Messages

I did a rough-and-ready repeatability test this week to get an idea of how much slop there is in the arm mechanism. Put a 5mm dial indicator vaguely aligned in the x-direction resting on the pick-up head which was vaguely in the middle of the bed. Stepped forward and back along X to reasonable limits on the indicator. I was surprised to find very little error when stepping back and forward ~ <0.01mm. 80 GUI steps ~= 2.01mm - so 1 step ~= 1 thou, actually .99 thou. And that was without the warm-up procedure. So surprisingly good I'm thinking - as the small dimension on an 0402 is 20 thou then 1 thou s.b. ample accuracy/resolution.

A.

... the RVxS as accurate x/y drilling machine - like it

Good point re dimples - have sent in another batch with them, also versions which may not need the brass bush, a Tool 3 as mine is beaten up, and tests of little holes in various materials. Arriving around the new year.

A.

Hmm, I had a go at making some brass inserts and also drilling out the small nozzles with tiny drills this afternoon. Not entirely successful - the inserts were OK-ish, but the drilling wasn't convincing. it could be that it needs someone who knows what they're doing The 'Tool 2' nozzles are fairly convincing straight off the printer so if anyone has spare brass inserts and wants to try them out, let me know. Otherwise I have a few more ideas re inserts and tiny holes - if anyone has any thoughts also let me know.

A.

>>> What is critical is the runout between the inside of the brass insert and the nozzle tip...

Ah yes, I see what you mean.

>>> drill the hole on a lathe

You obviously haven't seen the dodgy Clarke lathe we have here

I have the drilling stuff and the tube now, so when I get a spare minute...

Hey thanks Mike,

Yeah it did seem to get worse after I used the Platenclene

The silicone grease helps a lot ... so nearly there with the tool loading.

Also I did my first bona fide fix to the 'having to assist the z axis when starting' problem. It was the fastening of the vertical rack to the arm - the screws were a bit loose and the rack was disengaged from the pinion at the bottom of the z-axis travel. These ones here:

http://anagram.net/nuts/Versatronics/Rack%20fastenings.png

Obvious really...

Alan

p.s. the rack and pinion look very replaceable with slightly modified off-the shelf parts, and maybe with brass versions.

Hmmm, I having some difficulty getting the tools to load and release reliably. I've tried a little Platenclene on the o-ring spigot thing 'tool holder' and in the inside of the brass rings in the nozzles. Also spiyda's ball bearing tool position setting technique - these have helped quite a bit but I'm still not there. This machine has been sitting around not being used much for the last 7 years, if that makes a difference. The symptom is that the pickup sometimes happens a bit crooked and the nozzle doesn't go all the way onto the spigot and/or it goes through the nozzle release routine but doesn't. The latter is a bit fast, so I have not quite seen what's happening there.

Any thoughts?

Alan

OK a slightly different proposal for a possible cal grid position:

http://www.anagram.net/nuts/Versatronics/RVxS-grid%202.PNG

I'm thinking the whole thing is keyed for exact position on the 6mm dowels (which I just discovered come out and could be replaced for this purpose with longer ones). I think they might stick through the 7mm holes in the Cal04 bed and key the cal grid directly or whatever it is mounted on.

Spiyda: anything else about the cal you can think of? Does the cal routine look like it goes through all the accessible points on the grid? In what order?

Alan

BTW are there any problems / gotchas extracting the fixed camera / lighting or is that fine?

Alan

Ah eventually found the dodgy usb scope:

http://anagram.net/nuts/Versatronics/Nozzles/print%20vs%20original%20detail%201.png

http://anagram.net/nuts/Versatronics/Nozzles/print%20detail%20nozzle%201.png

Real one on the left, 3D print on the right. I deliberately put some tiny fillets on to make it go in and out of the rack easier. Have ordered some tiny drills...

>>> As far as the smoothness is concerned, as long as the camera sees it it as black, I don't think that will be a problem either.

It was the smoothness at the end of the nozzle I was bothered about, will try some v fine wet and dry.

