Single-supply piezo-preamp circuit for e-guitar

[Simon Schneiter]

¡Hello Everybody!

I'm working on a design using a piezo guitar-pickup that should be
conditioned for a A/D-converter (an then to a DSP and then ... but
thats another (long) story = ). The whole thing should be small and
not use too much power.

I've designed a circuit using single-rail supply, since the
A/D-converter requires the signal to be between 1V and 4V. I assumed
the peak output voltages of the piezo to be at +10V respectively -10V.
After looking around for quite a while I settled with a circuit I
found in "OpAmps For Everyone" which I found on the TI Website
(http://focus.ti.com/lit/an/slod006b/slod006b.pdf).

I read threads in this newsgroup and found some discussions about
designs for dual-supply opamps (e.g. "Piezo audio preamp works but I
would like advice" dating july'04).

I'm pretty lost about how to model the actual piezo-transducer and I'm
waiting to get the pickup delivered. So I couldn't do any mesurements
on them yet.(http://www.graphtech.com/product_display.asp?SearchType=brand&brand=Ghost)

I'm using a MAX4494 which I found after reading on the RANE-website
(http://www.rane.com/hilovolt.htm). The A/D-converter should be a
PCM1802 from Burr Brown (now with TI).

I'd be very happy if someone might have a look on the circuit I've
planned so far. I'm unsure if using single-supply altogether is a good
idea? Or should I look out for a opamp that has a FET-input stage to
achieve a even higher input impedance? Is it any worth to model the
piezo-pickup as a voltage source in series with a ~680pF capacitance,
or do I mislead myself using this in a spice-simulation? Is there
another circuit out there that fits my needs better?

I'm a graduate student in electrical engeneering, but not experienced
with anlog stuff. And I'd happy to get the signal in an decent quality
into the digital domain, where I'm more comfortable

I put some screenshots of my current circuit and the simulations I did
on my website:
(http://www.dirtybit.ch/temp/piezo/)
One shows the circuit, including operation-point voltages and current
markers.

Thanks for anyone giving me hints/comments/help on this!

Simon

===============
Simon Schneiter
www.dirtybit.ch
===============
Switzerland, CH
===============

 

[Rolavine]

From: [email protected] (Simon Schneiter)

I've designed a circuit using single-rail supply, since the
A/D-converter requires the signal to be between 1V and 4V. I assumed
the peak output voltages of the piezo to be at +10V respectively -10V.

I think your way too high there on your assumption, I've fed Piezo guitar
pickups, both buffered and unbuffered into a mixer board, my guess is between 1
volt and 100 mv max rms output.

I'm a graduate student in electrical engeneering, but not experienced
with anlog stuff.

Wait, how is that possible? And in this case it is just a simple ac coupled
single supply op amp circuit with a gain of about 5, sheesh! What the hell are
they teaching you these days?

Rocky

 

[Simon Schneiter]

I think your way too high there on your assumption, I've fed Piezo guitar
pickups, both buffered and unbuffered into a mixer board, my guess is between 1
volt and 100 mv max rms output.

My thought was to make sure I get enough headroom to avoid clipping.
But I sure will mesure the actual output as soon as I get the piezos
delivered.

Wait, how is that possible? And in this case it is just a simple ac coupled
single supply op amp circuit with a gain of about 5, sheesh! What the hell are
they teaching you these days?

Hmmm, I also feel a little bad about this. I sure learned about the
theoretical side of it. But I never got some hands-on experience.
That's why I'm asking for help from people who have this experience.

But hey! As I mentioned, I'm focused on the digital side (computer
vision, coding and signal processing, speech recognition and
synthesis, VHDL, ...). O.k. the name electrical engineering is
probably just misleading, as I'm using it in a generous way.

Simon

 

[justin]

[email protected] (Rolavine) wrote in message

You don't need gain, you don't need an op amp. To impedance-buffer
piezo pu use a J201 FET with 2k21 drain to ground, 3M3 gate to ground
and 47µ drain to output. At Vss=9V, Id Å500µA, linear headroom is
Å2Vpp. If overdriven It will saturate gracefully but if you are
digitizing the signal for the DSP purposes the TDH is irrelevant
anyway.

j.

