreaties, but it took me a good bit of sifting through a bunch of repetitive and useless (to me anyway) information before I finally managed to find this page.
Thanks for posting this design online. I’m going to try to build this. I’m curious as to the reason for not biasing the gate? Wouldn’t we want to have a quiescent working point in the saturation domain? Without gate biasing, I can’t see what happens when the piezo output has shifted polarity (i.e. half of the time) – doesn’t drain current just drop to zero then?
The negative gate voltage required to bias into conduction is achieved by the gate being fuctionally grounded via R1 (no gate current flows, to a very good approximation) . The source, however, is raised a little bit positive relative to ground and therefore the gate by the drain current flowing. As more drain current flows the source is raised more positive (=gate made more negative) which reduced drain current and so a balance is reached.
The value of VGS is poorly controlled hence the rant about select on test, though my results show most of the jfets I tried were fine.
Thank you for all the information provided here. I’m new to electronics but keen to learn and play with contact microphones. I was wondering if this amplifier would be able to drive headphones by itself and if so, where volume control could be added?
I was hoping to make a small, portable unit to listen live through headphones to sounds in trees, grass etc . Would that be possible with this circuit or does the output have to go via a recorder? Thank you for any help and apologies for a potentially silly question!
Nope, sadly this one won’t drive headphones by itself. To a first approximation the ouput impedance is the value of the drain resistor, 3.9k, if you go and put a typical 32 ohm headphone across that (through the capacitor C1, where 2*pi*f*C < 32 ohms at the lowest frequency of interest, ie bigger than the 4.7u shown) you will form a potential divider. Output will be 32/3932 or less than a hundredth, and you won't hear anything, particularly outside! The Chinese do make headphone amps you can buy on ebay, but if you're going to carry around a box like that you may as well pack a small recorder and be able to record anything interesting. If you want to build it without a recorder, run a LM386 on a piece of veroboard and take TI’s app note fig 9.1