Question on capturing microsecond pulse signal

[joeazphx]

Hi all, I work in a mechanical lab and got some issues on the electrical part.

We acquired some piezoelectricresistive pressure gauges. They are basically polarized polymer with sputtered metal leads. When compressed, the polymer acts like a capacitor and creates a charge. A current then goes through the leads if in a loop. The charge is expected to be proportional to the pressure.

The gauges come with BNC adapter. I connected it to a resistor (1 ohm) and measured the voltage on it with a oscilloscope Hantek 7504U. The compression is about 20 GPa (2.9x10^6 PSI) for 1 microsecond. According to the specs of the gauge, that's 6 uQ of charge, translated to 6 volt for 1 us on the scope.
The signal is supposed to be a semi-square pulse with kind of a plateau top. But I got highly fluctuating signal (attached picture).

The gauges were developed in early 2000s. Documents from that time showed clean square signal for fast compression. The technique was sold a few times and the current manufacturer only knows how to make the gauges but not how to set them up. Assuming the gauges are still good, I feel the way I connected the resistor could be terribly wrong (attached picture) for fast signal. Do I need active probe for this measurement?

Any suggestions are appreciated!

[Benta]

First suspicion: that whopping big wire-wound power resistor works more like an inductor than a resistor with such narrow pulses.
You need a film resistor or something like it that's purely resistive up to 20 MHz or more.

[Davecbyp]

Thats a railroad braking resistor, 250 watts, why in the universe is a 250 w resistor used with a uC charge?

And one-ohm sounds suspicious, Id expect more like K ohms. A 1206 SMT resistor might be more appropos.

One ohms basically a short circuit. Is there a rated load R?

The scope trace is an oscillation, measure the frequency and find whats inductive. There must be either inductance, or a mechanical analogy, for oscillation. Do the number of cycles coincide with pw?

At second look, it appears the scope is trashing some data, or theres a second signal. This is an app for an analog scope, except for DSO capture...

And PS, that waveform looks like what happens with a VNA that gets a way out of range signal. I dont trust data boxes... I saw similar when testing RF antennas.

Id rather see a 16 bit uC with a faster ADC input for capture, all youre looking for is magnitude, simple data, a uC can capture to memory. Maybe a Pic stick.

So, what kind of explosives are you testing? I want some!

[Uunoctium]

My first answer was in the opposite direction - signal conditioning by LP-filtering to get an stable display. More static, than fast. Therfore deleted.

If doin math right, 20GPa equals 200.000bar...that's
 a lot. Where to find? In heart of an fusion chamber?


One very intresting paper of analyzing fast pressure pulse from blast:

Sanchez, K.; Achour, B.; Coustou, A.; Lecestre, A.; Charlot, S.; Lavayssière, M.; Lefrançois, A.; Aubert, H.; Pons, P. :"Transient Response of Miniature Piezoresistive Pressure Sensor Dedicated to Blast Wave Monitoring". Sensors 2022, 22, 9571.
https://doi.org/10.3390/s22249571

(yep, they also struggled with your observerd oscillations. 20.6MHz)

[voltsandjolts]

We acquired some piezoresistive pressure gauges.

No, what you describe is a piezoelectric pressure sensor, not piezoresistive. Using the right terminology might get you better web search results.

Kistler are well known for such sensors, used in combustion engine chamber pressure measurments etc. They have reading material that might help you, and charge amplifiers etc. (which is what you need when instead of that resistor you should throw away).

2 million psi for a microsecond frankly sounds like nonsense. You're not a free energy nut job are you? If not, please explain what you want to measure, and folks here will likely help with amplifier suggestions.

[TimFox]

Piezoelectric devices are normally modeled as a voltage source in series with a substantial capacitance.
That capacitance is what you measure with a normal capacitance meter connected to the device without external stress.
Therefore, the voltage you sense across the termination (resistor, amplifier input, etc.) includes that important capacitance, which can resonate easily with inductive terminations.

[Benta]

(which is what you need when instead of that resistor you should throw away).

No way! Those resistors are damn expensive and may be top-notch in a different project. EG, as a load that can take a lot of bashing.

[joeazphx]

Great point, Thanks!