 

[Clifford Heath]

My thought was to make sure I get enough headroom to avoid clipping.
But I sure will mesure the actual output as soon as I get the piezos
delivered.

You might *want* clipping. I used piezo wafers as percussion triggers,
and when glued on 1/8" plywood under a rubber pad, a medium whack
with the handle of a screwdriver yielded pulses of several hundred
volts (with no load of course, just a high-impedance oscilloscope
probe). Make sure you have sufficiently low input impedance!

Clifford Heath.

 

[Pooh Bear]

¡Hello Everybody!

I'm working on a design using a piezo guitar-pickup that should be
conditioned for a A/D-converter (an then to a DSP and then ... but
thats another (long) story = ). The whole thing should be small and
not use too much power.

Ok, for starters your circuit will be noisy as hell since you're using an inverting
configuration and you have to add the thermal noise of the 10M input R to any analysis.

You should use a non-inverting voltage follower. I doubt that a typical piezo output will
exceed 3v pk-pk. If it does - attenuate the signal after the follower.

Do you have to run it off a single supply of 5V ?

The PCM1802 has a single ended input. You can get far better results using differential
input converters.

Lol ! It's internally differential anyway !

If you're worried about 'overdriving' the input - I would suggest adding some clamp diodes
to gnd and 5V on the analogue input ( pins 1 and 2 ).

The best way to be sure of 'safe' input levels is to measure the pk-pk maximum signal swing
available from the op-amp stage and scale it actually.


Graham

 

[Pooh Bear]

I think your way too high there on your assumption, I've fed Piezo guitar
pickups, both buffered and unbuffered into a mixer board, my guess is between 1
volt and 100 mv max rms output.


Wait, how is that possible? And in this case it is just a simple ac coupled
single supply op amp circuit with a gain of about 5, sheesh! What the hell are
they teaching you these days?

It's a mystery to me too.

Should mean I'll be handy for anyone needing competent analogue design. It just
doesn't seem to be taught any more. And certainly not well.


Graham

 

[Tony]

I think your way too high there on your assumption, I've fed Piezo guitar
pickups, both buffered and unbuffered into a mixer board, my guess is between 1
volt and 100 mv max rms output.

Most would be in that range. To a large extent, the mechanics
determine the output from a piezo, but open circuit piezos can
occasionally reach as much as 10Vp-p. As a result, many designers opt
for 18V or even 36V batteries. But all that's really needed is a
regular hi-Z buffer with low voltage supply; because the piezo is a
capacitive source, one simply shunts it with a larger capacitor to get
the voltage swings under control (and at the same time, make it much
more tolerant of lower input impedances). This is generally a far
better solution all 'round than trying to handle very high voltages at
very high impedances. My new piezo buffer runs fine on 3 AA cells for
decent battery life (along with a DSP to do the "Variax" thing).

Tony
Tony (remove the "_" to reply by email)

 

[justin]

I would definitely NOT apply clamping or amplification in this case.

These pickups work as differential pressure sensors, voltage output
is proportional to a velocity difference between a vibrating string and
a vibrating top. These excursions are minimal. You would have to whack
the strings really hard to get more than 2Vpp into a ³1M.
I'm assuming we are using a bridge piezo pu which is the only 'correct'
way of utilizing piezos for further signal processing.

In practice, instrument piezos are buffered by a JFET voltage follower
which, among other positive features _saturate_ rather than clip when
overdriven. Hard (diode) clipping at any stage is a big no-no as it
causes realy nasty transients which will fool DSP into 'hearing' what's
not there.

Further, you should 'phantom' power the buffer and place it as close to
the pu as possible and run low Z signal to the DSP.

Maybe you should take a look here:

http://www.rason.org/Projects/jfetamp/jfetamp.htm

http://www.till.com/articles/GuitarPreamp/index.html

http://www.geofex.com/Article_Folders/foolwfets/foolwfets.htm

http://www.till.com/articles/PreampCable/



j.