ICP Piezo Force Sensor Interface to microcontroller

Hello Greg,

You are correct, there are many ways of solving this problem, but by fixing some parameters, you may be able to narrow down your options. I have a couple of questions:

1. What determines the offset of the sensor output? Is it ratiometric? (i.e. dependent on supply voltage)

2. What resolution/dynamic range are you trying to achieve? From your question I understand that the full-range of your signal (at the output of the sensor) is 5V. If you just attenuate the signal and offset to fit it into your ADC, your dynamic range will be reduced by the attenuation factor.

3. Are you comfortable sharing which sensor you are trying to use? Sensor manufacturers, even when they say "low impedance" mean different things.

Regards,

Gustavo

Gustavo:

In answers to your questions:

1) The offset is determined by the sensor electronics. The sensor itself is a piezo charge output sensor with a built in mosfet amplifier. The amplifier converts the charge output of the piezo sensor to a low impedance voltage output. The impedance is listed as less than 100 ohms. The supply voltage to the sensor 18-24 Volts constant current to maintain a linear response. I am not sure if the signal is radiometric but regardless supply voltage needs to be 24Volts which will result in an offset of 10Volts and a signal range of 0 to 5 volts.

2) This is a hard question to answer as the range could change depending on the sensor chosen and test being run. As a simple answer, resolution of 0.1mv out of a 5V signal would be great. However, in reality this resolution would only be needed for a smaller range say to 1V and then switch to 1mv for full scale readings. The ideal system would incorporate digital control to adjust range and resolution as needed.

3) We have not chosen a specfic sensor but ICP is an acronym for "integrated circuit piezoelectric", and is a registered trademark of PCB Piezotronics. This is also similar to IEPE which is the standard for piezoelectric transducers and is also an acronym for "Integrated electronics piezo electric".

Thanks for your help

Greg

Hello Greg,

The problem I see with using the AD8276 directly to take your sensor measurement is that its output impedance is low compared to the output impedance of the sensor. You may be better off with an instrumentation amplifier, such as AD8226. Since both diff amps and inamps will subtract the offset by looking at the difference between their inputs, you would need an offset voltage that ideally is the output voltage from the sensor without mechanical load. This is a difficult problem. At some point you have to decide what is offset and what is a valid signal. How much this offset drifts and whether or not you can take a baseline without taking a hit on dynamic range will further complicate the problem. Furthermore, these sensors seem to have a discharge time constant that will come into the equation.

Take a look at the circuit idea I'm attaching to this post. The divider formed by R1 and Radj should give you the sensor offset without load. There are different ways of adjusting this based on your design requirements. It could be through a DAC, digipot, servo loop, etc. AD8276 is used to drive the reference pin of the AD8226 to mid reference. Further down the signal chain, you'll need some overdrive protection since the instrumentation amplifier is running off 24V (you can power AD8276 from 5V). Then, the PGA inside the AD7192 can take care of adjusting the range for you. I'm not including details on the source Iex and the reference circuit. Also, please note that you may need an RFI filter at the input of the AD8226 amplifier. For examples on how to do this, please take a look at page 22 on the AD8226's datasheet which can be found here.

Another alternative: You can use a dual version of AD8226, the AD8426. The second instrumentation amplifier can be used to create a fully differential drive for the ADC, as shown on page 25 of the data sheet. This circuit structure will give you better dynamic range and a single-chip solution, but it would require protecting both inputs of the ADC and it will consume a little more power. It would also have a little less bandwidth.

Let me know if you have any further questions.

Regards,

Gustavo

Gustavo:

Thank you for your detailed answer. I have a couple of questions about your schematic. In your comments you say that AD8276 is used to drive the reference pin of the AD8226 to mid reference. What value are you considering mid-reference? To me it means the mid value between what would be adjusted with the Radj resistor. For example, if I adjust the reference from 10-10.2 volts then mid-reference would be 10.1 volts. Is this correct? Then the value of Vref would be 10.1? volts?

I am going to incorporate this circuit into my design. Do you have any pointers for the Vref voltage source? I did a quick scan of your web site and there are many 10V reference sources but all of them seem to be fixed at 10.0 volts. Sorry if this is a basic question but I am much more comfortable with the digital side of circuits than the analog.

