Hi. I'm looking for some advice on an idea to help with a process at work.
We make wiring harnesses/looms for industrial engines. wires have idents printed along them then are cut to length and usually they need to be twisted together. wire lengths can vary form under 1M to around 5M, sometimes longer
We have a twisting machine that has a counter where you can set the number of revolutions before it will stop.
The process for twisting is to measure the length of wire, look up how many revolutions are needed for that length and number of wires to be twisted on a chart and enter that on the counter. clamp the wires into the machine to the left and the other end into a similar fixed clamp on a moveable mount to the right to keep the tension. when the guard is closed over the rotating clamp it will start the twisting.
This takes time and people take short-cuts, putting in a high number and stopping when it looks right. which causes quality problems with wires being too slack or too tight.
I have been thinking of a way to improve this, I could interface an Arduino to the counter and have some way to measure the distance of someone holding the other side of the wires and calculate how many revolutions are needed. I just need a reliable way to measure distance. what would be the best way to do this?
My first thought was to use a rotary encoder, like a mouse wheel connected to something that could be pulled the distance needed, like those retractable cords for keys but be able to reach at least 5M. then looking on here I saw some topics on sonar range finding. how accurate are those? i guess it doesn't need to be exact, within a few cm will do. how good would it be at detecting a persons distance or would would the sonar be absorbed by their clothes? is there any other way of measuring distance?
I suspect a sonar sensor would work great. Soft things like clothing or fabric are usually excellent targets at short range (1-5 meters is fairly short). At this distance hard surfaces work well, but the angle needs to be perpendicular to the sensing direction or the sound will reflect off at an angle and miss the sensor. The soft targets will absorb some power, but usually the sensors have enough power to cut through that.
I actually thought about this a lot. I was working on an automatic rubber band winder for elastic powered model planes. It was going to have a sensor that sensed the person holding the far end and gave them directions on where to stand to get the proper wind.
Put some research into the sonar sensor itself, as they have different sensing patterns. If the spread is too wide you might just detect a neighboring piece of equipment and not the desired person.
Some people have used a 1D LIDAR, but that tends to get kind of expensive. You could also have the user pull out a tape measure and have a camera read the number of the tape (Obviosly not with an Arduino, but with a PC or RPi it could be done).
I can't visualise how the two ends of the wires are located, but is there any way you can arrange it so that the wire is drawn over a pulley or past an idler wheel during the loading process, so that you can measure how long it is by counting revolutions of the wheel?
PeterH:
I can't visualise how the two ends of the wires are located, but is there any way you can arrange it so that the wire is drawn over a pulley or past an idler wheel during the loading process, so that you can measure how long it is by counting revolutions of the wheel?
this is very similar to the machine we have, although ours will be a lot older proeff.de/htm/en/html/ProductsProduct_InformationTWISTER.php
there is a video on that page of a different model but the idea is the same, except its the operator will be pulling the wires tight.
one of my first ideas was to have a loop of cord running the entire length of the bench and back round pulleys at either side, one that has an encoder on that will count revolutions that can be used to work out distance. then all the operator would have to do is when loading in the wires in pull the cord at the same time as he is pulling the wires tight. the problem is that the cord might get in the way and tangled up in the wires and there would have to be a way to reset the count after every time.
jroorda:
Some people have used a 1D LIDAR, but that tends to get kind of expensive. You could also have the user pull out a tape measure and have a camera read the number of the tape (Obviosly not with an Arduino, but with a PC or RPi it could be done).
LiDAR sounds interesting but I bet health and safety would have a problem with lasers pointing at people. I think the tape measure and camera is a good idea but would be over complicated and well outside my skills.
jroorda:
I suspect a sonar sensor would work great. Soft things like clothing or fabric are usually excellent targets at short range (1-5 meters is fairly short). At this distance hard surfaces work well, but the angle needs to be perpendicular to the sensing direction or the sound will reflect off at an angle and miss the sensor. The soft targets will absorb some power, but usually the sensors have enough power to cut through that.
I defiantly think this is the way to go, there is nothing extra for the operator to do and nothing to get in the way.
any recommendations on what sonar module to use? how accurate are they?
It looks to me as if the twisting machine is in two parts that would need to be set to the correct distance apart to suit the wire length. Is that right?
yes but the moveable part doesn't get used anymore, its worn out, the grips don't work well and takes longer to set up. instead the operator usually just holds the wire to save time. for longer wires when you can't reach the machine they have attached a cord to the guard so when they pull it the guard closes and sets the machine away. one issue with this is that if the person loses grip on the wire there is nothing to instantly stop the machine, ruining the wire. i'm planning to add a way to detect if there is tension on the wire and will stop if it is let go. could also use this to start the machine, like a double pull so it doesn't accidental start early when laying in the wires.
Chagrin: maxbotix.com might be the only option. They are very well supported and documented.
i think one of those will do the job, some have an I2C bus and looks easy enough to interface to an arduino. even has sample code in the datasheet.