I’m looking for a solution to easily measure chain wear and in particular, if it’s uniform across the entire chain.
@dlang suggested linking the left and right chain together into one long chain and doing the distance between motors measurements with different segments of the chain to see if the measurement is consistent. If each chain is worn the same uniformly, the measurements should be the same regardless of which portion is between the sprockets. However, managing the slack part of the chain, moving them on the sprocket, etc. would be tedious and I’d probably have to modify the code some to facilitate the measurements.
I was wondering if, alternatively, I could use my digital caliper and measure the distance between a set of bushings on the chain to get a reasonable result. My calipers are capable of measuring up to 6-inches or 24 links of chain. I think I can fit the calipers into the gaps between the bushings and “stretch” the links out with the calipers to make a measurement:
If I repeat this process along the chain and get different measurements, then I know the chain wear is not uniform. If I add the diameter of a single bushing to the measurement and subtract the length of an ideal chain (0.25 inches * the number of links) I’d get the amount of stretch. I think this would be relatively easy to manage to do…
I’m lucky to have a place high enough to hang the chains. I hung both from a fixture that held the top ends evenly and a tape measure alongside. Gravity provided even tension. I checked every foot to see if they were still even and the pins were “where they should be”.
I’m looking for a solution to easily measure chain wear and in particular, if it’s uniform across the entire chain.
@dlang suggested linking the left and right chain together into one long chain
and doing the distance between motors measurements with different segments of
the chain to see if the measurement is consistent. If each chain is worn the
same uniformly, the measurements should be the same regardless of which
portion is between the sprockets. However, managing the slack part of the
chain, moving them on the sprocket, etc. would be tedious and I’d probably
have to modify the code some to facilitate the measurements.
there isn’t going to be a lot of slack (just tip the machine closer to
vertical), and for moving the chain, let the motors do it (run both motors
clockwise X ticks and then pull tight, record the position of both motors, back
one motor off a little to create some slack, repeat)
I was wondering if, alternatively, I could use my digital caliper and measure the distance between a set of bushings on the chain to get a reasonable result. My calipers are capable of measuring up to 6-inches or 24 links of chain. I think I can fit the calipers into the gaps between the bushings and “stretch” the links out with the calipers to make a measurement:
If I repeat this process along the chain and get different measurements, then I know the chain wear is not uniform. If I add the diameter of a single bushing to the measurement and subtract the length of an ideal chain (0.25 inches * the number of links) I’d get the amount of stretch. I think this would be relatively easy to manage to do…
given that you are talking about errors in the range of 0.001", and that’s the
theoretical limit of your calipers, I don’t see manual measurements being
practical. You need to measure across many links, and you need to do so under
some amount of tension to keep the chain straight.
I’ll have over a full chain length’s of chain as slack when I start the process. The issue is keeping it from “kinking” if I can’t find a solution to take it up. My shed is way too small to lay it out.
If I measure 20 links (6.35*20=127.0 mm) and as maslow calculates my error to be 0.34% then it should measure out at 127.4318. This should be measurable with the calipers. I’d test it but I would literally be struck by lightning if I stepped outside.
the chain is a loop, if the length of the chain was exactly enough to go around
the motors, it would be tight top and bottom.
since you will have about an extra foot, the slack part of the chain will sag in
the middle, but I would guess that this sag won’t even get down to the level of
the workpiece.
If I measure 20 links (6.35*20=127.0 mm) and as maslow calculates my error to be 0.34% then it should measure out at 127.4318. This should be measurable with the calipers.
you have to have something to pull the chain straight as you measure it. normal
error is more like 0.15% at the extreme end of a bad new chain
your caliper has a theoretical accuracy of ~0.025mm (0.001"). This much error
per link will result in over about 12.5mm of error over 500 links.
so if you measure 20 links, your multiplier is 25 instead of 500, but that means
that you are at 0.5mm of error in chain length with perfect measurements (in
practice, your error is going to be at least 2-4x the theoretical limit of the
caliper
a ‘bad’ new chain with a slop of 0.15% will be off 0.09mm per link. better chain
will have smaller errors.
measuring with the encoders has a theoretical accuracy of 0.0003 in (0.008mm) so
you can afford to be 2x-4x as bad quite easily.