[joeazphx]

Thats a railroad braking resistor, 250 watts, why in the universe is a 250 w resistor used with a uC charge?

And one-ohm sounds suspicious, Id expect more like K ohms. A 1206 SMT resistor might be more appropos.

One ohms basically a short circuit. Is there a rated load R?

The scope trace is an oscillation, measure the frequency and find whats inductive. There must be either inductance, or a mechanical analogy, for oscillation. Do the number of cycles coincide with pw?

At second look, it appears the scope is trashing some data, or theres a second signal. This is an app for an analog scope, except for DSO capture...

And PS, that waveform looks like what happens with a VNA that gets a way out of range signal. I dont trust data boxes... I saw similar when testing RF antennas.

Id rather see a 16 bit uC with a faster ADC input for capture, all youre looking for is magnitude, simple data, a uC can capture to memory. Maybe a Pic stick.

So, what kind of explosives are you testing? I want some!

I did not realize there is a disadvantage for the large resistor until you guys pointed out. It totally makes sense now.

1 ohm was recommended by the developer, I guess to reduce the RC constant and improve responsiveness.

The setup is similar to the paper that Uunoctium mentioned.

[joeazphx]

My first answer was in the opposite direction - signal conditioning by LP-filtering to get an stable display. More static, than fast. Therfore deleted.

If doin math right, 20GPa equals 200.000bar...that's
 a lot. Where to find? In heart of an fusion chamber?


One very intresting paper of analyzing fast pressure pulse from blast:

Sanchez, K.; Achour, B.; Coustou, A.; Lecestre, A.; Charlot, S.; Lavayssière, M.; Lefrançois, A.; Aubert, H.; Pons, P. :"Transient Response of Miniature Piezoresistive Pressure Sensor Dedicated to Blast Wave Monitoring". Sensors 2022, 22, 9571.
https://doi.org/10.3390/s22249571

(yep, they also struggled with your observerd oscillations. 20.6MHz)

Yea the setup is similar to what's in that paper. I did some search and reached out to a few authors. Did not hear back much.

[joeazphx]

We acquired some piezoresistive pressure gauges.

No, what you describe is a piezoelectric pressure sensor, not piezoresistive. Using the right terminology might get you better web search results.

Kistler are well known for such sensors, used in combustion engine chamber pressure measurments etc. They have reading material that might help you, and charge amplifiers etc. (which is what you need when instead of that resistor you should throw away).

2 million psi for a microsecond frankly sounds like nonsense. You're not a free energy nut job are you? If not, please explain what you want to measure, and folks here will likely help with amplifier suggestions.

Great suggestion, thanks. I was looking at charge amplifier. Indeed it's piezoelectric. Sorry for the typo. I had both and one of them jumped out of mind randomly.

The setup is similar to the paper that Uunoctium mentioned. Actually this very much is a gentle moderate condition for testing, compared to what we routinely do. We can go an order of magnitude higher.

[jheissjr]

Wire wound resistors have inductance. The capacitance of the sensor with the inductance of the resistor could be a factor. What is generating the 20 GPa? Do you have a picture of the sensor?

[joeazphx]

Piezoelectric devices are normally modeled as a voltage source in series with a substantial capacitance.
That capacitance is what you measure with a normal capacitance meter connected to the device without external stress.
Therefore, the voltage you sense across the termination (resistor, amplifier input, etc.) includes that important capacitance, which can resonate easily with inductive terminations.

Thanks! That is something I need to understand better.

Like e.g. this type of sensors, I know their capacitance. Assuming I can use a non-inductive terminator (film resistor?), I do not need that capacitance to calculate the charge, right?
https://piezopvdf.com/shock-gauges-sensor-Bauer-pvdf/

[voltsandjolts]

2 million psi for a microsecond frankly sounds like nonsense.

The setup is similar to the paper that Uunoctium mentioned. Actually this very much is a gentle moderate condition for testing, compared to what we routinely do. We can go an order of magnitude higher.

Hmm, OK. Please excuse me for saying but while the mechanical engineering you must have for such a blast wave setup is impressive, it's a stark contrast to the electronic engineering as shown. I suggest you contract an EE for a while.

[Marco]

Regardless of the noise, it can clearly drive a 1 Ohm resistor to a significant voltage, so a charge amplifier is complete overkill (and in the application perhaps needlessly fragile).

