I have a piezo sensor that is putting out 1.1mVAC when resting when a Fluke is connected set on AC Volt. It is quite "steady" at 1.1mVAC, and at times goes to 1.1mVAC and to 1.2mVAC.
Above 1.4mVAC reading there is definitely an external trigger going on, and operational readings are expected to take place at the 2-10 mVAC level. Putting aside improving the sensitivity of the sensor itself, what would be the best approach to amplify it for an ADC-read at a minimum 1V scale but preferably a 3V scale of an MCU such as the Nano. Need to measure the signal amplitude (not just exceeding the threshold). So need to amplify it by min. 1000 times at the 1mVAC reading and 300 times at the 10mVAC reading..
Simple OP AMP? Instrumental OP AMP? Any pre-conditioning before the OP AMP?
It's "simple" to make an amplifier with an op-amp but low-level signals and high-gain amplifiers are highly susceptible to noise pick-up. (Any noise gets amplified along with the signal).
Peizos are high impedance and that makes noise sensitivity worse...
If the vibration/activation is strong enough you can get more voltage out of a piezo... maybe several volts.
You may need shielding, and if there is a long wire between the sensor and the amplifier you WILL need shielded cable.
Your multimeter has filtering/averaging/smoothing so you may not be seeing the noise. If the voltage isn't fast-changing you can use software smoothing and/or hardware filtering (a low-pass filter) to reduce the noise/instability.
Also if the sensor is picking-up an impulse, the meter won't catch the peak. If it's sensing a continuous vibration the meter reading should be pretty accurate.
You might need a separate power supply or a separate voltage regulator to clean-up the power supply noise, including any noise that gets-into the power supply from the digital circuitry. (It's not unusual to double-regulate the power supplies in microphone preamps and phono preamps which also work at low-millivolt levels.)
Also if you need accurate output near zero the op-amp should have positive & negative supplies.
I agree, you are measuring noise pickup. Shield the wires, and place the whole test on top of a grounded shield, I guarantee you will see a huge reduction in the voltage.
To really understand the noise, you need to view it with an oscilloscope.
You mentioned all the additional issues that I discovered since I posted this. Quite remarkable!
There is nothing connected to the sensor, only the Fluke...set to VAC. So this is the natural freq/volts of the sensor itself in its current environment. The "sensor" is just a metal plate with a piezo material layer. The Fluke neg ground is connected to the metal and the + of the Fluke to the piezo layer itself.
The sensor is quite stable on the Fluke, but a total mess on the OScope. As it responds to air flow hitting it, it's quite linear on the Fluke but not on the Oscope. I just hooked up the Oscope and getting about a 25 khz average with huge random jumps. And a Vpp of about 50mVAC that seems centered at around 1mVAC. But at times the freq goes to 100khz or below 25khz. At times the mVAC jumps to 300mVAC.
I would be happy with a FLuke-type circuit that would amplify the 1mVAC.
That would be a job for an op amp. Piezo sensors don't put out any voltage at all unless they are vibrated mechanically. So you are either measuring some acoustic noise, or some electrical noise in the environment.
Amplifying the signal will also amplify the noise, so it won't help you separate the signal from the noise.
If you can't get rid of the noise, you can low pass filter the signal. But you have to be confident that the resulting bandwidth is sufficient for your needs. Also it won't filter out any low frequency noise. So if the noise is broadband, it will still mix with your signal, in the chosen pass band.
Thank you. Since I hope to measure air flow, there is not much I can do to the sensor itself for noise. Except shielding.
In this case looks like mostly I need to do bandwidth in the expected reading range as part of the op amp circuit and then signal processing on the MCU for averaging and throwing out spikes.
Any op amp (pre-conditioning) circuit suggestions that does not "kill" the 1mVAC signal? Should I make it DC with a capacitor inline?
The "sensor" is just a metal plate with a piezo material layer. The Fluke neg ground is connected to the metal and the + of the Fluke to the piezo layer itself.
Oh, a bare piezo disc. But how is it installed? How does it measure flow? The second question you can ignore if you like, I'm just curious. But the installation is still a mystery. If you are testing it just lying on your desk like that, see my previous comments about that.
What is the time scale on your scope photo? How are the probes connected? Are you using 10x or 1x on the probe?
Can you show a photo of the entire test setup as it is, when you take measurements?
Have you tried doing it in a different location? Some locations have a lot of electrical noise from LED lights, voltage converters, all kinds of electronics...
You shouldn't have to "shield" the disc, but you definitely should ground it.
I see no leads on the disc. How are you connecting to it?
Alligator clips, with positive alligator clip isolated from metal plate. Did not try different locations...this is representative of the environment . I turn on and off a 2 speed table-top fan.
It's facing the air flow with only the Fluke leads connected. I just encased it in aluminum foil except the front, but resulted in no difference on the scope readings. Using 10x probe and 25us window. Oscope + connected to piezo layer (Fluke +) and Oscpe neg to metal plate (Fluke -)
Okay that is a photo, but it really doesn't show the connections. Remove the foil, it's probably not helping anyway. Please show the entire test setup, wires, probes, scope and all.
I asked about the installation because it's fairly obvious that you can not use the bare disc as it is. So I would like to know your plans, as it does affect the answer to your original question.
I turned off all the lights. No noticeable change to the oscope. When I took the + lead off the piezo, the Vpp jumped about 3 times from about 16mVAC to 54mVAC.
Setting the oscope period to 25 ms (do not know how to set it to 20ms) and repeating the same as above, did the same 3 times jump but at different VAC scales.
Will try to make it as short as possible. How would you go about amplifying the 1mVAC signal to about 1V with a regular op amp? Even if with the noise. Inverting op amp with resistor from sensor in series and a capacitor to make it DC? Looking at a range of +8mV and -8mV sine(?) wave, that is centered at around +1mVAC.
No, for a high impedance source you must almost always use the non-inverting configuration. About 1M input load resistor should be about right. It should be a JFET or similar op amp with high impedance inputs.
