Sign up Calendar Latest Topics Donate
 
 
 


Reply
  Author   Comment  
vogel

Member
Registered:
Posts: 74
Reply with quote  #1 

Hi,

Does anyone know where it is possible to find datasheets or details about the sensors used by ultrasound meter manufacturers?
I've gone through the SDT, UE systems or Sonotec websites and I couldn't find any details (such as working principle, sensitivity, frequency response, signal to noise ratio ..) about their sensors. 

Thanks


Walt Strong

Sr. Member
Registered:
Posts: 709
Reply with quote  #2 
"I couldn't find any details"

Good luck on your quest! I have been working with ultrasound for a long time, and the OEMs you mentioned appear to keep sensor details as proprietary. There in no ISO or similar standard for sensitivity or frequency response or calibration method. The meter you mentioned typically are digital volt meters with dB referenced to 1 micro-volt. dB values will vary with meter and sensors when measuring the same sound! You would be shocked at how inexpensive some of the sensors are from third party suppliers. I am talking about the sensing element (usually piezo-electric, MEMS are now coming along) and not those with a special case/housing, magnet, stinger, or cable.

Why are you asking the question?

Walt
vogel

Member
Registered:
Posts: 74
Reply with quote  #3 
Hi Walt,

thanks for your answer. The reason I'm asking if is simple curiosity. I started working with vibrations before ultrasound, so when you see things like the one you mention "dB values will vary with meter and sensors when measuring the same sound!" you cannot be but shocked .. would anyone expect that from a vibration meter?

So I had some spare time and I wanted to do some research comparing some measurements and trying to put some "science" into my knowledge ... are some sensors better than other for certain applications? Are specific frequency ranges better than others for certain applications? And I would like to do it by understanding the physics, not by reading and believing a selling brochure.

Would you have any reference of those third party supplied sensor? I'd like to try one.
Walt Strong

Sr. Member
Registered:
Posts: 709
Reply with quote  #4 
Many of the common sensors for leak detection, machine fault detection, and integrity seal testing operate near 39-kHz. The Emerson-CSI meter had lower frequency settings, and some go higher. The sensors actually a broad response unless "tuned" mechanically or electronically.

Search: digikey ultrasonic transducer

https://www.bing.com/search?q=digikey+ultrasonic+transducer&form=EDGEAR&qs=SC&cvid=7f3424566d3046a38d6520387796c77f&cc=US&setlang=en-US&elv=AY3%21uAY7tbNNZGZ2yiGNjfNO3ejW1W67lis5*CchzKYjWh5OhAzzcQqXIvvQtdqXEoPAZTqAuTgui3zRt6lzoch76b3n4q*zi6gdOR1z3fMK

https://www.digikey.com/products/en/sensors-transducers/ultrasonic-receivers-transmitters/527

Search: pvf ultrasound sensor

https://www.bing.com/search?q=pvf+ultrasound+sensor&form=WNSGPH&qs=SW&cvid=bec449c735e245629e7657b28772b079&pq=pvf+ultrasound+sensor&cc=US&setlang=en-US&nclid=FB9B320077853E98E92461E6A83309C7&ts=1519329446251&elv=AY3%21uAY7tbNNZGZ2yiGNjfNO3ejW1W67lis5*CchzKYjUOyMidu02u39DZjpQXJ%21rrf*iLGKsxa*w6VEBNlLeNYCLfQOulGEO8Kd6kYgE6Zi

https://www.sparkfun.com/datasheets/Sensors/Flex/MSI-techman.pdf page 34 I have transmitter and receiver kit

Search: pvdf ultrasound sensor

https://www.bing.com/search?q=pvdf+ultrasound+sensor&form=EDGEAR&qs=PF&cvid=50fcae22d33c46248ea820a8683e3ae2&cc=US&setlang=en-US&elv=AY3%21uAY7tbNNZGZ2yiGNjfNn2cmNmFpe%21qOsWBtZn6XYQ8%21jRfmRAes8I8yLceame21btSMsfZUmkTMwgNkOTPcCQwjZE1u3F0vJygQcLFF5

Search: mems ultrasound sensors

https://www.bing.com/search?q=mems+ultrasound+sensors&form=EDGEAR&qs=PF&cvid=40d2f049a1b043b58f3773dd0f2bca32&cc=US&setlang=en-US&elv=AY3%21uAY7tbNNZGZ2yiGNjfNn2cmNmFpe%21qOsWBtZn6XYrJjS28QSNe9mauoh6rDso%21R*1o8zULh%21hjvTww8okAsFq*AhkuCwbDrzH6Luwo5k

This should get you started! Good luck with the calibration issue!

