Problem with a Loom!

[Ubergeek63]

If the shuttle is spinning then you will get multiple pulses per quarter rev. If stationary then you will get one.

Mike.

A quick measure of the picture shows the shuttle makes between one half and one revolution as it travels it's length. That alone is enough to cause the rod magnet idea to fail more often than it works.

Dan

 

[Rolf]

[snip}

Rolf,
The mirror is mounted on a surface with a HOLE across the surface, when any string of any shuttle is cut off or finished then the surface has no support (leakage of the taut string) and will be exposed due to centrifugal force. Make sense?

Sounds like they don't want any change in shuttle tread tension. How is this possible, the shuttle spool must have some drag and it must be increasing as the diameter of it gets smaller? So, is the range of tension so optimized that a minute addition can not be tolerated?

Wish I could open the ZIP'ed video clip, maybe I would understand what is going on better.

Rolf

 

[wizard]

Rolf,
The shuttle's string swings are not so noticeable at all. You think the spool's string has a lot of swings? It does not at all.

Here is a picture of a typical shuttle.
You can see the spool place (the blue line), and the related string (the red line).

In accordance with the reflector, I put a cube surface having a mirror on it under the arm and I said what would happen in the other post…

 

[wizard]

here are 3 typical pictures which I have found via the net:
You can see the reflector system on them too.

Hope they help you guys to understand me more clearly!

 

[user_88]

Wish I could open the ZIP'ed video clip, maybe I would understand what is going on better.

Rolf

Try This:
WinRAR archiver, a powerful tool to process RAR and ZIP files

 

[xanadunow]

Thanks Wizard,

I understand the process now, and very helpfull here was the drawing from this patent thingi.

The task as I understand, is to stop the process when one of the four strings breaks (or the spool on the shuttle ends).

In the xyz space, position of the string on the shuttle is a variable that can not be easily monitored.

But, the string from the shuttle is always in a very predetermined and constant (x,y) position with respect to the arm - as shown on the drawing you supplied anotating the string as the red line. If a photo sensor (transmitter and receiver) were mounted on this arm in the exact position where the string always passes (with the string being in the middle between them), then any movment of the arm can be totally ignored because the sensor does move with the arm just as well as with the string always between them.

As long as the string is present, the (photo) Rx can not see the Tx, if the string breaks - an instant detection occures.

Some tunning of the sensor would be required (I am thinking of a window comparator past the photo RX stage, and initial luminosity(current) adjustment on the photo TX side. Infrared is probably the best choice to avoid false triggering from ambient light. Also, the string vibrates diagonally to some extend and this, just as well as the thickness of the string versus the area of the Tx/Rx IR beam - is the reason for requirement of some signal processing past the photo TX-RX but within the sensor. Ultimately, the output from the sensor is a Go/NoGo type of a signal.

The string is not touching the photo TX-RX, I think this was the requirement.

Four shuttles, therefore four sensors with outputs ORed to produce the "stop all" signal.

Makes me wonder, what stops the machine when any of the "vertical" strings in the sock breaks as it being passed from many spools from the ouside diameter of the big wheel to the sock (those strings looking like numerrous spokes in the big weel).

Ofcourse, I must have missed something but I wonder what.

I guess a problem could be in the fact that the arm constantly moves doing 360 degrees rotation and there is no way to attach these 4 wires. But this is the only logical point to detect the string. Hence, there may be a requirement to incorporate a battery (rechargable at the end of the shift) and do it in the same fasion as it is done with wireless mouse (mouse has 2 of such Tx/Rx optodetectors inside)..

Regards,
xanadunow

 

[eblc1388]

A quick measure of the picture shows the shuttle makes between one half and one revolution as it travels it's length. That alone is enough to cause the rod magnet idea to fail more often than it works.

Dan

Don't get all fixated with the solution proposed.

I have said to use one magnet, but there is the possibilities to fix two with 180 degrees or even three at 120 degrees apart. If the shuttle is rotating along its axis that slowly.

The microcontroller just counts how many pulses per one whole revolution of the loom and keep a running average of the pulses count.

Any accumulated count per loom revolution of 75% or less would means one shuttle has stopped rotating. It is easy to set the trip point at 80% that of the current running average per loom revolution.

 

[wizard]

You are welcome xanadunow,


Yes I want the system to stops working when one of those 4 spool's strings breaks or ends.


