Hi Karl,
Your predicament must have struck a chord because I was dreaming about
answering your plea last night. Lets see if my answer makes sense while I'm
awake. You didn't say whether your more accurate parts were located near the
center of the vat and the less accurate ones were towards the outer edges.
What I know about SLA accuracy;
One of the first things that comes to mind would be an out of round beam
condition. Either laser problems or mirror clipping. You could check for
this, first by looking at the xy ratio, 1 is ideal. Then, wearing your laser
safety goggles of course, by using the "move beam" option in the beam
routine and program the x,y to the extreme locations x=0
y= 65,536 etc. and looking at the beam shape projected on a piece of card
stock. An out of round beam would produce worse curing profiles the farther
from the center of the scan region.
Another thing I have noticed is the need to run test parts after a PM. My
guess is that slight changes in the resin level as a result of maintenance
can change the so-called shrink factors. I just make some sticks nine inches
long (250) length and typical wall thickness for my application, in each
orientation, in the center and near the edges and measure the results to
determine the actual compensation required. Christmas trees bear little
resemblance to any actual geometry I might be building so are of limited
use.
Another possibility;
Ed Gargulio, bless his soul, used to explain the workings of the SLA in
great detail. One of the things he mentioned was that due to the scanner
design, one of the vectors, I forget which X or Y can't scan a straight
line, but scans an arc in the absence of compensation. Therefore one of the
reasons for calibration. To characterize this non linear distortion and
correct for it at run time. If your cal file gets corrupted all accuracy
bets are off. Usually your FE will have made a back up of this file after
the last calibration. This brings up the point that due to the point source
scanning your most accurate parts will be those located near the center of
the scan area. The farther off axis you get, i.e.. towards the edges of the
vat the less likely you are to hit the target. With the beam that is. 3D
presented data at one time that showed a more accurate user part made on a
500 as opposed to a 250. That makes sense due to the longer focal length,
hence less off axis scanning and better scanning system. If they had done
the obvious and made a user part scaled to 500 capacity, 19 inches rather
than 9 inches etc. It would have shown that it makes a less accurate large
part.
Another possible source of inaccurate scan regions;
That would be some schmutz(gunk) on brewster window. Fairly easy to see and
clean up. Could be splashed up from the vat or a big piece of dust. By now I
hope you have gotten lots of suggestions and are on your way to making great
parts. I'd love to hear what is takes to resolve this problem.
Andy Scott
Lockheed Martin Aero Sys
> ----------
> From: KDenton@williams-int.com[SMTP:KDenton@williams-int.com]
> Sent: Tuesday, September 07, 1999 10:19 AM
> To: rp-ml@bart.lpt.fi
> Subject: Build Problems with SLA and Confused!
>
> Hello all,
>
> Yes I'm back on the list and in need of help!
>
> With out describing the parts in full this may be difficult but I'm
> hopeful
> that someone will have an answer because its completely baffled me!
>
> I have been building a part that looks like a turbine and the orientation
> is
> such that the build lines are in the direction of air flow. This provides
> us with the best possible looking blade and the least amount of cleanup
> both
> with regard to supports and general part cleaning. The parts are small
> and
> we are building them on an SLA-500 using SL5180 and QuickCast 2.0. The
> room
> that the equipment is in is climate controlled such that the temperature
> has
> not deviated from 72 degrees F and the humidity has not deviated from 32%
> in the fours years that the lab has been in operation. Resin temperature
> is
> at 31 degrees C. When the first set of these patterns were built in April
> and May of this year they were within .002" of tolerance and we satisfied
> the customer (engineering) with the required number of castings. Knowing
> that the part would be required again in several months I placed the build
> files ( .bff) on a separate hard drive for later use. Two weeks ago I was
> asked to rebuild several patters and now they are .020" to .030" out of
> tolerance. In one case the part is .020" out on one dimension and right
> on
> another dimension. One part comes in large while another comes in small.
> Literally the only thing that has changes is a new laser and all other
> parts
> seem to come in. These parts by the way are segments and when enough are
> complete they are assembled to a fixture. At this point I am thinking
> it's
> the human variable and not the machine that is causing things to go crazy.
> Any all comments would help.
>
> Sincerely,
>
> Karl Denton
> Lead Engineer
> Williams International
>
> For more information about the rp-ml, see http://ltk.hut.fi/rp-ml/
>
For more information about the rp-ml, see http://ltk.hut.fi/rp-ml/
This archive was generated by hypermail 2.1.2 : Tue Jun 05 2001 - 22:52:39 EEST