SpudFiles

btrettel, thanks for the PDF, I'lll take a look at the spice model.

I still think you should use an actual instrumentation amp for the circuit. (A real instrumentation amp, not just an opamp designed for use in an instrumentation amp). The chip is a bit pricey, about $7, but I suspect it'll behave better than anything you can do with a single transistor or FET. One real advantage of a true instrumentation amp is that the designers have done all the work to get the amp to be stable and fairly resistant to noise. If you want I've got a pretty good instrument amp circuit that runs off a single 9V battery. The input impedance is about 2MegOhm.

In your schematic I think you can omit C1. Just use the internal capacitance of the piezo for that cap. C2 then effectively becomes the voltage divider pretty much all by itself. In my experience a typical 0.75"D piezo will put out 30V or so when it is inside a firing combustion chamber. That means you probably need the voltage divider to get the voltage down to what the amp and soundcard can handle. Don't want to saturate the amp or AD converter or toast anything. I think an 0.47uF cap would be about right. That'll give a 10 to 50 voltage reduction (based on 10 to 50 pF capacitance of the piezo), which should be about right.

jimmy101 wrote:btrettel, thanks for the PDF, I'lll take a look at the spice model.

I still think you should use an actual instrumentation amp for the circuit. (A real instrumentation amp, not just an opamp designed for use in an instrumentation amp). The chip is a bit pricey, about $7, but I suspect it'll behave better than anything you can do with a single transistor or FET. One real advantage of a true instrumentation amp is that the designers have done all the work to get the amp to be stable and fairly resistant to noise. If you want I've got a pretty good instrument amp circuit that runs off a single 9V battery. The input impedance is about 2MegOhm.

In your schematic I think you can omit C1. Just use the internal capacitance of the piezo for that cap. C2 then effectively becomes the voltage divider pretty much all by itself. In my experience a typical 0.75"D piezo will put out 30V or so when it is inside a firing combustion chamber. That means you probably need the voltage divider to get the voltage down to what the amp and soundcard can handle. Don't want to saturate the amp or AD converter or toast anything. I think an 0.47uF cap would be about right. That'll give a 10 to 50 voltage reduction (based on 10 to 50 pF capacitance of the piezo), which should be about right.

Save the money. He is not interested in precision value measurements. He will be running uncalibrated. Any high fidelity audio amplifier with sufficient bandwidth will do fine. An amp for a DVM or scope is not required, why pay for it. Very high input impedance to make the RC time constant greater than 10X the event time is more important. 2 meg for the typical series capacitance of a piezo may be too low. I don't have the time now to do the math. I'll leave it to you to figure the 10-50 pf 2 meg time constant.

EDIT

An online calculator is here. This shows the need for a low leakage input amp. A 2 Meg load on a 10 pF series cap has a TC of only 20 uS. 2,000 Meg input impedance starts to get you into usable territory for the frequencies of interest.

http://www.cvs1.uklinux.net/cgi-bin/cal ... _const.cgi

If the capacitance is as high as 0.01 uF an impedance of 100 Megohm will give a 1 second TC. Values of only 2 Megohm make a poor preamp for a piezo for low speed events.

Tech

But you are forgetting that he'll have to use another cap as a voltage divider. If he doesn't then he may well toast the amp. Indeed, his drawing shows the second cap.

Can the FET take a +/- 50V or more voltage swing on it's gate? I don't know much about FETs but a standard transistor would be cooked by that voltage swing on it's base.

I figure it'll take about an 0.47uF cap to drop the voltage to a reasonable level. The 0.47uF cap plus 2Meg of resistance gives an RC of about 1 second, not 20uS. The voltage to the amp would be limited to something in the vicinity of +/- 1V.

Save the money. He is not interested in precision value measurements.

Yes he is interested in precision measurements or he wouldn't even be trying to do this. He might not be interested in calibrated measurements but calibration has nothing to do with precision.

Technician1002 wrote:Good choice. They get past the problem of the crappy mic inputs on laptops. Be aware, they do not have floating (balanced) inputs like pro gear so ground loops and power supply hash often creep into the results

I know what ground loops are but I've never heard of power supply hash and Google wasn't very helpful. Is it just noise? I figure some noise is acceptable as long as its frequency is far higher than the signal as filtering would be trivial.

