We need to measure a rotational speed of the shaft diameter 1200 mm, and only the shaft surface is available for speed pick-up. Preferred output is 10000 PPR with full quadrature. The first idea was to use a magnetic band with incrementally coded scale (used in linear encoders), but how to make accurate joint to complete the circle? Or, maybe there are some other methods? Please advise.
Eng-Tips is the largest forum for Engineering Professionals on the Internet.
Members share and learn making Eng-Tips Forums the best source of engineering information on the Internet!
-
Congratulations JStephen on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!
Large shaft speed measurement 4
- Thread starter thinker
- Start date
Non-contact speed sensors with reflective tapes or proxmimity probes with a notch/keyway/welded pad in the shaft.
Muthu
Muthu
electricpete
Electrical
That would give one pulse per revolution which is good for some purposes, but apparently this guy needs 10000 PPR
=====================================
Eng-tips forums: The best place on the web for engineering discussions.
=====================================
Eng-tips forums: The best place on the web for engineering discussions.
-
2
- #5
Skogsgurra
Electrical
One magnetizing head working on the steel surface like the recording head of a tape recorder) and two pick-up coils positioned 1/4 period away (for quadrature) from each other does it.
The tricky 'make ends meet' thing can be fixed quite easily by looking at the picked up signals and adjusting the signal frequency so that the last period ends where the first period starts.
It takes several rotations before you have everything in place, after that you can enjoy a stable high-resolution speed signal.
If you only need to measure speed and do not need to have an update more than once per revolution, then reciprocal measurement is a no-brainer.
It may be tricky to have 10 kPPR. That will make the period of the magnetizing cycle equal to around a third of a millimeter. That can be done, but it takes a nice shaft surface. Putting a lower resolution pattern, like 1 or 2 mm period is quite easy.
Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
The tricky 'make ends meet' thing can be fixed quite easily by looking at the picked up signals and adjusting the signal frequency so that the last period ends where the first period starts.
It takes several rotations before you have everything in place, after that you can enjoy a stable high-resolution speed signal.
If you only need to measure speed and do not need to have an update more than once per revolution, then reciprocal measurement is a no-brainer.
It may be tricky to have 10 kPPR. That will make the period of the magnetizing cycle equal to around a third of a millimeter. That can be done, but it takes a nice shaft surface. Putting a lower resolution pattern, like 1 or 2 mm period is quite easy.
Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
Gunnar's solution is very elegant.
However if I understand quadrature signals correctly, we only need 2500 magnetic periods to create 10000 PPR, so the magnetic period will be around 1mm on the circumference of a 1200mm diameter shaft.
I'm further curious if anyone knows how fast the shaft can travel beneath such a magnetizing head and pick-up coils (leading to the max rpm for this solution).
Thanks,
Sean
However if I understand quadrature signals correctly, we only need 2500 magnetic periods to create 10000 PPR, so the magnetic period will be around 1mm on the circumference of a 1200mm diameter shaft.
I'm further curious if anyone knows how fast the shaft can travel beneath such a magnetizing head and pick-up coils (leading to the max rpm for this solution).
Thanks,
Sean
Skogsgurra
Electrical
It is understood that you get four times the PPR. If you, for instance, specify a 1024 PPR encoder, you get an effective resolution equal to 4096 increments. That is how all encoder vendors specify their devices.
I do not think that the speed of a 1200 mm shaft is more than one or two hundred RPM - at most. So the frequency will not be extremely high. Many unknowns, the biggest problem is to keep the distance between pick-up and shaft surface close and constant.
Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
I do not think that the speed of a 1200 mm shaft is more than one or two hundred RPM - at most. So the frequency will not be extremely high. Many unknowns, the biggest problem is to keep the distance between pick-up and shaft surface close and constant.
Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
- Thread starter
- #8
electricpete
Electrical
Gunnar / Mobius - can you explain the principle more?
We apply AC at the stationary head I assume. If shaft was stationary we create a traveling wave which result in ac sensed a the stationary pickups.
When the shaft starts rotating, I can imagine that eddy currents would change the pattern seen at the pickups, but it's hard for me to understand what relationship it will have to speed or position.
=====================================
Eng-tips forums: The best place on the web for engineering discussions.
We apply AC at the stationary head I assume. If shaft was stationary we create a traveling wave which result in ac sensed a the stationary pickups.
When the shaft starts rotating, I can imagine that eddy currents would change the pattern seen at the pickups, but it's hard for me to understand what relationship it will have to speed or position.
=====================================
Eng-tips forums: The best place on the web for engineering discussions.
Skogsgurra
Electrical
You record a periodic pattern on the shaft surface with the recording head. When you are satisfied with the pattern you remove the recording head. Then read the signals with the other two heads. Similar to a tape recorder.
One problem with this methd is that you need to use either hall devices or flux gates to read signals near zero speed. But if you only need to measure speed above a certain limit, then you can use inductive pick-ups.
Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
One problem with this methd is that you need to use either hall devices or flux gates to read signals near zero speed. But if you only need to measure speed above a certain limit, then you can use inductive pick-ups.
Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
electricpete
Electrical
Aha interesting. That makes more sense. Thanks.
=====================================
Eng-tips forums: The best place on the web for engineering discussions.
