Hello, I wonder if anyone can help!?
Before christmas I had my machine setup, calibrated and working wonderfully. I’ve got back in the shed this week to fire her back up to find the centre position is way out and it’s cutting oval shapes, not quite circles etc.
I have rigorously calibrated the machine 6-7 times now and the same results happen again and again. I’m so frustrated as I know the machine worked brilliantly before.
Any suggestions welcome!
Hmmm I’m not sure what could happen while sitting that could result in a loss of calibration.
Six to seven times seems like enough that doing it again seems like it will result in the same results.
I’m pretty lost, but let’s try to think of some other things to try. I wonder if maybe one of the encoders is not signaling the controller properly? It might be worth unplugging and re-plugging all of the connectors.
What happens when you run Actions -> Test Motors / Encoders?
Could there be any physical change which might have an effect?
Thanks for getting back to me.
I have already wiped the software from my mac and also did a full reset after reading another thread where you suggested to wipe the system and start again. I then took out all the connections and basically started as it was a new machine.
I then did the test motors / encoders and they all passed.
Could it be an issue with the arduino or the controller itself or some crazy chain issue? I’m so lost!
I thought I should also mention, it always seems to think the centre is in the same location after every calibration. It isn’t changing to random positions, it’s always the same, just in the wrong position!
This then makes the cuts all the wrong size and circles are oval shape.
That’s good news. At least we’re getting consistent results. We can work with that.
This sounds like for some reason the calibration isn’t getting the right numbers for some reason. Let’s try two things.
Open up the settings and see what the distance between the motors is currently indicated to be, then check with a tape measure to see if it appears to be reasonable.
Let’s check to make sure that the settings are changing during calibration. A permissions issue could be preventing the machine from saving it’s new calibration. Going through the calibration even quickly should result in that number changing so we can see if the value is being updated and saved.
Awesome, thanks for giving me hope!
I will get in the shed later and check both of those and report back.
Let us know what the results are and we’ll work from there
Ok the measurements are, 2994.85mm in the software and the measurement from the middle of each cog is 3050mm on the tape measure.
that’s a significant enough difference to cause a problem. Try chaning the
setting to match the tape measure measurement and then do the rest of the
Thanks David, just to confirm where should I start the calibration from if I manually update this setting?
Am I correct to use the measurement from the middle of the left cog to the middle of the right cog?
You should be able to update that setting and then skip the calibration entirely and do Actions → Set Chain Length Automatic.
If you change that value and then close and re-open ground control is the new value remembered?
That is exactly correct
Ok so I’ve entered the measurements in manually and restarted the software. It did remember my new measurement.
I then did the automatic chain measurement and it’s still exactly the same.
Great OK so we know that the calibration is able to remember the computed values.
Let’s try deleting the settings file entirely and starting fresh.
When you started fresh, what exactly did you do? Did you do these:
Great. Yes when resetting I followed those instructions which I saw on another forum post.
If you are using a kinematics type TRIANGULAR the following may solve your problem:
I was getting ovals instead of circles and fixed the problem by running the triangulation test cut pattern. The chain sag correction value for triangular kinematics adjustment value is needed to generate perfect circles.
One thing to consider when building one of these is wood deviation. Weather, moisture, cold days and hot days effect the frame and the plywood. Just something to consider.
those sorts of errors should result in a few mm worth of error. It sounds like
the original poster is having far more severe problems.
dlang, I agree, there would be MM worth of deviation, but in my experience that is already too much deviation. Repeatability is important. Have not built a frame or structure yet. Want to build some rigidity into the design and want to see if this system can be adapted to a rail system, similar to conventional CNC, using 3d printed parts, bearing wheels and tube rails. Was even thinking about using stepper motors. Will post my progress after some CAD work and Kinematics to work out the kinks.
If you are building a machine with a gantry, built it to be a coreXY machine and
use more traditional controllers, you can put it at an angle like the maslow,
but it will have far more in common with those other machines.
the genius of the maslow is the low-cost (and low speed) motors with the router
hanging on the chains, but that is also the big weakness. If you introduce a
gantry system, but keep the angled chains, you are keeping the worst of both
designs, not the best.
steppers have the problem that you don’t know if they move or not (especially
when powered off). With the angled chains of the maslow, this is especially bad
as there is no way to auto-home the machines to detect here the chains are. The
high ratio worm drives that the maslow uses prevents this, but adding the same
type of gearbox to a stepper produces insanely high numbers of steps needed to
move any reasonable distance.
but with a corexy design, you can home the machine the way you would a regular
CNC, and you can go with ‘standard’ servos or encoders and big DC motors with
modest gearboxes rather easily.
On the current Maslow, repeatability is actually very good, it’s accuracy that’s
dlang, you are correct in your interpretation. That is why I am trying to work things out in a simulation program. Thought if I keep the maslow angle design where the table is on it’s side I can keep the foot print smaller than a table. Hopefully I can get things worked out.