LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

Magnetic Variable Reluctance Speed Sensor Measurement

Hello everyone,

 

I'm currently working on measuring the RPM of a TurboCharger. I have placed the Magnetic Speed Sensor (Jaquet ATS-1 DSE 1010) in the TurboCharger and have connected the Sensor to a myRIO system, which is connected to my computer.

 

According to the Data Sheet the Output Signal is an Analog Sine Wave with a Speed Detection Range of 8'000 to 400'000 RPM. For now I can't get to those RPM's, since it's not connected to the AirFlow system. But I would still like to get the LabVIEW started, before putting high pressure air through.

 

When I spin the TurboCharger by hand, I see the change in the output signal. It looks similar to a sine wave signal, but there is still a lot of noise too.

 

Could anyone help me regarding measuring/getting the Frequency, Amplitude and RPM in LabVIEW? Is there any way to process this signal and remove the noise? Appreciate any help! 

 

Best regards,

Download All
0 Kudos
Message 1 of 6
(915 Views)

How about attaching a PDF of the Data Sheet?  What is 8'000 RPM?  Is it 8000 RPM or 8,000 RPM perhaps?  If this was a "magnetic" detection system (I'm not an engineer, so I don't know the proper name for what I'm thinking of, but something that measures the magnetic "strength" of a spinning magnet, maximum when the magnetic "north pole" points north, minimum at south), this would give a sinusoid at 133.3 Hz, with 400,000 RPM being 6667 Hz.  This type of sensor, however, I'd expect to produce a simple (lower-frequency) sinusoid when you spin it by hand -- I'm puzzled by the "single sinusoidal waveform every so often" nature of what you observe.

 

Bob Schor

0 Kudos
Message 2 of 6
(873 Views)

Hello Bob,

 

Thank you very much for your reply!

 

Yes, I meant 8 000RPM to 400 000 RPM. I have attached both Data Sheet's I got. I have also attached a PDF from Honeywell, which kinda explains the concept behind this type of sensor.

 

I'm used working with Pressure Transducers, Thermocouples and a Coriolis Mass-Flow Sensor. These were pretty straight forward, since I just received a Voltage signal and could translate that directly to Temperature, Pressure, Mass-Flow.

 

In my mind, I have to be able to read a Frequency and an Amplitude out of the Raw Signal. Any help is appreciated! 🙂

Download All
0 Kudos
Message 3 of 6
(871 Views)

I think I get it!  This gadget is designed to be placed near a shaft of steel that has a flat machined on its surface.  If there were no flat, then as the shaft spins, the "magnetic reluctance" (the "disturbance in the Force, Luke") caused by the (relatively constant) steel "disturber" would be minimal, i.e. would be a constant, but if you machine in a flat, then as the shaft turns, when the flat is hear the sensor, the reluctance will vary, apparently producing a (single) sinusoid as it passes the sensor.  So this represents a "contact-less" method of getting a "once per revolution" signal.

 

If that's true (and someone out there who is a bonified Engineer, instead of a physiologist, can say Yea or Nay), then all you need to do is make a little circuit to make a TTL Pulse when the sinusoid is significantly larger than the noise (I don't have your data figure in front of me, but I recall you had a decent S/N) and send it to an NI Multifunction module with a Counter/Timer circuit to get the Frequency, number of TTL Pulses per unit time.

 

Bob Schor 

0 Kudos
Message 4 of 6
(839 Views)

You wrote 800 000 RPM  that result in about 13.3 k pulses per second

your slow turning picture show about a 1:16 ratio of pulse repitition to pulse length

13.3k times 16 results in a frequency of ~213 kHz  so you need at least a samplerate of 500kSPS  (and more to capture features of the pulses ....)

As you can see in your screen.png , you didn't catch the full sensor response anymore!

 

For frequency (and RPM) measurement I would recommend a counter with configureable threshold.

 

If your DAQ can sample fast enough, you can use peak detection (maybe in a two step way, first find positive and negative peak of one pulse with a coarse window, by peak distance choose a optimal window length and do a second peak detection)

 

depending on the data you can get, maybe a pulse fitting can improve the results

Greetings from Germany
Henrik

LV since v3.1

“ground” is a convenient fantasy

'˙˙˙˙uıɐƃɐ lɐıp puɐ °06 ǝuoɥd ɹnoʎ uɹnʇ ǝsɐǝld 'ʎɹɐuıƃɐɯı sı pǝlɐıp ǝʌɐɥ noʎ ɹǝqɯnu ǝɥʇ'


0 Kudos
Message 5 of 6
(777 Views)

@FMERAL wrote:

Hello Bob,

 

Thank you very much for your reply!

 

Yes, I meant 8 000RPM to 400 000 RPM. I have attached both Data Sheet's I got. I have also attached a PDF from Honeywell, which kinda explains the concept behind this type of sensor.

In my mind, I have to be able to read a Frequency and an Amplitude out of the Raw Signal. Any help is appreciated! 🙂


At 400,000 RPM (assuming one pulse per revolution) that is 6667 Revolutions per sec (6667 Hz)

However, the signal is appox at a 17.5% duty cycle. (base off of pixels in your graph)

Assuming I have done my math correctly (IE Do the calculations again!) 6667/.175 = 2.6248e-5 = 26.25 usec pulse

As Bob Schor points out, you will want some external HW signal conditioning to boost the signal out of the noise. One possible suggestion would be to use a flip-flop circuit to turn the pulses into a square wave.

 

Hope this help.

 

 

---------------------------------------------
Certified LabVIEW Developer (CLD)
There are two ways to tell somebody thanks: Kudos and Marked Solutions
0 Kudos
Message 6 of 6
(716 Views)