=====================================
Eng-tips forums: The best place on the web for engineering discussions.
Skogsgurra
Electrical
You mean zero pulse?
That is not needed when measuring speed. Only when positioning. It can be done in the same way. But you need an extra track, just like you need in an encoder.
I shall add that I have not used this technique IRL. But a guy named Orvar Dale did when he was experimenting with what later came to be the Torductor/Pressductor. This was back in the fifties and electronics was vacuum-tube based. I shall google the thing to see if it is being used today.
Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
That is not needed when measuring speed. Only when positioning. It can be done in the same way. But you need an extra track, just like you need in an encoder.
I shall add that I have not used this technique IRL. But a guy named Orvar Dale did when he was experimenting with what later came to be the Torductor/Pressductor. This was back in the fifties and electronics was vacuum-tube based. I shall google the thing to see if it is being used today.
Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
Skogsgurra
Electrical
Google was not my friend here. 'Tape recording' doesn't seem to be used any more. Laser correlation techniques have taken over. There are several ready-to-use systems available.
That would, of course, be a faster and easier solution. But probably less fun :-(
Accuracy and time delay can be a problem in laser techniques. It depends what the specs are.
Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
That would, of course, be a faster and easier solution. But probably less fun :-(
Accuracy and time delay can be a problem in laser techniques. It depends what the specs are.
Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
Skogsgurra
Electrical
Sometimes it is only about specmanship. Sometimes, there is a real need for high resolution.
High PPR numbers are necessary when positioning, but seldom for speed measurement. There are simple encoder functions built into standard DGBBs and they seldom have more than a few hundred PPR. That has proven adequate for many speed measurement purposes.I did some work for SKF selling their 'Sensor Unit' to the salesmen and the main problem was to make them understand that high PPR is not always necessary - sometimes not even desirable. There are a few slides about that in
Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
High PPR numbers are necessary when positioning, but seldom for speed measurement. There are simple encoder functions built into standard DGBBs and they seldom have more than a few hundred PPR. That has proven adequate for many speed measurement purposes.I did some work for SKF selling their 'Sensor Unit' to the salesmen and the main problem was to make them understand that high PPR is not always necessary - sometimes not even desirable. There are a few slides about that in
Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
-
2
- #18
Both Heidenhain and Renishaw make tape scale encoders that can can be wrapped around a shaft. The tape scale comes with an adhesive backing. The read head is mounted separately. The optical pitch is 20 microns or 50,000 cycles per meter, 200,000 pulses/meter. The A or B channel maximum frequency is above 200,000 Hz. A 1.2 meter diameter at 30 RPM will be less than that. There is an Index pulse.
You will have 750,000 PPR.
You will have 750,000 PPR.
electricpete
Electrical
I think high pulses per revolution can be useful for elaborate torsional vibration measurement systems also.
One thing monitored is speed oscillation. If machine nominal speed is fmachine and modulating speed is fmodulating, then the resulting spectra has components at frequencies |fmachine +/-fmodulating|, and the higher-frequency of these is at frequency of fmachine+fmodulating. Based on Nyquist criteria, we need to "sample" the speed (computed from position) at a rate exceeding 2*(Fmodulating+Fmachine) in order to avoid aliasing. A common relatively high modulating frequency is twice line frequency (this frequency excitation occurs as result of electrical unbalance and can excite resonances such as turbine blades if not properly detuned). So for this case we may need pulses at > 2*(7200+30)cpm. ~ 15,000 pulses per minute. Since the machine is rotating at 30cpm, we would need 500 pulses per revolution (15000 ppm/30 rpm = 500 ppr) to be able to sense that particular speed oscilation. Note gear mesh can also be sources of high frequency torsional excitation.
I believe even tighter requirements can apply if it is desired to measure torsional strain (twist) by comparing position at two opposite ends of a shaft... because we are trying to measure a small angle which may be changing over time. The angle associated with pulses must be much smaller.
=====================================
Eng-tips forums: The best place on the web for engineering discussions.
One thing monitored is speed oscillation. If machine nominal speed is fmachine and modulating speed is fmodulating, then the resulting spectra has components at frequencies |fmachine +/-fmodulating|, and the higher-frequency of these is at frequency of fmachine+fmodulating. Based on Nyquist criteria, we need to "sample" the speed (computed from position) at a rate exceeding 2*(Fmodulating+Fmachine) in order to avoid aliasing. A common relatively high modulating frequency is twice line frequency (this frequency excitation occurs as result of electrical unbalance and can excite resonances such as turbine blades if not properly detuned). So for this case we may need pulses at > 2*(7200+30)cpm. ~ 15,000 pulses per minute. Since the machine is rotating at 30cpm, we would need 500 pulses per revolution (15000 ppm/30 rpm = 500 ppr) to be able to sense that particular speed oscilation. Note gear mesh can also be sources of high frequency torsional excitation.
I believe even tighter requirements can apply if it is desired to measure torsional strain (twist) by comparing position at two opposite ends of a shaft... because we are trying to measure a small angle which may be changing over time. The angle associated with pulses must be much smaller.
=====================================
Eng-tips forums: The best place on the web for engineering discussions.
- Status
Similar threads
- Locked
- Question
- Locked
- Question
- Locked
- Question