I’d test it but I would literally be struck by lightning if I stepped
outside.
the chain is a loop, if the length of the chain was exactly enough to go around
the motors, it would be tight top and bottom.
so if you have motors 10’ apart 20’ 2.5" of chain would be stretched tight
using the cable sag calculator I just posted the link to,
22’ of chain (what ships with the maslow) would sag about 3’ in the center
25’ of chain would sag about 5’ in the center.
if you put something behind the machine so that this sag is pretty much straight
down from the sprockets, this isn’t a problem (it’s only a problem if this pulls
at enough of an angle to the sprockets to want to jump off)
since you will have about an extra foot, the slack part of the chain will sag in
the middle, but I would guess that this sag won’t even get down to the level of
the workpiece.
I really like this idea! Never thought of that. It wouldn’t be that difficult for the machine to do this either. This could be a nice automated way of analyzing the chain health on a link-by-link basis, as well as validating chain tolerance parameters along with verifying the motors are in sync. Heck, it might even allow us to numerically measure backlash.
You could take one of the sprockets used for slack management and attach it to the sled (or some other weighted object) and hang it on the slack chain… this would keep the slackside tensioned to minimize chance of chain issues. Put the sprockets at a decent height and you don’t need to worry about tilting the frame
I think what this would do is determine the difference in stretch of the chain link feed out by the left sprocket and the chain link fed in by the right sprocket. I don’t think we will have an absolute measurement of a link. But it will likely test variability since the odds are that if there is variability, then the wear won’t be exactly the same between the link feeding out vs. the link feeding in. If wear is uniform then each measurement should be close to the same. To do a per link measurement this way, you need one chain perfect with 0 slop.
I’ll note that with the frame at the normal angle (11-15deg) and a counterweight on the slack chain sides, the chain enters the sprockets at a vertical angle (gravity = 0 deg) and not the sprocket angle. This has to wear at some point, some amount, on the chain.
I expect that we’ll begin to see this effect after some time on the chain, and obviously, the greater the counterweight and/or frame angle, and the more use the machine sees, more this would theoretically occur.
I plan to take a 20 link segment, put one end of the caliper at one end of the segment and the other end of the caliper at the other end. I think I can insert the jaws of the calipers in-between the bushings and then using my hands, stretch the 20 links apart. If wear was 0.34% uniformly, the caliper should read 127.4318 mm for the 20 links. This is easily within the ability of the caliper to measure. My limit of measurement would be approximately 0.02% wear based upon a theoretical accuracy of 0.025 mm.
I like the idea of automating it by running the chain around the sprocket and I’m confident it will show variability if its present, but code needs to be written to do that whereas measuring using a set of calipers would not take nearly as long to quantify whether or not there is variability.
But, if we could solve a large set of simultaneous equations to arrive at individual wear values per link (or group of links), that would be awesome. I just don’t know if we’ll have enough measurements to calculate all the variables. My gut feel is will be shy by one, but I could ponder it some.
So, it might be simple enough, using numpy, to solve simultaneous equations to arrive at a value per link… as long as the number of links is an odd number. I did 999 equations using numpy.linalg.solve and it didn’t even hiccup.
Well, it still might be a bit of a challenge to do it… ideally, you would stetch the motors apart rather than turn the sprocket to apply tension (keep each sprocket at 12 o’clock).
So, when using the chain measurement routine, when the process is done, the left sprocket returns to 12 o’clock? I have assumed that the right sprocket stays at 12 o’clock (does not move). Per @dlang’s above suggestion, how do you preserve the 12 o’clock integrety on both sprockets?
