note that the signal is moving at ~200m/us, and we are needing to measure <0.1mm
in length. that is going to take VERY fast electronics to do.
and it will be an extremely noisy signal to analyze, you not only have the links
showing up, but also your contact to the chain, and the fact that the signal
goes both ways in the chain (is the short measurement towards the sled or away
from the sled? you have no way to know for sure)
are you talking about machine resolution or measurement resolution?
the limits to machine resolution right now are the softare and calibration, our
precision (repeatability) is much better than our target of 1/64"(~0.4mm), but
our accuracy is not that good. There are a number of people who are reporting
getting in the 1mm range on accuracy, but many others are struggling, and the
big effort right now is trying to solve this.
sorry, upgrading the chains won’t help the type of breakage we have had
reported.
look at the pictures of the breakage, you could go with a MUCH heavier chain,
and the particular joint that broke would not be significantly stronger. It’s
not a joint that is supposed to have any stress on it, and in normal operation,
it is not at all important (in fact, having that piece free to rotate would be a
slight advantage in many cases)
We have just been using the chain wrong on everything except the top mount
triangulation kit.
If you built an inexpensive TDR that you could put on each chain and run it in real-time, then you could switch away from chains to something else… say a light wire. At that point, it doesn’t matter if the cable stretches. But I don’t know the feasibility of building an inexpensive TDR that can give you sub-mm accuracy… Not too sure about the feasibility of building an expensive TDR either.
That’s why I was mainly thinking of using it for a cable based system. Looking into driving cables like a lead screw with anti twist. Or only use it for continuity check.
it’s heavier, so the chain sag problem will be larger, and at some point we
end up with the weight of the chain applying more pull on the sled than the
minimum we want
it’s bigger, so the sprockets are bigger. which means you need more powerful
motors, higher resolution encoders, etc (or at least higher gear ratios)
it’s probably more expensive (although using bicycle chain may be slightly
cheaper). the cost of the chain is currently not a large part of the machine
cost, somewhere around $1/ft when purchased in bulk, so <$25
Since sag and flexibility act in different directions we may be able to brance and treat the moment arm as a solid while still having the flexibility of chain.
Am i mistaken, or is #25 chain not actually roller chain?
I seem to remember that roller chain has actual rollers, whereas #25 ansi chain does not have rollers. Instead it has bushings. IIRC, #40 chain is the smallest chain that uses actual rollers.
Edit: i found where i had read that
interesting reading… more than you ever wanted to know about chains…
On the surface, there appears to be several issues with the TDR concept:
They are difficult / expensive to build for precision: Ive not seen hobbyist TDRs like this.
To be accurate, they require a “launch Cable” ahead of the Cable Under Test.
They are not fast; the $30k units I’ve used have a refresh rate measured in single-digit Hz.
They rely on good cable insulation and reactance isolation.
Recognizing the bend in a cable (where it hits the take-up spool) would rely on a very sharp (cable-damaging) bend, and it may still be impossible to get repeatability on measuring to the same precise point on the cable where the spool begins to bend it.
Recognizing the increased conductivity of the cable as it touches the take-up spool would rely on a bare cable which would be subject to humidity, ambient ionization, RF interference, and other variations that could be difficult (impossible?) to compensate for.
Maybe a laser TOF would be more feasible, but the affordable, off-the-shelf designs do not hit sub-mm accuracy. It might be good as a rough position audit for safety cut-off when something breaks, but not for moving the sled in a precise line.
I have designed a different function of the maslow. Moving the motors from the outside edges, and onto the actual platform with the motors. that way the spacing between the motors is always the same it will never change, then cables wrapped on grooved drums are spooled and measured (trying to figure out how to manipulate software to understand that pulled distance for calibration) and at the tops were the cables actually anchor are a type of swiveled anchor point. so no need for bricks anymore as the motors become some of the weight, also the sled i made out of epoxy granite and has a super smooth hard surface glides well and will last for ages. the benefit here is the motor and electronics and everything can be contained in one unit and stored safely away. once i’m done working bugs out with it i’ll debut here. but some food for thought for now. cables work well just like someone said basically don’t leave hanging around as they will develop some “memory”
Awesome, I’m definitely interested in cable drives. A rest would allow the cables to sit without constant tension. I’m working on a new frame design that can fold up.
Going to work on some concept art and CAD of the systems I’m looking to develop. Have CNC machines and 3d printers to help others as well. Collaboration is the strength of open source.