According to the shuttle and the arm set I think we have 2D motion not 3D.


Well I did not use any sensor under the line, just a mirror; the IR TX and RX were mounted in the IR system at above the Loom system (Just like 3 above picture).
They are hidden under the arm and rotate in respect to it, when a string is breached or cut off then the mirror comes out of the arm and reflects the light emitted by the TX IR into the RX for an instant per every revelation.

According to the numerous strings, fortunately they have just a one axis motion. So there is a spring for any of those strings (they are more than 300). When one of them is broken then the spring hits the metal body of the system and this why the system stops working. Those strings are tough when are not broken. Makes sense?

You have to consider that using of any tick in respect to the shuttle's string causes tension. So I think the best solution is working on the spool rotation along its axis?

 

[ericgibbs]

hi LC,
Like your magnet idea, my thought would be if you fit say 3 magnets to the shuttle will the 'metallic' parts of the main assembly cause drag on the shuttle due to the magnetic interaction.?

 

[eblc1388]

hi LC,
Like your magnet idea, my thought would be if you fit say 3 magnets to the shuttle will the 'metallic' parts of the main assembly cause drag on the shuttle due to the magnetic interaction.?

This is a valid concern. More magnets is needed only if the shuttle is rotating slowly.

I have the assumption that the "space" underneath and above the shuttle are absence of metallic metal because it will be occupied by the spool of thread that the shuttle has to carry. Also the magnet(s) fitted onto the shuttle shaft which is at a smaller diameter than a fully wound thread spool of shuttle when the loom is working.

 

[xanadunow]

More questions about the shuttle

It appears to me that these shuttles do have a mechanical rather than an individual drive, i.e. there is one drive motor for all four of them?

I guess this is why the tension of these 4 strings is incorporated to the overall mechanical equation of the device and once initialy set will never change.. if the string is intact.

My logic says that we need to capture the string at the position that is known and constant and I have presented a possible solution. This is the same point at which you have started (the arm).

The next thought is the answer to the question what happens when the (horizontal sock-wise) string breakes? Well, the vertical string feeders will miss capturing the broken string, will make it's up/down exchange and the next shuttle will lay the horizontal string in parralell with the previous horizontal string that worked (i.e. 2 horizontal strings will be captured together).. could we somehow detect that? E.g. a macro camera looking at the progress of the sock? Can it recognise the change in the pattern of the sock by some measurable means? Because the pattern will change. With one string missing there will be two strings captured together folowed by one captured alone in every full cycle of 360 degrees and this will continue for as long as there is one string missing..
Can you sample the production pattern against the correct image stored in the memory and and basing on this, make the decision to stop the process? The reasoning behind this is intuitive observation of things as they happen. A human observer will pick up the broken string by the change of pattern and nothing else. Maybe the sensor should do the same - observe the pattern. Just a thought and a possible solution.

I am steering away from the shuttle monitoring, there is nothing there I can see that I can measure easily enough for a viable solution. Not only the string travells from one side of the spool to the other abut also the diameter of the spoll changes as the string is being used - what would capture that?

I guess the main problem is preventing you from sensing the tension of the string, it would possibly be - the easiest solution..

I'll be watching this thread (or string I should say ) with interest.

Regards,
xanadunow

 

[Ubergeek63]

Don't get all fixated with the solution proposed.

I have said to use one magnet, but there is the possibilities to fix two with 180 degrees or even three at 120 degrees apart. If the shuttle is rotating along its axis that slowly.

The microcontroller just counts how many pulses per one whole revolution of the loom and keep a running average of the pulses count.

Any accumulated count per loom revolution of 75% or less would means one shuttle has stopped rotating. It is easy to set the trip point at 80% that of the current running average per loom revolution.

I am not being fixated, I am a design engineer that knows the pitfalls.

In this case you are talking about beat frequencies, or in layman's body of experience, the strobe effect. A fan blade in the sun is a blur. A fan blade in AC lighting shows a distinct number of blades, the number seen is the multiple of the frequency of the light source that the blade is spinning at.

The same thing will happen with the magnet on the shuttles. At certain spool diameters the sensor will completely miss the spinning and shut down his loom. It might be alright for a hobbyist, but that is production equipment and time is money. And that is the reason for the initial query: to save the time and materials when the shuttle runs out while your back is turned.

Dan

 

[Ubergeek63]

Ubergeek63,
The factory does not like to be involved with replacing or charging the batteries.

