Noob UK build

[ChrisK]

My arm calibration seems to be working out pretty well in that all arms are pretty close to each other. I'm a fair bit out and tomorrow I will hook up the computer to re-adjust the values. At the moment I'm getting 38.5mm movement for a 40mm command on x and z with 38.58 on y.
My J-head arrived yesterday and offering up I'm a bit befuddled. I always thought that the J-head should slide into the hub and the sideways bolt holds would be used to pass bolts through the groove in the top of the hot end but mine does not fit. Now I know that all holes should be gone over a bit with a drill but the difference is quite significant. I'm measuring up the hub ID as 15.6 and the hot end external diameter at 16. Should I try and drill out the hub or is the norm to use an adaptor?? I know Tim has used an adaptor and brandon, rather usefully, has linked to a few adaptors on another thread. Has anyone mounted their J-head in the hub?

[Nicholas Seward]

I am pretty sure I made the ID 16.1mm. I will check.

[timrwilliams]

I seem to remember my J head was rather a tight fit, and needed some hub filing work to get it to seat.
As you know, that wasn't the main reason I used an adaptor, that was cooling improvement.

http://www.thingiverse.com/thing:33465/#files from Piit79 (to whom - many thanks) was the design I used.
The SCAD file was helpfully provided, so I managed to adjust the stub end dimensions to be a sliding fit for the hub. Then a couple of filed grooves to take the fixing bolts.

[ChrisK]

So, after a long spell away from my GUS I'm back on getting this to work even though I wimped out and bought an off the shelf cartasian printer (CTC).
I managed to find myself an Anubis hotend from a gent I met at one of the 3D printer meetups at Hackney Hackspace which now means my arms don't get near the singularity and collapse. My geometry now is probably very close to the norm.
I've fixed a few firmware settings such as setting my minimums to 0. Before that following a home command my printer would think it was well into the negatives and even a small positive move would result in the arms shooting out of bounds .
I'm still struggling with the bed levelling calibration in Segmentise. Firstly my co-ordinates were throwing up 'maths domain error' but now I have better values the g-code generated for some reason is in the negative domain. Running the unchanged, downloaded segmentise gives the right sort of movement but as you would expect not necessarily near the printer bed. A cheeky cheat I discovered to test values one by one for the maths errors is that even though segmentise expects 6 minimum values you can re-use the same value more than once. I used this trick to swap out readings one by one until I found the ones that broke the maths.
I do have a problem with my Y motor. Sometimes I can hear a rythmic light ticking which sometimes manifests as slight movement. If the arm is trying to move when this is happening it affects the movement causing moves to be disrupted, possibly what was causing my innacurate readings before the problem became bad enough to notice. On a related note I may well need a chipset fan as the thermal glue that holds the heatsinks on my stepper drivers liquefies.



At the moment I am using the GUS as a simple teaching delta robot building gcode routines by hand in Repetier. I am planning on using a model hotend file to make a number of tools such as hooks and stamp holders so the kids can use GUS to programme simple tasks like moving parts from point A-B and using stamps to mass produce X-mas cards

[timrwilliams]

A board fan is a good idea. I always fit one on new builds. It may solve your Y axis problem. I don't know what hardware you have, but most stepper drivers have over temperature protection which can cause strange effects.

On the subject of boards - I am starting to feel way out of date. I started with a Sanguinololu (how old is that, but it worked) and recently changed to RAMPS 1.4. But judging by reports, 32 bit is the way to go. I am not sure I can face another learning curve!

I too had some fun playing with Segmentize values. To challenge it, I tried 6 x mid position (122.4,122.4,122.4 in my case). It actually produced some fair gcode, the print lifted at one side, but was not too bad. It set me thinking and I realised I don't fully understand the maths.

Other posts on the forum suggest that GUS MK2 with in-firmware maths is close to a result. I look forward to more reports.
I think it likely that Mk1 will never produce really accurate prints and I like your idea of using the pick--n-place ability. As a delta robot GUS certainly has a WOW factor. I sometimes run it with M302 - (allow cold extrusion) just to sit and watch the action (Yes, I know I should get out more!)

Seasonal best wishes