Walt
VibGuy~5

Member
Registered:
Posts: 69
Reply with quote  #5 
I've asked the question many times-looking for frequency response curves, but got nothing.
According to the sales guys, they use a +/- 2 kHz band pass filter that can tune you from 18 kHz up to 100 kHz.
We mostly work at 30 kHz for bearings and 40 kHz for leaks, but lower frequencies for structure borne steam and valves.
As with any equipment and sensor, it's all about trending. Any dB value will depend on their reference voltage for making the dB calculation. They don't tell you what it is and you can't change it.
No point trying to compare it to anybody else, as I guess this reference voltage will be compared to the background electrical noise-everybody will have different values.
We're still getting used to ours-everyday is a school day.
Walt Strong

Sr. Member
Registered:
Posts: 709
Reply with quote  #6 
The frequency bandwidth of many ultrasound sensors is surprisingly much wider than 2 or 4 kHz. The response is not flat, but sensor output can still be useful. 

SDT uses 1 micro-volt reference. The actual voltage measured by the "microvolt meter" depends on the sensor sensitivity which is never provided. The dB value is converted from linear = 20x LOG10 (V/Vref) where Vref = 1.0e-06 volts. Some issues are what happens when sensor is partially damaged, but still appears to function, or when a sensor is replaced. How close is the output, especially for Trending? I would be careful about Trending, since there are a lot of variables that you may or may not have control over. Ask three or more people to make a "bearing housing" contact measurement within a short period of time that minimizes machine operating variations, see what you get! The "art of measurement" is very important whether for audible sound or ultrasound.

SDT made (many years ago) a small airborne sound source that was intended for demonstrations and integrity (leak) testing. I converted it into a shop/field calibrator by making an airborne (microphone) measurement at an on-axis repeatable distance (say 10" to 12") and recording meter value. On the back side of the unit, I made a drill-point indentation into the metal case. This became my reference source/location for the contact sensor stinger. A low reading could also indicate an issue with the sound source!

Have fun on your UT journey,

Walt
VibGuy~5

Member
Registered:
Posts: 69
Reply with quote  #7 
4 kHz according to UE.
I think the Vref should always be the electrical noise floor, then the log of 1 would give you zero.
Pruftechnik used a Vref above this for their shock pulse measurement. The value would bottom out about -28 dB, as the log of a fraction is a negative number. Never made much sense to me and explaining to customers how negative shock pulse was related to the log of a fraction usually meant you lost them straight away.
We only trend ultrasonic readings using a magnet at an agreed location, but even at the same speed, different levels in our tanks can give different suction pressures which give a saw tooth effect on the trend. I'd have more faith in the old VA. Stethoscope allows you to hear what's going on better.
We don't trend air leaks, but you get a feel for how bad a leak is after using it for a while.
Time at the machine-best way to learn anything.
vogel

Member
Registered:
Posts: 74
Reply with quote  #8 
In my opinion, the reference used for the dB calculation is not an issue, as long as you state it. Like in acoustics, if you measure a sound at a certain dB SPL, I can use another instrument and replicate the measurement since I know that the reference is 20uPa. This is the way acceptance testing of machines (gas turbines or blowers ...) that involve sound measurement works ... There must be an agreed reference level.

Quote:
Originally Posted by Walt Strong
The frequency bandwidth of many ultrasound sensors is surprisingly much wider than 2 or 4 kHz. The response is not flat, but sensor output can still be useful. 


So this is basically my problem, how do you know the sensor frequency response without a datasheet (other than testing)?

Then my question comes, if the frequency range of the sensor is wider why is it needed to filter it to such a narrow bandwidth? The more you narrow it, the bigger the risk of leaving your source frequency out of the measurement frequency range. Maybe you try to use use the sensor at its resonant frequency to amplify the signal?

Quote:
Originally Posted by Walt Strong
The actual voltage measured by the "microvolt meter" depends on the sensor sensitivity which is never provided. 


And this is my second problem, how can I use a sensor without knowing its sensitivity? This must be one of the few fields in test & measurement where this happens (if not the only one). What happens when you replace the sensor?




Walt Strong

Sr. Member
Registered:
Posts: 709
Reply with quote  #9 
"Then my question comes, if the frequency range of the sensor is wider why is it needed to filter it to such a narrow bandwidth?"
My guess is from the Old Days of analog circuits that were used for frequency demodulation, and it provides an adequate signal for listening.

"Maybe you try to use the sensor at its resonant frequency to amplify the signal?"
This is common practice.