Recharging or replacing 4 AAs every three months would be a bother?

I might be off but I know the ICs do not take any serious current. Even at 10mA you are at 10 days if you forget to turn it off at night.

Dan

 

[Pommie]

I am not being fixated, I am a design engineer that knows the pitfalls.

In this case you are talking about beat frequencies, or in layman's body of experience, the strobe effect. A fan blade in the sun is a blur. A fan blade in AC lighting shows a distinct number of blades, the number seen is the multiple of the frequency of the light source that the blade is spinning at.

The same thing will happen with the magnet on the shuttles. At certain spool diameters the sensor will completely miss the spinning and shut down his loom. It might be alright for a hobbyist, but that is production equipment and time is money. And that is the reason for the initial query: to save the time and materials when the shuttle runs out while your back is turned.

Dan

This is a very good point and one I hadn't considered. However, this will only happen if the shuttle does less than one revolution as it passes the sensor. Looking at the pictures it looks like the shuttle is (at least) 5 time smaller than the central hub and so it should do 25 revs per orbit. A correctly sized coil should pick up multiple pulses.

Mike.

 

[Ubergeek63]

This is a very good point and one I hadn't considered. However, this will only happen if the shuttle does less than one revolution as it passes the sensor. Looking at the pictures it looks like the shuttle is (at least) 5 time smaller than the central hub and so it should do 25 revs per orbit. A correctly sized coil should pick up multiple pulses.

Mike.

I believe you are confused as to the geometry of the thing. They are making bags not socks. If we assume a 1M diameter you are talking about one pulse for between 7cm and 30cm of shuttle motion. Do you think a coil of that size, assuming it is sensitive enough, is practical?

Dan

 

[xanadunow]

More on pattern recognition

It may sound complicated at first and one may think of utilising a PC to do all the hard work, but it may not be required.

If you could exstract the luminance from the video signal of the camera looking at (say) couple of square inches and take samples (pictures) at frequent interwals and use a "Sample and Hold" of the average value of luminance in a definable time frame, a comparator of the value "hold" against the next value "sampled" can be used to make a distinctive decision whether to stop or to continue with the process.

A possible pitfall of this approach is that the difference in the average luminance between the sample (pattern) that is good and sample (pattern) that is faulty may not be descrenable sufficiently enough from the luminance's "jitter" of successive "good" samples. The "window" could be just too narrow, but it is worth to try and perhaps experiment with the cameras angle to get the best results.

The expense here is in the camera, some coax and some circuitry in a box the camera's cable is connected to. We bring the composite video in, we sample the luminance only at equal intervals, we compare the "sample" with the previous "hold" sample and we make the decision. The process is dynamic but it must be descrenable.

What do you think of this solution?

Regards,
xanadunow

 

[Pommie]

Think metal detectors. Pulse Induction metal detectors induce a current in the target object and then look for eddy currents. They work reliably at 12". With a magnet providing the pulses it will be relatively simple.

My assumption on the geometry is that the shuttles provide thread that goes around the sack hence my assumption that the relationship between the two speeds is the square of there diameters. How can a shuttle go around a hub and not provide enough thread to match the circumference?

Mike.

 

[Ubergeek63]

Think metal detectors. Pulse Induction metal detectors induce a current in the target object and then look for eddy currents. They work reliably at 12". With a magnet providing the pulses it will be relatively simple.

My assumption on the geometry is that the shuttles provide thread that goes around the sack hence my assumption that the relationship between the two speeds is the square of there diameters. How can a shuttle go around a hub and not provide enough thread to match the circumference?

Mike.

Straight ratio, there is no square term in the circumference. 1M×PI around the sack and 1-3CM×PI around the shuttle.

Dan

 

[Pommie]

Yes, my bad, a straight relationship. From your guesstimate that would still result in 30 (to 100) revs of the shuttle to one revolution around the machine.

Mike.

 

[Ubergeek63]

Yes, my bad, a straight relationship. From your guesstimate that would still result in 30 (to 100) revs of the shuttle to one revolution around the machine.

Mike.

Which means you 15CM of sensing to ensure that you do not miss a pass. Combine that with the possibility of receiving anywhere from one to four pulses as the various shuttles pass the sensor becomes a bit of a math problem, if you're lucky.

That is why I was saying a battery operated sensor set and a single 4KHz inductor as an "antenna".

Dan