Sign up Calendar Latest Topics Donate
 
 
 


Reply
  Author   Comment  
Curran919

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 599
Reply with quote  #1 
This is a continuation of this thread earlier this year.

I've been playing with codewheels, zebra tape and tachometers. I've got 5 different brands of infrared phototachs in my lab, and none of them work as I think they should. Only 2 actually give me a square wave on a zebra strip. The rest all need retroreflective tape.

The ones that need reflective tape have no visible laser and still need to be quite close (<3") to the target to trigger, which I would expect to be larger if they were indeed a laser tach with an invisible beam.

Of the 2 that work with diffuse zebra tape (Jaquet FRA 119 and Moviport D711), I'm getting strange results. The pulse length is discretized in time, such that edges are only detected at multiples of 1/3600th of a second. I have encoders that do not have this problem and seem to give a continuous signal, but these tachos seem to be interrogating the surface with pulses every 3.6kHz. The only piece of information in either manual that substantiates this is the max shaft speed of 99,999rpm. With a 1PPR target (assuming 50% duty), that would equate to a required sample rate of 3.3kHz. Seems close...?

Is this a common behaviour? Both of these sensors are also 35+ years old, so maybe this is an outdated tech? After all, although the sensors are intended to read from several pulse-per-rev wheels, they are not intended to give more than the average speed signal, which is unadulterated as long as there are less than 3600 pulse-per-second. However, the torsional vibration is completely destroyed. I could get a new IR sensor, but I'm not convinced that they wont have the same problem since it does not appear to be a malfunction.
RustyCas

Avatar / Picture

Admin
Registered:
Posts: 1,810
Reply with quote  #2 
First let me say that I understand little of what you said above (I’m an ME after all). But I just received a demo unit of a SpeedTach that Robert Laschinger at Planet Instruments is offering. With the 2130/2140 it functions as a SpeedVue and also a tach. According to Robert, unlike a standard laser tach that just looks for the reflected beam that corresponds to 1X shaft turning speed, his tach laser is actually emitted at a fixed frequency and the receiver looks for a signal that is returned at that frequency, thus rejecting other reflected signals (sunlight, fluorescent or led lighting, etc). I have no idea if that relates to your problem. But....

I am almost certain that Robert completely understands what you are talking about, and if he doesn’t know of an existing solution, can likely design one. His contact info can be found at https://www.planetinstruments.com/contact/

__________________
"The trend is your friend"
John from PA

Sr. Member
Registered:
Posts: 963
Reply with quote  #3 
Curran, you mention “infrared” and then switch to “laser”.  I’m sure you realize there is a difference.  

In my experience doing torsional work I never got an infrared device to work well, if at all.  Most of the work using infrared devices would date to the timeframe you mention, 35 years ago.  I almost always had to rely on an active magnetic pickup (or proximity probe) looking at a split gear over the shaft.  Years later I did find some success with laser but the reliability factor just wasn’t there.
Curran919

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 599
Reply with quote  #4 
I could interpret that in two ways, I guess:
  • Its a specific wavelength of light (so the frequency he refers to is in Terahertz), or
  • He is pulsing the light at a lower frequency (as I think is happening with my sensors) so you can account for the ambient light when the pulse is 'off' and 'subtract' it from the light when the pulse is on to get a truer reading of the surface color.

In that latter, it would have to be much faster than 3.6kHz to be useful for torsional vibration.
Curran919

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 599
Reply with quote  #5 
Quote:
Originally Posted by John from PA
Curran, you mention “infrared” and then switch to “laser”.  I’m sure you realize there is a difference.


Naturally. By that I meant that (like a laser tach) they didn't work with standard zebra tape. However, unlike a laser tach, there was no visible laser. So i opined that perhaps there were infra-red lasers, which clearly exist, but probably not in this application.
Curran919

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 599
Reply with quote  #6 
BKSV documentation had some good data on this (bottom of p2):

Quote:
Originally Posted by BKSV
Type 2981 uses a continuous-wave laser. A tacho probe based on a continuous-wave laser avoids the phase jitter from tacho probes based on pulsed or sampled lasers and provides the precise rotational speed and phase information needed for order tracking, phase or balancing applications. It also minimizes the size needed for the reflective target.

OLi

Sr. Member
Registered:
Posts: 1,914
Reply with quote  #7 

I use the ROS-P and the similar laser version (same vintage) and a coptech box that include a 2907 I think tacho chip (it can be done with a 555 timer chip also) it is a frequency to voltage converter and the ROS-P and or the input stage of the converter chip make the pulse a constant width pulse train, not a perfect square wave, that is for phase measurement when balancing in my book. So the constant width pulse train where the frequency is converted to a voltage in the chip, given specific filters, integration etc. give a varying signal level that vary with the frequency variation of the input pulse train and the output is near a perfect DC level if there are no variation and it is a AC signal that you can do your FFT on and find what torsion frequency you can find if the signal vary.

If you do the magic some other way, other things may apply and your mileage may vary but that's how I do it. 


__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
Curran919

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 599
Reply with quote  #8 
Oli, the ROS-P requires retroreflective tape, no? How do you get a multiple-pulse-per-rev to do your torsional calcs?
OLi

Sr. Member
Registered:
Posts: 1,914
Reply with quote  #9 
no it's for normal 3M reflective used for speed signs and such here or similar. I printed 3M strips w. 50/50 black print.
ROS-P have a built in trig function so output is digital, even more so after the suplly box where it is 5V TTL output that I use.
Only brand I seen so far that strictly require polarized reflective is Sick and I got a big roll of that too, sigh, why make it easy but no zebra of that.
I am normally not that picky, I just so far wanted to verify that there are a Torsion issue or not and if so the frequency. It is just rarely I need to get the details and then it may warrant a encoder, finding a cogwheel or even strain gauges. I just want a quick and cost effective way that most of the time verify no problem.

