Cylinder position feedback

G

Thread Starter

Glen Boston

I need to measure the position of a long stroke cylinder, 6000mm. The environment restricts the use of external feedback devices (optics, cable transducers). I could use a transducer mounted inside the cylinder, bore, but the retrofit price is over $25K each, and I have seven cylinders to do. I heard once about a slave cylinder that could be mounted remotely, in our case at the valve stand. This slave cylinder would be small and could have the position feedback installed in it. The stroke of the slave would be directly proportional to the large master cylinder. I would like to have at least 5mm accuracy.

My question is has anyone used an arrangement like this, and how accurate would the feedback be?
 
M

Michael Griffin

The only way I could imagine the "slave cylinder" working as you describe would be if it were in series with the "real" cylinder and was being used to measure the oil flow. This would require a cylinder of the same size (oil capacity) as the "real" cylinder though. It also wouldn't account for seal leakage in the "real" cylinder so the two would likely get out of sync. If there is some other reliable way of doing this, I would be very interested in learning about it.

I don't know what your mechanical restrictions are, but I would suggest looking at an externally mounted magnetostriction sensor. You might see if you could attach it to whatever you are moving rather than the cylinder itself. Also, do you need to actually measure over the entire stroke, or is there a smaller region of interest?

5mm accuracy on a 6000mm stroke would be better than 0.1% accuracy. If this is a large device, there may be more than that amount of mechanical play in the system, so it matters just where you are measuring it.
 
K

Ken Emmons Jr.

A standard LVDT can be used remotely. Its just a cylinder and a coil and could be used remotely if coupled to your system properly. You can get an LVDT and amplifier fairly cheap (as compared to $25k). You should be able to achieve amazing repeatabilities and good accuracy depending on the LVDT you buy. The datasheet should rate accuracy for you. The output is typically analog, depending on the amplifier you can use 10V, +/-10V, 4-20mA, etc.

~Ken
 
R
Glen,

Perhaps a magneto restrictive transmitter they are certainly accurate but the probe would be 6000 long so I'm not sure how it woud allow for the piston.

Another possibility would be an ultrasonic level transmitter bouncing off whatever you are moving with the cylinder.

Hope this helps,
Roy
 
M

Michael Griffin

The only way I could imagine the "slave cylinder" working as you describe would be if it were in series with the "real" cylinder and was being used to measure the oil flow. This would require a cylinder of the same size (oil capacity) as the "real" cylinder though. It also wouldn't account for seal leakage in the "real" cylinder so the two would likely get out of sync. If there is some other reliable way of doing this, I would be very interested in learning about it.

I don't know what your mechanical restrictions are, but I would suggest looking at an externally mounted magnetostriction sensor. You might see if you could attach it to whatever you are moving rather than the cylinder itself. Also, do you need to actually measure over the entire stroke, or is there a smaller region of interest?

5mm accuracy on a 6000mm stroke would be better than 0.1% accuracy. If this is a large device, there may be more than that amount of mechanical play in the system, so it matters just where you are measuring it.
 
R
Glen,

What material is the cylinder made of? Aluminium, SS, steel if it's non-ferrous, the magneto restrictive might be a good choice, Magnet on piston probe straped to the outside. Still like U/S though, doesn't require any cylinder mods.

Roy
 
C

Curt Wuollet

I would think he is talking about a small cylinder mechanically mounted alongside the large cylinder so that it's stroke is equal and it's displacement could be measured to indicate the stroke.

There would  be challenges to 5/6000 accuracy.
I would think an LVDT  or MS semsor could be enclosed to offer the same environmental factors with better accuracy.

Regards
cww
 
M

Michael Griffin

In reply to Curt Wuollet: If you could mount a small cylinder beside the main cylinder and be mechanically coupled to it, you could mount a sensor in the same place and be done with it. If the small cylinder isn't mechanically coupled but but the hoses are connected in parallel (to divide the flow), then the two cylinders won't be synchronized.

A sensor of some sort is evidently required. It is difficult however to give a realistic recommendation without knowing what the machine looks like. If it were me though, I would be looking at whether it is possible to measure the position of what the cylinder is moving rather than the cylinder itself. With something this large you could be off by several millimetres just in mechanical play. Having the measuring sensor as part of the cylinder might in fact be exactly the *wrong* thing to do.
 
P

Peter Nachtwey

The external and parallel cylinder idea will work PROVIDING there are no forces that would bend the rod. If so the rod diameter may need to be increased. That is a long distances if the slave rod is to be unsupported.

The accuracy is more of a mechanical issue than a feed back issue. MTS rods will provide the necessary resolution. This has all been done before. I still would prefer the rods mounted inside the main cylinders just for the simplicity of design.

Now how are you going to control all those cylinders?