Best regards,

greg

Greg,

I am considering mid-reference to be the middle of the ADC's reference. In other words, if you supply your ADC with 5V reference, then it would be 2.5V. You will want to drive the reference pin this way because your signal will be bipolar, and this way you can acquire it with the ADC. However, and like I said before,  you may benefit from getting twice as much signal range if you implement the differential drive. I had a little more time to sketch it, so I am attaching it to this post. While the previous idea would allow you to acquire signals from from your sensor in the range slightly less than 7.5V to 12.5V (assuming 10V offset) the differential drive would allow you to go from a little more than 5 to a little less than 15V.

Radj is a different story. You want to adjust the voltage going into the input of the first instrumentation amplifier to set your "zero" coming from the sensor. The voltage resulting from this divider should be around 10V, as you've correctly pointed out.

Vref can be 5V. There's a number of options. I'm not the apps engineer for references, but I know that if you wanted to run directly off 24V, you can use something like AD586. However, 24V input for a 5V reference is a lot of power and you may want to stay away from that. That means that you should drop that voltage with a regulator or more simply with a zener. Then you can feed it into a reference like ADR435. I believe you can also buffer the reference voltage with an amplifier with enough current drive to power the ADC. An example on how to do this can be found on page 16 of the following data sheet (AD7690).

Regards,

Gustavo

Gustavo:

I am getting a little confused on the output of the sensor. My understanding is that the signal ranges from 10 volts to 15 volts depending on the load on the sensor. As it is a piezo, the charge increases with a change in load and then drops back to a baseline value. In my mind it is like half a sine wave. In AC coupled mode, a capacitor is used to remove the DC component and the values can be read from using an ADC. I cannot use the AC coupled mode due to the time constant of the sensor in AC coupled mode. I do not think the signal is bipolar centered around the offset. I thought the AC part of the signal was simply the fast response time of the sensor. However, I could be way off on this. The sensor is a standard piezo with an charge amplifier built in.

The mid-reference makes sense to me with regard to a bi-polar signal. I can also take care of the voltage reference as per your circuit. I was thinking it needed to be adjustable with regard to the sensor zero. I will have a stepdown voltage available so the reference will not be powered from the 24V source.

Greg

Greg,

My understanding from glancing over the ICP sensor documentation is that even if tension is never applied to the sensor, there is a discharge time constant. In other words, if you think of a constant force pulse that will raise the voltage from 10V to 15V, there will be an exponential voltage drop. If the pulse is long enough to drop to, say 14V, then when the force is removed, the output would drop to 9V and eventually go back to 10V.

In other words, even though you're dc coupled to the sensor, there will be an ac coupled like behavior. However, if your application won't see a long pulse (and therefore the recovery voltage will be small), then you can modify the first schematic to drive the reference to a smaller voltage (for instance, less than 1V) with a resistor divider and an amplifier like the OP1177. This is shown on page 20 of the AD8226 data sheet.  The nice thing about the fully differential output is that you get the negative swing without giving up your full positive swing.

If you have less than 18V available, then you should be able to use ADR435 for your system reference. I am attaching a variation to the last circuit to generate a more stable offset null for the sensor output (assuming Vref=5). R1 will allow the circuit to adjust below 10V and Radj will move the voltage higher. R1 probably can be something like 1k and Radj should be at least more than twice R1.

Regards,

Gustavo

PS> I have attached a second schematic showing an example on how to implement a programmable offset null. Note that if you prefer the original implementation using a potentiometer (for manual adjustment), it would be a good idea to add a series resistor to limit the adjustment range. Good luck!

Message was edited by: Gustavo.Castro. Added programmable offset null

Gustavo:

I'm with you. I am going to use the full differential input as per your last schematic. It looks like it will have the most flexibility. I will also check out the ADR435 for the reference. I will get back to you when we get a prototype together to let you know how it worked.

Thank you for your help with this application. Your input is greatly appreciated. I always use ADI components when I can because I have gotten such great support from your company.

Best regards,

Greg

Gustavo:

Quick question on your last circuit diagram. Is the AD8274 powered from 5 Volts or 24 Volts?

Thanks,

Greg

Hello Greg,

It is powered from the same supply as the inamp, or 24V according to the schematic. Actually I am glad you asked, since it just made me realize I had a small mistake in my schematic. To get the correct gain from the AD8274, you have to connect the 12k resistor to ground. I'm sorry for the confussion.

Regards,

Gustavo