Toothed Drive Belts

[Sixexe]

I was considering what would make an RV machine go out of calibration and thought that the only parts that would have a major contributory factor would be the toothed drive belts wearing or stretching.  The (new to me, thank you Gopher) RV4s I have seems to be less accurate in making placements in the board area nearest to the fixed camera.  I have taken a look at the toothed drive belts driving the coaxial shafts and noticed that there is indeed damage to one of the belts which coincides with the positions of least accuracy.  Has anybody replaced the drive belts before or knows the specification for the belts or where to get them from?  I would have thought it essential to ensure the belts are in tip top condition before going to the trouble of trying to calibrate the machine.  If possible I would like to replace all 3 toothed belts with new.

[Mike]

I don't know anything about these belts, but wonder if there may be some 'running in' time for a new belt.
I expect these will be fairly standard parts - let us know how you get on finding replacements .

[phonoplug]

Another significant cause of placement inaccuracy is the cameras. No, not being funny, but if the centre of the camera isn't where it was when calibrated this will add an offset to everything.

If the fiducial camera, this can produce confusing offsets as they will vary between boards depending on the position of the fiducials on a given board, and produce rotational offsets.

[ianh]

Hi, I also have a problem with the fiducial camera location calibration.  Does anyone know how to calibrate this camera?

[Sixexe]

I have been taking some measurements and have come to the conclusion that the 2 main drive belts underneath the machine are 13mm wide 950mm long with a 2.5mm pitch, the top belt above the arm (housed under the cover with the vacuum pump) is 18mm wide 610mm long and has a 5mm pitch.  The lengths and pitches are standard but the widths not.  I have ordered a new set of  Synchroflex belts from Transmission Developments Co Ltd for £46 + carriage and VAT, they can make non standard widths to order.   I will let you know how things go when I receive them.  Has anyone thought about possible problems that I could have in removing the belts and not placing the new belts  in exactly the same position?  I can see issues with the top belt but not the bottom belts.  Regarding calibration of the fid cam, could this be done somehow by lining it up against the fixed cam and comparing the centres of the images?

[Mike]

I don't think it's as simple as that due to the geometry - the offset and direction of any error varies hugely across the placement areas due to the camera rotation.
I think it would be possible to figure out the calibration process, but nobody's got round to looking seriously at it yet - there are a few discussions in other threads, plus the drawing of the calibration grid, which is just a grid of circles at 0.5" intervals.
As far as is known, it images each dot in turn to measure the offset from the nominal position, and takes a few hours.
My approach would be to get the grid printed on a typesetter or photoplotter on a stable non-shiny material, mount it on a rigid plate (maybe 3.2mm PCB), and augment the camera lighting with a LED ring to reduce the effect of ambient light, plus a neutral density filter if needed to bring the exposure back to the right range.
Something that would need to be established is the absolute position of the grid on the placement area, and establishing a reference point.
I think it could probably all be figured out with some tinkering with the factory menu (after backing up the RVdata.dat of course!), and observing what gets written to the file with the grid in different positions. 
 

[Gopher]

I would concur, it must be possible to work out and be a fairly simple procedure once the process and zero point, calibration sheet location etc are deduced. However one needs to work it out the hard way after much backing up as Mike suggests unless you can track down an engineer with knowledge of the process.
In the case of this machine, I found the area in question to be consistent in its inaccuracy which meant I could compensate for it with some adjustment to the placement in RV Place or even by avoiding that area when placing the board. It also seemed to suggest (to me at least) it wasn't caused by a wear/damage issue as I would expect that to be very inconsistent?
I think it would be a very brave thing indeed to attempt a calibration without first getting the process working with the current drive belts. However you are clearly much braver than I as you have already removed covers and looked in places I never dared to while it was a critical machine.
Whatever the procedure may be I would suspect the machines existing lighting is adequate as the other 2 calibration routines allow you to tweak the vision parameters specifically to get a good image on Tool 5 and I would imagine a similar facility being available to get a good image of those little dots.
Is it possible, that this being Versatronics the calibration grid is actually nothing more than FR4 with loads of fids on it?