>>> I have some brass tube that could be turned down, but I couldn't find any that was exactly the same both internally and externally..

Something like this may do it, by my calcs 0.2mm oversize on the OD and 0.2mm undersize on the ID:

http://www.tubestore.co.uk/ourshop/prod_2038041-127mm-12-Dia-x-250mm-Unpolished-Brass-Tube.html

may just fit anyway, 0.2mm isn't much .... or this, 1mm undersize on the ID:

https://www.metals4u.co.uk/brass/c2/tube/c30/12.7mm-x-1.6mm-(12-x-16swg)/p13496

Alan

Hey Spiyda,

>>> you are more than welcome to the parts I made / had made to take this further..  ( rotational calibration thingy, camera on mount  and  grid plate with slightly iffy grid print)

Thanks that would be great - but I'm hoping we might convince you to re-kindle some interest

I think you were nearly there - seems to me there's (a) just checking that the grid is accurately positioned where it's meant to be - if, in fact, that's required and (b) verifying that the resulting cal file is at least as accurate as the previous one.

I'm up for spending some time to finish this off - anyone else? Mike? Spiyda, any chance of finding some more enthusiasm / time for the last couple of tasks?

I've found a camera which seems to the job and have CADed-up Versatronics' mount design, so it would be easy to print it up. I have to get some stuff water-jetted over the next week or so, so could get some cal platforms cut for anyone who's interested.

Alan

>>> That looks like it's shifting nibbles of data at a time

Ah v good, I just twigged yesterday that the card is an 8-bit ISA card - so that makes sense. Also, I just noticed this:

http://anagram.net/nuts/Versatronics/Contents%20of%20ramp%20dot%20sys.png

Alan

OK here are the first prints:

http://www.anagram.net/nuts/Versatronics/Nozzles/Premium%20nylon%201.jpeg

http://www.anagram.net/nuts/Versatronics/Nozzles/Various%20materials%201.jpeg

- there are a few materials left to try but those were my best guesses.

On first look:
+ they dimensions are fairly accurate and spare brass bushes just push in nicely,
+ the detail is not as good as real machined ones - I was hoping it would be a little better,
+ the nozzles need drilling out - I guess I knew this up front
+ the tips of the nozzles may need some gentle smoothing,
+ they seem strong enough, but maybe not as strong as machined acetal.

I'll see if I can find my dodgy usb microscope tomorrow and post some more detailed images. Base cost (w/o shipping & taxes) varies from $3 to $8 p.u. depending on materials.

A.

Ah think I've figured out the camera positioning - the camera adaptor shown in the CAL05 model mounts direct to the brass bar on the z-axis. The camera is either 15mm or '17.17mm' diameter and is held with the two m3 grub screws. I was expecting it to be another WAT-704 but apparently not as those are 18mm diameter. Of course, it could be some camera + light rig. I think the slots on the CAL05 model, can be sensed by the R-axis zero sensor.

Something like this:

http://anagram.net/nuts/Versatronics/Possible%20camera%20mount.PNG

http://anagram.net/nuts/Versatronics/Camera%20adaptor%20orientation.PNG

Alan

OK, I don't know exactly what this means and it may well not all be correct, but here is a visualisation of the two cal matrices in my rvdata.dat with the tool etc positions from the same file. It is possible the cal matrices don't start at cartesian 0,0 (as my previous graphic) and it is possible that the rows/columns may be inverted - as I made a guess which may the matrices are serialised into the file. Anyway, FYI:

http://anagram.net/nuts/Versatronics/RV1S_152%20RBMAR07.png

The green lines are the 1st matrix and the red lines the 2nd - if indeed my guess about the matrix dimensions is correct. The overall shape of cal data looks bona fide, but who knows. There looks something fairly wrong in that the cal data spans the whole bed - which is clearly wrong as the arm can't actually do that - but that's as far as I've got so far. Constructive comments welcome.

Alan

Very Nice,

I'm up for some PCBs and happy to contribute if anyone is sending them out, otherwise I'll add them onto our next order here.

Alan