Technician1002 wrote:The circuit is a typical single ended AC amp. It does pass an AC waveform including the negative half. The output can swing between the supply rails of ground and +. It operates as a class A amplifier.

Makes sense.

I'll do a little more research into this before putting the pieces together as I don't know much about it.

jimmy101 wrote:In your schematic I think you can omit C1. Just use the internal capacitance of the piezo for that cap. C2 then effectively becomes the voltage divider pretty much all by itself. In my experience a typical 0.75"D piezo will put out 30V or so when it is inside a firing combustion chamber. That means you probably need the voltage divider to get the voltage down to what the amp and soundcard can handle. Don't want to saturate the amp or AD converter or toast anything. I think an 0.47uF cap would be about right. That'll give a 10 to 50 voltage reduction (based on 10 to 50 pF capacitance of the piezo), which should be about right.

Interesting idea. I'm not certain it'll work in practice as the internal capacitance in the series model is just an approximation but it might work. I'll probably use one regular capacitor and a variable capacitor to figure out how much attenuation is necessary.

Highly precise measurements are unnecessary. Precision to 1 or 2 psi would be very adequate. Whether or not a simple FET would be adequate for this is not something I wouldn't know. Maybe down the road I might look into an instrumentation amp., but for the moment I'll be doing some tests with the 353s I have that have an input impedance of 10^12 ohms if I remember correctly. The IGFET would be better for this if its input impedance really is essentially infinite.

I do want calibrated measurements too. My intention is to use some of these sensors to see how the pressure in pneumatic and spring guns varies with time in different locations. To calibrate the sensor I plan to buy a cheap pressure gauge with good precision (I think the one in my McMaster-Carr order has a precision of 2% in the mid-range) and see how the response and pressure are related (which should be linear, but it wouldn't be difficult to compensate for non-linearities).

The ultimate goals are to improve my computer simulation of pneumatics guns and have more accurate and precise inputs.

jimmy101 wrote:Tech

But you are forgetting that he'll have to use another cap as a voltage divider. If he doesn't then he may well toast the amp. Indeed, his drawing shows the second cap.

Can the FET take a +/- 50V or more voltage swing on it's gate? I don't know much about FETs but a standard transistor would be cooked by that voltage swing on it's base.

I figure it'll take about an 0.47uF cap to drop the voltage to a reasonable level. The 0.47uF cap plus 2Meg of resistance gives an RC of about 1 second, not 20uS. The voltage to the amp would be limited to something in the vicinity of +/- 1V.

Save the money. He is not interested in precision value measurements.

Yes he is interested in precision measurements or he wouldn't even be trying to do this. He might not be interested in calibrated measurements but calibration has nothing to do with precision.

Good point on precision and calibration. My bad and I knew better.. Good catch on that one

Many IGFET's are very sensitive to the voltage to the gate. Proper design with a IGFET front end feeding a high voltage transistor is needed.Bipolar transistors than can handle over 600 volts can be found. Common places high voltage transistors are use are switch mode power supplies running on line voltage and sweep circuits in CRT displays and televisions.

http://www.angelfire.com/80s/sixmhz/flyback.html
More info on driving flyback transformers in the link with real world transistor tests. With a floating bias supply a IGFET can feed one of these in a class A amplifier and still keep the voltages on the IGFET within spec.

Gate to source can be kept within 10 Volts and Source to Drain can be kept within 30 volts and still drive the gate with a 100V AC signal in a source follower configuration.

Cool, a SixMhz link (some of my fav's, though I don't think he's created any new pages in ages )

Bipolar transistors than can handle over 600 volts can be found.

I believe those are Vec or Veb values. Often things like the Veb over voltage is much less. For a generic NPN transistor it'll get cooked if the base drops more than about 6V below the emitter (Vbeo) IIRC.

In this app, I'm just a bit worried that when you smack the piezo hard enough it'll ring like a bell and the output voltage will swing negative about as far as it swings positive. Or, the polarity of the piezo is not particularly well defined and the amp stage might not like getting a -30V spike when it is designed for +30V spikes.