If it needed 10k to get to a volt maybe.

[Obsidianxenon]

(which is what you need when instead of that resistor you should throw away).

Why on Earth would you throw that perfectly good 250W resistor away?

[voltsandjolts]

No way! Those resistors are damn expensive...

Why on Earth would you throw that perfectly good 250W resistor away?

Guys, of course I didn't mean literally dispose of it, I meant throw it away figuratively, from your design workbench and from your design thoughts. And make a charge amplifier.

Regardless of the noise, it can clearly drive a 1 Ohm resistor to a significant voltage, so a charge amplifier is complete overkill (and in the application perhaps needlessly fragile). If it needed 10k to get to a volt maybe.

How much of that voltage is ringing due to the resistor inductance? Charge amp has a number of benefits, including negating the cable capacitance to due the virtual earth input. If you actually care about accuracy and reproducable results, use a charge amp, with sufficient bandwidth. If you want to play around, use a resistor.

[Marco]

including negating the cable capacitance to due the virtual earth input.

With 1 Ohm load the RC time from cable capacitance is likely measured in picoseconds.

If they do use a charge amplifier, it's going to need a chunky opamp to keep up, also sprinkle some TVS around. This is not small signal, it's energetic.

[voltsandjolts]

It's not the time constant I was thinking about, it was the reduced signal amplitude due to sensor output charge being used to charge cable capaciitance. But the setup shown is so bad, who knows what else is wrong. I mean, maybe they're using an electrical trigger to detonate an explosive charge which creates 2 million psi. Presumably the oscilloscope was close to the test setup, so maybe the electrical trigger pulse emf is actually what the scope picked up, and was mistaken for the sensor signal. Maybe the vibration from the detonation caused microphonics in the ceramic caps of the scope front end, and thats what the scope screenshot shows. I'm done guessing, if they value the accuracy of the measurements they're attempting, they need someone EE qualified on site, end of story.

[Marco]

It's not the time constant I was thinking about, it was the reduced signal amplitude due to sensor output charge being used to charge cable capaciitance.

The time constant is just a sanity check to see if that can even matter, in this case it can't. Even if the source capacitance is large enough to matter, the cable capacitance still doesn't. With 1 Ohm load there will be no significant capacitive divider effect ... you can just ignore it.

With 1 Ohm the source capacitance likely doesn't matter either.

[voltsandjolts]

The scope screenshot is at 500ns timebase.
The resistor is not the type which has non-inductive wiring placement, it has inductance.
At this time scale, the value of the resistor impedance is likely defined by it's inductance.

Also, you seem sure the scope screenshot shows the output signal of the piezoelectric sensor.
I'm sure it's a scope screenshot.

[Marco]

The scope screenshot is at 500ns timebase.
The resistor is not the type which has non-inductive wiring placement, it has inductance.
At this time scale, the value of the resistor impedance is likely defined by it's inductance.

You can still make a guess about the energy in the signal.

Also, you seem sure the scope screenshot shows the output signal of the piezoelectric sensor.
I'm sure it's a scope screenshot.

I'm also taking into account what they said the datasheet said and the manufacturer recommended.

Measure it with a better resistor and coax on 1x to get rid of some sources of potential noise, sure ... but if 1 Ohm load works, a charge amplifier is a bit silly. A 1 ampere current doesn't need amplification and cares little about most realistic sources of capacitance.

[voltsandjolts]

... but if 1 Ohm load works, a charge amplifier is a bit silly. A 1 ampere current doesn't need amplification and cares little about most realistic sources of capacitance.

Ah, I see, you're still convinced the resistor impedance is a purely resistive 1 Ohm.

[Marco]

I am convinced they can fit a 1 Ohm resistor which is still close enough to 1 Ohm impedance at 1 Mhz ... and if that works as recommended by the manufacturer a charge amplifier is a bit silly.

[moffy]

... but if 1 Ohm load works, a charge amplifier is a bit silly. A 1 ampere current doesn't need amplification and cares little about most realistic sources of capacitance.

Ah, I see, you're still convinced the resistor impedance is a purely resistive 1 Ohm.

A suitable surface mount or combination should be fine, appropriate ones work at RF frequencies.