__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
planet-i

Avatar / Picture

Member
Registered:
Posts: 5
Reply with quote  #10 

Curran,

Quote:

  • Its a specific wavelength of light (so the frequency he refers to is in Terahertz), or



Yes, a 650 nm red laser (461 THz!) can easily be filtered with a simple red lens, or more precisely with a Visible Band-pass Filter 650 nm. Most laser tachometers, including the Monarch PLT200 and Emerson SpeedVue 430, use a visible band-pass filter in front of their photosensor. The filter is used to block ambient light (noise) from the sensor.

As a side note, from the previous thread, sunlight contains a broad spectrum of wavelengths which includes that of the red laser. The 650 nm wavelength component of bright sunlight can pass through the optical filter and overpower the signal from the reflected laser. This problem usually does not occur with bright fluorescent or LED lights because their spectrums are not continuous and usually contain much less energy in the 650 nm band.

Quote:
 
  • He is pulsing the light at a lower frequency (as I think is happening with my sensors) so you can account for the ambient light when the pulse is 'off' and 'subtract' it from the light when the pulse is on to get a truer reading of the surface color.

Yes, the SpeedTach sensor that Rusty mentions incorporates this type of filtering in addition to the visible band-pass filter. Instead of a continuous-wave, the laser is pulsed on/off @ 15 kHz and a synchronous detection method is used. The signal from the photosensor is passed through a 15 kHz band-pass filter to only detect the reflected light from the modulated laser. The constant (DC signal) ambient light is eliminated as well as any light not modulated at the 15 kHz carrier frequency. This greatly improves sensitivity and in many cases allows the sensor to work without the need for retro-reflective tape. Only a small change in reflection is all that is needed. This filtering is highly effective in eliminating background noise (i.e. bright light, 120Hz fluorescent flicker, shadows from fan blades, etc.), but like all sampled systems it has bandwidth limitations (< ½ sample rate).

Your application would most likely need a fast (high bandwidth), analog (non-discretized in time) output sensor.

Are you processing the zebra strip sensor output as a digital or analog signal?

 

Curran919

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 599
Reply with quote  #11 
Aaah, glad to see I put the pieces together correctly, Robert (?). I had no idea before last week that optical sensors were typically 'pre-sampled'.

I have found these continuous wave laser tachometers from B&K (which I already own, luckily) and PCB/TMS. If I had zebra tape with retroreflective vs. matte black regions, then I suppose this would work fine for the torsional vibration. The only commercially available retroreflective zebra tape I found was two colours, but both reflective, which doesn't help. Like Oli said, I can find some retroreflective sheets and print a pattern on them. Gotta get the ol' inkjet working again, I guess.

So I imagine optical encoders are using the same transducers as an infrared sensor, and the only difference is the beam-interruption style vs. reflective style. I assume the encoders can be used as 'continuous-wave' because of the lack of ambient light. I wonder if the shaft is sufficiently shielded from ambient light, whether the reflective-type optical sensors can also be used with continuous wave? I'd prefer a commercial option, but I suppose I can breadboard a few components too.

I'm recording the TTL signal analog (through NI-9234 at 51.2kS/s, which is fine for low speed and low PPR) and digital (NI-9361) as well as in a 2GS/s digital oscilloscope to get a good overview (for validation). Plus most of the bells and whistles for preprocessing (frequency divider, bessel filter, differential amplifier, etc.).

The continuous-wave encoder seems to have a pretty sharp rise time even above 30k PPS, so the sensor bandwidth is fundamentally high enough, but I could also imagine that interrupt-style sensors have a higher bandwidth than the reflective-style.
planet-i

Avatar / Picture

Member
Registered:
Posts: 5
Reply with quote  #12 

I have good results printing custom zebra tapes with a Brother P-Touch label maker and the P-Touch Editor PC software.

The editor software allows you to design and print individual labels up to 1 meter long. The easiest way to draw the bars is to create a text box with a repeating capital ‘I’ using the ‘Arial Narrow’ font in ‘Bold’:
Like this -> IIIIIIIIIIIIIIIIIIIIIIIIIIIIII…
Then select ‘Free Size’ on the text box to scale to the desired lines/cm.

The P-Touch labels come out of the printer laminated and are available in Black on Reflective Silver. I have only used the black on white labels from Brother, so I don’t know if these will be reflective enough for your tachometers.

Here is an example of the labels I printed:
IMG_1861.jpg

Note: The lower cost P-Touch label maker models will work but only have 180 dpi resolution, so check that the model supports 360 dpi and works with the P-Touch Editor software. The higher resolution is recommended to lessen any granularity effects on the width of the lines when scaled and printed.

Are you using LabVIEW to process the signals from the NI daq cards?


Robert Laschinger
Planet Instruments, Inc.


Curran919

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 599
Reply with quote  #13 
Ah cool. I think the silver tape would work. Its only flat reflective instead of retroreflective, so you'd need to get the laser angle of incidence pretty orthogonal, but I guess doable. I also notice they have a black on clear that you can just lay over retroreflective tape. Thanks for the idea.

The digital signal I've done up some hacky .vi while I am validating the method. Then I'll incorporate it into our standard homebrew DAQ .vi. The analog stuff is all processed in Matlab.
Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.