Absolutely agree with you there.
Yep, I also agree. It will need a lot of testing to adjust/verify. Just an observation: the chain sag doesn’t appear to be much different, other than it goes in the intended direction down in the problem areas.
My main concern is the force on the motors. While the rotational force is actually reduced, the sideways loading on the shaft is probably about double what a standard tensioner would be. That could cause bearing failure if it’s too high.
I have considered the same arrangement… one fly in the ointment is that the top pulley on each side would need to be at least 10’ high to cover the full motion of the sled, top to bottom on both sides. Also, the design needs to create more pull towards the outside the closer the sled gets to the bottom corners… it seems like there will be no change in performance with this layout, as the weights just cancel each other out. Maybe using the slack chain with pulleys and bungees could create what we want, but the added complexity would likely create other problems. Perhaps the best answer is to make the top beam wider to increase the motor spacing by maybe a foot on each side? More chain, of course.
Block and tackle setup took care of the 10ft. But I agree, I think a 12 ft top bar may be the best chance.
I like the idea of making the top beam wider to give gravity a better shot at redirecting the sled to the desired location, I just have a fairly small space to work in currently. What’s the largest top beam someone has installed to date?
I’ve created a new topic to expand on the non-router-sled related discussion of the tensioning system that everyone is asking about.
It’s here if interested: Tensioning system to improve performance in bottom corners
Reply to this thought is in Tensioning system to improve performance in bottom corners
chain sag is different in that it’s more predictable when the only down force
The current calibration is able to make a fairly good guess at the correct
constant to account for it, and then it compensates for the sag consistantly.
With the additional tensioners on the bottom tied to the length of both chains,
the math to figure out how much force is being applied (and therefor how much
tension there is on the chains, and what that does to the resulting chain sag)
is FAR more complex, and I expect has lots of variables in what the lengths are,
how much tension the stretchy cord applies (which changes over time), etc.
I don’t need an elaborate answer to this but how is chain sag compensation done without knowing the weight of the sled and angle of the frame? Are just average values used?
the calibration measures the effect of the chain sag and iterates the constant
until the calculation with that constant matches the measured effects of the
This avoids the need to weigh the sled, the chain, etc.
Wouldn’t that also accommodate the force of tensioners?
only if the effect of the tensioners is nice and linear. But all these tricks
that tie the tension on one side to the chain length on the other in various
non-linear ways will throw off the calcuations
at leat, I think that’s what will happen That’s why I say that it would be
good to have someone test with the stock weights and test with this and see what
the benchmark test shows (you may want to actually cut more of a grid across the
workpiece to detect other non-linear effects, the accuracy benchmark assumes
linear error and so it is only doing the minimum amount of cutting to detect
going to a 12’ top beam should give the sled at least twice the force to move it into the bottom corners, so that is by far the simplest and most predictable way to improve performance there.
The other thing to do is to reduce the amount of force needed. Make the sled slick, and experiment with making the frame closer to vertical.
We know that tilting it back further (say to 20 degrees) makes the machine completely unusable (When bar was trying to build the new sled design, at one point he misunderstood the directions and built an otherwise current machine with this angle, so we know it’s bad)
in the early days (original frame and long before chain sag compensation) we had someone try with angles down to 5 degrees from vertical and the finding then was that at 5 degrees he started having other problems.
so experimenting with machine tilt could reduce the friction and the existing force would move the sled better.
I think trying these things would be a good idea before adding a significant amount of additional complexity is a good idea.
I think I will try 5 degrees tomorrow. I have a benchmark at 15 degrees. I can run a benchmark at 5 to compare. Any other tests I can run at 15 and 5 degrees that will give you any insight?
I’m assuming you’ll do the calibration test cuts at 5°? I just learned today that those measurements dial in the chain sag correction along with other things.
one key question is if you have problems with the router pushing away from the
workpiece as you lower the Z axis into the work.
a good benchmark for testing in the corners is to do a L (in the bottom left),
do you trace the same path going down, right as you do going left, up?
Left, then up is a particularly problematic action and may round the corner as
there isn’t lots of force to move the sled left, so it’s got a good chance of
sticking until the left chain starts pulling up, at which point it will swing
left, but as it’s being pulled up, it will round the corner
Down, then right could end up with a squiggly vertical if the sled is not
sliding smoothly don the workpiece.
you cn shift the workpiece over, and do it again at a different angle (and with
different amounts of weight on the sled)
does going closer to vertical help the L to be sharp?
when it’s sharp, can you reduce the weight on the sled and have it remain sharp?
is there any tendency for the sled to push away from the workpiece as the bit is
drilling into the workpiece? (do a reasonably deep pass so you can see this
if you can get a video of each test, that can tell us a lot as well.
There is the calibration test cut and the accuracy benchmark test pattern.
the calibration test cuts are short cuts. The accuracy benchmark is four squares
plus a handful of short cuts.
If you want to do the accuracy benchmark, you need to do the calibration with
the same angle/weight.
Before you start doing that, try the L cut in the bottom left that I describe
above. When you have a combination of angle and weight that works well for that
test, go ahead and do a full calibration and the accuracy benchmark to check and
see that it’s working everywhere else.
Hi David, we’ve moved discussions about the tensioning system to a new thread. I can paste your responses to the new thread if you need me to. (Tensioning system to improve performance in bottom corners)
That’s fine with me. I generally reply to posts by e-mail and usually don’t pay
much attention to what thread they are in.
Please make sure you post enough context that it’s clear what I’m replying to (I
try to include that in each post, but don’t always)
Just wanted to say that I finally squeezed in 2 hours of time to work on the Maslow and I updated it with the thick stretchy string and those fantastic wall mounted pulleys you this thread turned me on to.
New top tensioning system complete! Also added a spacer under my motors because I was just over the 15" minimum motor height before. This should push me up to almost 17".