Machine Calibration

Started by basemetal, April 08, 2011, 11:52:10 AM

Previous topic - Next topic

basemetal

I see very few posts providing information on machine calibration and settings. Does anyone have documentation relating to varouse aspects of calibration. I have some information that i can post on certain aspects of calibration that i have managed to obtain and perhapse now is the time to pool our knowledge.

;D

Gopher

I can't see it hurting, I suppose one could even share FDR's or CDF's we're especially proud of

Mike

The only thing I've had to do was tweak some tool block positions, which was pretty easy with the cal mode of RV Place.
Some things will need jigs but I suspect that between us we could probably figure out the entire process to cal everything - fortunately all the software functionality appears to be included in RV Place. 
There are some things in the bunch of drawings I have that look like they may be cal tools (e.g. This Grid) - haven't had time to look at everything in detail or play with the software.

Maybe since Dave's now so busy selling Vivo's he'd care to share... ;)

basemetal

We have had to do a little more than tool position re-sets. Re-adjusting the fiducial camera and setting the OTF camera focus for example (including camera centering and getting the ratio set up correctly) as well as the fixed camera calibration. While the machine certainly placed well before the adjustments it certainly performed far, far better after the adjustments. Of course its also easy to mess up settings. I am also a little concerned that some of the responce to questions posted on this forum is perhapse a little misleading. Having said this i am also concerned that the training i got with my first RV4 was slightly flawed, especially in respect to the fiducial camera (we currently have four machines with two in everyday production).

The method that i was taught when setting up the initial fiducial marker positions on a new PCB was to use a small tool and obtain the fiducial mark position from the board by using the tool and NOT the fiducial camera. Of course when placing the parts one runs the fiducial setup automatically and this process uses the camera to find the center of the fiducial marker. Initially i was frustrated at having to apply board biases to components untill i noticed a pattern. If the fiducial camera x/y offset is out by only a little then when placing, the component placements will be distorted - imagine the board as a sheet of elastic and being stretched the same amount as the fiducial camera missalignment (sorry - imagine the board being fixed in space and the component placement position matrix being the elastic sheet). These can become big problems on large boards and large step and repeat boards. Once the x/y offset alignment was done on the fiducial camera - no more component biases. And the adjustmnet is so easy - special tools required, a blob of prestik.

So if you want to check the x/y offset settings on the fiducial camera put a blob of prestik on the PCB, use the small nozzle and make a very small indent on the prestik, then switch to camera and the cross hairs should align exactly with the indentation.

Anyone know the passwords for the calibration tabs in RV place and what each password does?

davidc

the fid cam is reffed agains the the head. ie the nozzle . the nozzle is the datum... so setting up is done with the nozzle not the cam as the fid cam offset in SW is then applied to the position of the nozzle. if you use the fid cam to set up then it appiles the offset again...
:)

set up with the nozzle. then you can use the fid cam ..  :)

davidc

yes i know what the passwords are...  :) i need some pcbs making for a job.. does any one here make them for free... ;)

davidc

Also Pauls comments allude to the offset of the Cam to the Head.. the point at which you align the head to the fid cam is crucial..

when you buy a machine at this price point you wont get a uniform performance re the relationship of the 2. So get it right in one place and its wrong in another. so you have to decide where you want it to line up.

Gopher

When the machine was new my board biases were 0. However I do recall errr Richard finding this surprising so I suppose my machine must have been unusually good. Sadly it's no longer the case.

I am sure plenty of people would build a few boards off the books in return for an equivalent favour. I am equally sure it's probably well within the grasp of those who code to grab those passwords from wherever they are stored.

Mike

Quote from: basemetal on April 12, 2011, 02:11:01 PM
Anyone know the passwords for the calibration tabs in RV place and what each password does?
tabs ? what tabs...?

I just noticed that replacing the place.dll (mentioned in this thread) from the standard install with the later one changes the cal menu behaviour :
with old DLL, clicking 'Factory' brings up a password dialog with P45 pre- entered, and you then get to the cal screen
with the new DLL, after entering the password it prompts to create new Dats, and if you say no it just returns back to the normal screen.
I'm guessing maybe the newer place.dll  needs different cal data? or maybe there is just a version check.

In addition to the 'P45' password in RV Place, there also appears to be 'versa' - is this what you were looking for?



basemetal

Yes, Mikes answer seems to point in the correct direction. On two machines running the same version of software i get different results as described in Mike's post. I have not checked the DLL version but will now. Having purchased one machine as a brand new VSMT (Versatronics) machine and the other as a completely refurbished and re-calibrated machine from the same supplier - purchased at a later date, it begs me to ask "what does refurbished and re-calibrated" mean if the calibration software is an old version?

I beg to differ to davidc comments. The x/y fiducial camera offset is the offset between the camera center and the nozzle center. As the arm moves about the workspace the angular relationship between the fiducial camera and nozzle changes - in fact between extremes of movement x and y will transpose as the angular position of the fiducial camera arround the nozzle changes by 90 degrees. If one looks at the error between nozzle tip position and camera center view point it is nowhear near the error that would be generated by transposing the x/y offset values. In any event to correct for this angular rotation error in software is trivial. Therefore it is obvious that the x/y offset is adjusted by the software to compensate for this angular rotation depending on where the arm is in the work area.

It therefore makes little difference where the offset is set up in the work area, although i have set up mine in the center of the table and about 6-8 inches forward of the rear PCB holder.

Also you should not consider the nozzle as the datum, this only confuses the issue. The only datum (or reference point) is a point fixed and immovable on the table itself, everything else is variable. If you think of a 2D array of points, a grid or matrix if you like, then everything else is on its own grid which can be adjusted and calibrated - even the PCB should be considered as being on its own grid (The fixed camera then becomes a simple x/y offset calibration).

Having purchased a new machine and always having to fiddle with biases for a few years was a real pain, now no more biases and its the best its ever been, a weekend of my time well spent.

when you buy a machine at this price point you wont get a uniform performance re the relationship of the 2. So get it right in one place and its wrong in another. Disagree, my fiducial to nozzle error accross a work area of about 200mm x 100m is now no more than 1 thou.

if you use the fid cam to set up then it appiles the offset again... Err no! I wil post a writen procedure for what we do but in the interim, set up in the normal way BUT switch to camera and align the first reference, then switch off the camera and the nozzle points to exactly the same spot as the camera was pointing, only then move to the 2nd reference and then switch to camera and set the reference point with the camera, once set with the camera, turn off the camera and the nozzle will point to the 2nd reference, then do the board heights. If the fiducial camera was set up correctly and in my case out by less than 1 thou, i can get far better resolution than using a tourch and small nozzle!

yes i know what the passwords are... So why not post them for the benifit of all? i need some pcbs making for a job.. does any one here make them for free... That depends, if it is for the benifit of the RV comunity then i am game. ;D


basemetal

Is there a file area where we can place information and documents? Drawings, pictures, documents and the like?

Mike

Quote from: basemetal on April 13, 2011, 08:50:44 AM
Is there a file area where we can place information and documents? Drawings, pictures, documents and the like?
For things specifically related to a post topic, attatch it to the post (you can attatch up to 4 files up to 2Mbytes each). Anything more general, email to me at mike @ whitewing.co.uk & I'll figure out somewhere sensible to put it

Gopher

Quotewhen you buy a machine at this price point you wont get a uniform performance re the relationship of the 2. So get it right in one place and its wrong in another. so you have to decide where you want it to line up.

An interesting statement, your new machine is at this price point, possibly even lower if you scale the original price to todays money, are you saying your new machine also offers a compromise on accuracy?