I am unsure about how the chain would interact I am sure @bar or someone else would be more qualified to tell you but I did notice that I was able to get better and better cuts with the center alignment sled because then it made sure that the bit was perfectly centered. I also found that the more accurate I was with the value from the center of router bit to the first chain made a difference too.
Hope this helps.
Thanksā¦ Iāve rebuilt the sled twice and am fairly happy with the alignment (did the last one by hand with a template I printed out). I need to rerun the chain calibration to do some double checking, but last night I got values of around, iirc, .3 to .4 which is significant difference (10 mm or so) in chain length from manual to automatic. I had to stop before I really looked into it. Itās a bit hard to believe thereās not something else going wrong other than chain stretching.
For me, I ended up making probably 5 sleds, I would say if you are worried about the chain, they are not that expensive to just get 2 new ones (I think $15 each from Maslow store) and then I would redo all of the calibration steps and then cut out the center alignment sled. I know you said you have one that is great and maybe it is perfect, I just know that I tried some by hand and some with machine and every time it beat me once I got it even remotely close to properly calibrated. I will say that I noticed that almost every time I did a sled without that center alignment that when I would rotate the sled I would notice the bit slowly make a crescent moon shape and that is why I went through so many sleds. I wanted it within about 1/32" over a 10" circle, which granted isnāt perfect but for what I am using it for it would good enough until I learn more and continue to dial the machine in.
Hope this helps, happy cutting! 
Itās likely I will grab another set of chains, but Iām looking into other chains (perhaps higher quality) to see if I can get better results. I have a 12-foot top beam, so the stock chains wonāt work out of the box (need to add some more length to them).
With my machine currently only accurate only to 1/16th an inch over 12 inches, I can do much better by hand and a printed template I made. I took the sled design with the center alignment, deleted the ring kit alignment holes and cut it out (so just the center alignment, brick mounts, and the 18-inch profile). I then used the template with the center alignment hole as a guide and manually drilled holes for the ring kit alignment and my router base. I got the ring kit installed and it appears pretty symmetrical.
For the router holes, I used a block of wood I drilled through on my drill press as a guide and ended up being able to install two screws from the backside of the sled all the way through the threaded base (so those aligned well). The third hole must have been off by a hair and I couldnāt get it perfectly aligned. I ended up installing that screw top down and it seemed go into the wood well enough from that direction. I havenāt done a rotation test on the sled yet to measure the level of āwobbleā but I will do tonight to see. I donāt think thatās the cause of my calibration issuesā¦ I really think its the distance between motors and the chains.
That sounds like you are even more detail oriented than I am! I like it! haha.
I also have a Maslow with extended chains due to a long cross memeber and wanting to be able to cut just past a 4āx8ā and still have a little extra chain. I chose a pseudo bolt together frame with a 2x6 and 2x4 in a T shape to make my top beam and I also had to add 18" of chain to the end of each of mine.
Are you thinking that the chain is stretching? On the chains that I have I noticed that on my wooden frame the pull tension could actually rotate my motor slightly so I had to go back with more aggressive bolts instead of screws to remedy this situation. Because it was slightly throwing off the measurement between but then I also noticed that I was not being precise enough with the top down measurements and that when I got those more precise by marking a line on my scrap board so that it was always that same location it helped significantly.
To me it seems that ever little thing helps and for me it was a long learning curve but I enjoyed seeing the progress.
Iām somewhat confident of my mounts. Iām using a 10-foot unistrut top beam with a 2-foot piece bolted to it on each end (1-foot of each 2-foot piece overlaps the 10-foot piece and they are bolted down with two 1/2-inch bolts). To mount the motors, I cut down a block of wood to just slide into the unistrut channel and then screwed the mounts to it through the slots in the unistrut. I really donāt detect any movement of the mounts or the unistrut extensions. Regardless, itās hard to believe I would see a centimeter of difference between a manual measurement and a machine measurement if all there was was motor mount flexing. And since I measure by hand with the sled attached and at the top of the frame, itās under tension as well.
Regardless, I picked up a couple of spare mounts through ebay and plan to drill a hole(s) through them to facilitate mounting them to the unistrut with a good size bolt(s). The existing slots donāt really work well (maybe someone has a good solution).
That sounds like a very stout frame, I have one with the wood I described above and then another with square steel tubing which had zero flex, I was looking at unistrut but I decided to just weld my second frame from scratch so unfortunately I cant help you with the unistrut mounts.
You are right thought, that is a pretty significant difference from the machine calibration to the measurement by hand. I am curious where the software is measuring from, like does it try to measure top dead center? Or is it measuring from another location and then figuring the distance from there. I think that would be a @bar question just to make sure that both measurements are coming from and going to the same locations?
The machine measures by just calculating how much chain it fed out to get from the left sprocket to the right sprocket (it backs out the extra two links that are added to wrap a little around the right sprocket). If the chain is truly 6.35 mm per link, it should be the same as what I get when I measured by handā¦ but it doesnāt match. My chain would need a per-link distance of 6.372 mm per link to get it to match.
A section of the chain can become stretched, though wear or during a mechanical mishap. I found that had happened to one of my chains. Doing the motor measurement is tedious, but taking the measurement using both ends of both chains would be thorough. Thatās easy for me to say - itās well beyond my patience though.
It might be that flipping the chains end-for-end and left to right every so often would be something to consider, like rotating the tires on a car.
Iāve had plenty of mechanical mishaps (chain wrapping around sprocket) but with such a difference, itās hard to imagine that the 1 cm error is limited to small section of chainā¦
On my stretched chain, the damaged was spread over about 5 feet on the sled end.
topcenter to top center.
The problem you are running into is slop in the chain, there has to be some give
for the parts to rotate, but if you think about it, with ~3m of chain and 6.35mm
links, you have ~500 links. If each of them is actually off by 1/1000"
(0.0254mm), that translates to an error in chain length of 12.7mm
a thousandth of an inch is not much clearance between moving parts.
I suspect we are at our accuracy limits until we account for this properly.