Another thing I was thinking about last night... If you have enough input resistance to the FET/MOSFET it might start to act as a very sensitive electric field detector. Basically, the gate looks like an FET in an unused stage of a chip. FETs are very sensitive to voltage but require virtually zero current so the FET starts to slam back and forth between "on" and "off" in response to small changes in the local electric field. (Hence the "always tie unused FET inputs to something, don't let them float" in digital designs.) The piezo will act as a capacitor between the gate and ground but I wonder if ~10nF is enough.

There is a really great demonstration circuit for what an untied FET gate can do. It'll easily detect the change in the local electric field caused by lifting your foot off the floor, or turning on a TV 10 feet away, or touching a pen to your hair.
http://www.eskimo.com/~billb/emotor/chargdet.html


The 1Meg resistor on the FET is optional. The "short antenna" can just be the FET's own lead, you don't need anything longer. I would think the FET could be repalced by a MOSFET. Reading the above link, it says to install a 100pF cap from the gate to the ground to significantly reduce the sensitivity. So, perhaps the ~10nF of the piezo is enough.

Now that I think about it, I was looking at very high impedance homebrew amps some years ago. There are a couple homebrew amps for ion chambers (radiation detectors). An ion chamber can detect the radiation of weakly radioactive materials. IIRC, the circuits used a pair of transistors wired up as a Darlington pair amplifier. To reduce the amp's sensitivity to local electric fields the amp is mounted directly on the ion chamber. Might be a good idea in this application to mount the amp on the piezo, several feet of wire might act as a heck of an antenna.

See http://www.techlib.com/science/ion.html ... %20Circuit scroll down to the "Experimenter's Ionization Chamber" section.

jimmy101 wrote:

Bipolar transistors than can handle over 600 volts can be found.

In this app, I'm just a bit worried that when you smack the piezo hard enough it'll ring like a bell and the output voltage will swing negative about as far as it swings positive. Or, the polarity of the piezo is not particularly well defined and the amp stage might not like getting a -30V spike when it is designed for +30V spikes.

Another thing I was thinking about last night... If you have enough input resistance to the FET/MOSFET it might start to act as a very sensitive electric field detector. Basically, the gate looks like an FET in an unused stage of a chip. FETs are very sensitive to voltage but require virtually zero current so the FET starts to slam back and forth between "on" and "off" in response to small changes in the local electric field. (Hence the "always tie unused FET inputs to something, don't let them float" in digital designs.) The piezo will act as a capacitor between the gate and ground but I wonder if ~10nF is enough.

There is a really great demonstration circuit for what an untied FET gate can do. It'll easily detect the change in the local electric field caused by lifting your foot off the floor, or turning on a TV 10 feet away, or touching a pen to your hair.
http://www.eskimo.com/~billb/emotor/chargdet.html

Very good points. As with any high impedance circuit shielding is a requirement. A properly biased Class A amplifier will handle a sine wave without clipping in either direction just fine. + or - 30 volts should not matter. As for reverse polarity on the gate. Remember this is a class A amplifier, unless severely driven. Normal ESD precautions need to be taken including reverse polarity protection and over voltage protection. Bias will need to be provided. It just needs to be very high impedance. I didn't say this was an unshielded breadboard compatible circuit. Encapsulation in metal like a condenser microphone is recommended for this reason.

Tech

Any thoughts on the affect of long leads? Figuring the amp is a foot or so from the piezo element. You could use a shielded cable but that adds capacitance and might not work all that well anyway since the circuit probably doesn't have a true earth ground. Would a twisted pair lead be a better idea than a shielded cable?

In the little piezo amp I made for another project neither of the piezo's leads went to ground, they went to the inverting and non-inverting inputs of an OpAmp (actually an intrumentation amp). So the amp was working on the voltage difference across the piezo instead of the voltage difference of one side of the piezo relative to circuit ground. Noise picked up on one of the piezo's leads would be matched (more or less) by the same signal on the other lead and the two noises would tend to cancel each other out.