[Sixexe]

I hope you can all see the picture attached.  When a belt looks like this it's time for a new one!  I would be extremely supprised if this was not the cause of the problems.  I am not trying to attempt a calibration but replace the drive belts.  I was also commenting that I'm sure damaged/worn belts would put the calibration out and i guess it would be a waste of time calibrating the machine if the belts are not in good condition.

[Gopher]

Hmm yeah I see your point. I assume that is one of the thin belts I have never exposed, I actually can't think of a reason why a replacement should not drop in here, it's when the arm in involved things get more complicated I assume the geometry/location and positioning of switches and sensors start kicking in.

[Sixexe]

Yes this is one of the thinner belts under the machine, I can't see why replacing these should cause problems.  If I put the stepper motors back having say rotated it whilst out do you think this would cause a problem or does it not matter because it is a stepper motor and has no absolute position reference (hypothetical question, at the moment!).  The other belt is not as bad but I don't see the sense in changing just one.  

[Gopher]

This sounds like logical reasoning to me. Does everything else in there look straight flat and as it should be?

[Sixexe]

Yes everything else looks all tip top, a bit of dust and a few loose sections of drive belt tooth that have fallen off.  I have attached a view showing inside for those who would like to see but don't want to pull their machine apart.  The little PCB at the front has some opto sensors which detect slots in the edge of the belt guides on the pulley's.  The PCB has an old ST6 micro-controller OTP, I might check to see if the read protection has been enabled, if not I will read the code if I've got the time.

[spiyda]

Just a word of caution

As the machines age, the belts will die of old age..

If the belts get this bad, the exposed steel wires will cut into the pullies..

its only a couple of screws to take the front panel off and the belts can be inspected with a flashlight.

If they show signs of deterioration, its quicker and cheaper to change the belts than to try and source and then replace the pulley!

which by the way is
a 24 tooth T2.5 pulley
13.15mm width across the teeth
12mm bore
overall length (with the plain shaft for a grubscrew) of 22mm

[Mike]

Has anyone found a source/spec for the belts?

[spiyda]

I have fitted synchroflex belts for now

http://www.bearingshopuk.co.uk/synchroflex-t25-950mm-24922-p.asp

Synchroflex T2.5 950mm
Belt Width: 13mm

I still have the old pulley in until I source a new one,

the originals are marked     T2.5 950mm  0103

the new ones are marked     234 Synchroflex T2.5 950mm

Only time will tell how they perform

edit   ..

someone asked me how long it takes to change the belts..

bearing in mind I had not seen the underside of the machine before..
its not difficult, just a bit long winded...  it took about 3 hours..  
Now I've done it once, it would probably take about an hour next time

disconnect the cables under the rear
disconnect the upward looking camera and remove
remove the rear black steel bed cover to expose the tension adjustment screws
loosen the tension adjustment screws
tip the machine up (supporting the arm)
undo the bottom cover
cut off the cable ties that hold the loom to the case to give a bit of slack
undo the plate that has the connector socket on it and feed it through the base plate hole
remove the bottom cover
undo and remove the opto sensor board
undo the bracket the supports the wires across the pullies
(the camera wires then either need desoldering or some other means of disconnecting..  I desoldered)
undo and remove the stepper motors and old belts (you have to feed the electrics through the belt, like a crazy chinese puzzle))
fit the new belts (doing the opposite of removing the old ones !)
refit the motors and tension the belts..  nice and tight, you should be able to just twist the belt 45 degrees or so)
reassemble.. with new cable ties at the appropriate point

there is no need to adjust or align anything except the belt tension..

but the tool positions will need recalibrating as the toolholder has to be removed to remove the rear top cover plate