Peter Nachtwey
 
G
From the responses so far, I'll assume no one has tried this. Here are some responses to your questions:

The cylinder is steel.

The idea is to measure the stroke remotely, so mounting anything to the existing cylinder is out.

I would like to use a smaller cylinder, like 900mm, that we already have available and mount a transducer in it. This "dummy" cylinder would be mounted near the valvestand in anothe area, out of harms way. The cylinder is controlled with a proportional valve.

The current control uses inductive proxes mounted to the cylinder head. There are brass plugs in the cylinder shaft at key points. The proxes are set so that they are "off" when the brass slug is present, "on" when they detect the steel cylinder rod. This design works, but is a maintenance problem. The cylinder rod cannot rotate, or the proxes will not read. The cylinder head area is suseptible to damage, and we are constantly adjusting/replacing.

This is a steel mill environment.

I would love to install a transducer inside the cylinder, but the retrofit cost is prohibitive.

Thank you all for your questions.

Glen
 
M

Michael Griffin

The following is for too short a stroke (1.8m rather than 9m), but the sensor fits entirely within the cylinder without drilling the rod.

http://www.cpi-nj.com/trans.htm

You might ask them if they can do a special version with longer stroke but less accuracy. I have never used this product, so I can't make any
recommendations.
 
Glen,

So you have ruled out ultrasonic as well? That's a pity, all you need is a small target, say 6" square or the signal could bounce off part of the moving portion.

Roy
 
W

William Sturm

Could you use a flowmeter and integrate the amount of oil going in? You should have at least one prox switch on the cylinder to use as a reference position. You could use the prox to periodically reset the integrator. I have no idea how accurate this might be, but some basic calculations should prove if it is at least feasible to try.

Bill
 
Bill,

Wish I had thought of that, something like a gear meter with count up/down pulses (quadrature I think).

Roy
 
I worked on a system using the idea suggested below by Bill Sturm. It worked very well.

A PLC integrated oil flow measured by a Coriolis flow meter. Several proximity limits were used to re-reference (update the integrator) along a 9 m travel. The integrator was reset at the beginning of each motion cycle and the error at the re-reference proxes was minimal (could probably have done with out most of them).

Bill Code
 
P

Phil Corso, PE

Glen, I have a a circuit I thought about using in 1997. It uses an LDT to provide the accuracy of an optical encoder in very long-distance applications. If you want the name of the designer, contact me off-line!

regards, Phil Corso ([email protected])
 
P

Peter Nachtwey

> This is a steel mill environment. <

Nasty, I know you said you want the external cylinders to be remote but then how does the external cylinder track the main cylinder if they are not physically connected?

> I would love to install a transducer
> inside the cylinder, but the retrofit
> cost is prohibitive. <

The alternatives may cost a lot more by the time you get tired of screwing with it. At least a MDT rod inside the cylinder will have a half life of more than a week.
 
M

Michael Griffin

Bill Code: That sort of system might have a problem if you tried to hold position anywhere except at end of stroke or stalled against a solid stop. A leaky seal would cause slow creep that couldn't be detected. Whether that is a problem or not depends upon the details of the application.

For flow meters though, try the following. http://www.awcompany.com/

I haven't bought their stuff, so I can't make a recommendation on them. However, they make positive displacement flow meters that are meant to work with high pressure hydraulics. I would suspect that this type of flow meter may work better for this application.
 
G
Thank you all for your replies.

The flow meter idea sounds interesting. I hadn't thought of that. I might even have one around here that I could experiment with.

Peter – maybe you didn’t read the original post. The retrofit cost is in excess of $25,000 each, and I have seven cylinders to modify. That gives me all kinds of room to experiment, without loosing my job. I already have small cylinders with transducers that I can experiment with. There are also a lot of pipe fitters willing to help. As long as I use existing spares, it doesn’t cost anything (as far as the BIG MAN is concerned).

> I know you said you want the external cylinders to be remote but then how does the external cylinder track the main cylinder if they are not physically connected? <

That is part of the objective of this post. I don’t know and was hoping someone had tried it and could tell me. The concept was given to me by a millwright, who thought he had seen it done in some past life. I thought it was interesting enough to post here, as there is usually someone who has been-there-done-that on this forum.

Glen
 
P

Peter Nachtwey

> The flow meter idea sounds interesting.
> I hadn't thought of that. I might even
> have one around here that I could
> experiment with. <

It will not be accurate enough. Oil compresses and leaks by the piston seals when under load.

> Peter – maybe you didn’t read the
> original post. The retrofit cost is in
> excess of $25,000 each, and I have seven
> cylinders to modify. <

I can read. I am just skeptical about other ways really being cheaper in the long run.

> That gives me all
> kinds of room to experiment, without
> loosing my job. <

OK, but you know the best way to do this.

Peter Nachtwey
 
Top