"And this is my second problem, how can I use a sensor without knowing its sensitivity?"
I guess by blind faith, but you need your own verification, as I indicated.

"This must be one of the few fields in test & measurement where this happens (if not the only one)."
I have been in the field of sound and vibrations for 50-years, and I was shocked many years ago with the same issues you have. I have not seen any progress either!

"What happens when you replace the sensor?"
You better have your own way of checking it, as I said.

Let me know if/when an ISO Standard comes out for this stuff!

Walt
VibGuy~5

Member
Registered:
Posts: 69
Reply with quote  #10 
The 20-100 kHz frequency range and the +/- 2 kHz bandwidth came to me by word of mouth-no paperwork to back this up.
If you're looking for a technology that will stand up to to an audit or serious review, then I'd advise looking elsewhere or coming back in 5 years. In my limited experience with ultrasonics, the big players are only entering the world of repeated readings, spectrums and time waveforms. I've asked for sensitivity details and Frequency response curves, but got nothing.
The reason for the +/- 2 kHz bandwidth is that it's primarily a listening technology and this works quite well-40 khZ for airborne leaks, 30 kHz for structureborne bearings, etc. If you were listening at all frequencies at once, you wouldn't make much sense out of it. Yes, you could be missing something, but then you can change frequency while you're listening if you want to.
'there must be an agreed reference level' -not so. It's their toy and they can do what they want with it. Put an SDT system against a UE system under the exact same conditions and I doubt they would be within 10% of each other. Not that you can compare under lab conditions anyway, as you don't have sensitivity details or reference levels in the first place.
Walt Strong

Sr. Member
Registered:
Posts: 709
Reply with quote  #11 
I agree with VibGuy comments. The current ultrasound products are not very good if you try to treat them the same way as vibration; such as for trending and certified annual calibrations. Ultrasound can be a great tool to supplement vibration monitoring programs. It is an excellent tool for detecting poor lubrication and early fault development including looseness, shaft coupling misalignment, loose parts, and poor lubrication, belt drive faults including low tension, belt damage, and sheave misalignment electrical faults including arcing and corona discharge, air and gas leaks. I have made all these measurements including using ultrasound for integrity testing of ship cargo holds, acoustical enclosure, and refrigerator (demonstration). Not all ultrasound meters and sensors are created equal, so try it before you buy. Just don't make a big deal out of trending voltage decibels without proper interpretation of what is being measured. Make sure you have your own method for making calibration/operational checks to be certain the meter and sensors are in good working order for repeatable measurements.

I have conducted several ultrasound training sessions, but here is a good independent resource:

http://ultra-soundtech.com/author/jim-hall/

https://theultrasoundinstitute.com/

Can you hear me now??


Walt
vogel

Member
Registered:
Posts: 74
Reply with quote  #12 
Thanks a lot to both of you for your advise. It's really helpful.

Honestly I'm still full of doubts ... like ok, let's assume that 30 kHz is a good frequency to detect structureborne/lubrication bearing faults, but how can I know that my sensor is able to measure at that frequency if I have no datasheet? How do I know that it isn't better to measure at 25 kHz or 35 kHz because my sensor response is better there?

I've been to some SDT training and I have used a bit these devices in the field, but I guess that I still have to do a kind of a deep dive into this technology ... I'll come back with feedback and doubts as I go through this learning process.
Walt Strong

Sr. Member
Registered:
Posts: 709
Reply with quote  #13 
"ok, let's assume that 30 kHz is a good frequency to detect structureborne/lubrication bearing faults,"

Okay, who told you this was the "best"? Consider the source of information!

"how can I know that my sensor is able to measure at that frequency if I have no datasheet? How do I know that it isn't better to measure at 25 kHz or 35 kHz because my sensor response is better there?"

If the meter you have has multiple choices for frequency setting, then you have two choices: follow the OEM meter settings or try one or more settings on a substantial number of machines and bearings to evaluate what is best for you. 

I have an old SDT 150 meter with several factory and homemade sensors that I still use. It is a single frequency analog circuit with digital dB meter. It detects everything that I previously mentioned. Measurement skill and interpretation are the essential difference between success and failure.

Walt
VibGuy~5

Member
Registered:
Posts: 69
Reply with quote  #14 
Don't take anyone's word for it. Best thing to do is compare for yourself and become a scientist/trailblaser.
I presume you have VA equipment? Find some bearings which you know from experience have lubrication issues and others with bearing defects.
Then listen with your ultrasonic equipment and vary the frequency. If your equipment allows, take a time-waveform and spectrum. Compare and contrast.
You may find that you'll end up teaching the ultrasonic providers a thing or 2 about their equipment.
Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.