I hired a new guy and he insists upon doing one DCC alignment and never doing another alignment, he lets the feature control frame control the alignment for him. I don't think that is a good practice, does anyone have an opinion on this?
I don't do more than one DCC alignment very often (iterative though).
I think you need to provide a lot more information about what is going on, and what you think is wrong and what you think is right, before anyone can really begin to answer your question properly.
I have a lot of programs where a DCC iterative is all I need and then I'll use XACT to build my FCFs (just like your new guy).... but if you've got funky geometry going on you may need additional alignments (which is what you're thinking and in a lot of cases you're right).
I'm with Vinnie, I'd really like to see a program you're concerned with.
I change alignments like I change my undies (weekly, lol, j/k).
When I program, I like to make the routine as robust as possible, and since I work with injection molded plastic, we could have stuff inadvertently bent or a sink or short that's over .100" from nominal which would normally cause a stall... I program interim translations rotations and levels any time I project from part xyz zero more than about an inch to make sure routine will robustly run.
At end of routine, I create a final alignment per print, and output all dim's. Putting the dims at the end also keeps routines compartmentalized and visually capable for the minions to understand where to look for the "red" (out of tolerance).
We make work instructions for each part number with the setup instructions for standoffs and mounting, so everything is the same each time. As a result, all our programs are 100% DCC, including the alignment.
We machine extrusions. Most of the time, the base material has more tolerance than the machined features. I look at as a little finesse is needed sometimes. How about a milled pocket that is super shallow and the base material is allowed to more more than the depth tol ? I would check the depth, 0.0 on the bottom of the feature and then move my probe "away" from the surface to check the width/length. I will never stall out and rub the bottom of the feature when checking the width. I can come up with hundreds of examples like this. Do I have program with 1 manual and 1 DCC align ? Of course. I also have programs with 75 alignments or more that may be 45+ minutes long.
Some of these parts have up to 20 different datum structures Primaries in X,Y and Z. We work in 5 of the 6 working planes. Just think that one should always level to the primary working plane. Thanks for your help.
I'd like to redact my previous statement.
Most of the time we measure parts (hydraulic valve spools, yay cylinders!) I create an alignment for every feature.
I have to align to the part itself, and I do this by zeroing on a circle, taking 4 more and level/zeroing to a cylinder, zero the Y, and zero W (the angle of the rotary table).
Each of these are done in seperate alignments, and then one final alignment to create the "Full DCC Alignment". This has to be done before any measurements can take place, and for most of my specific use cases, every 30° or so. So there are plenty of alignments going on, I just like that for posterity. Some of the tolerances are about ± .0001", with approximately 75 features with 3 to 5 dimensions each.
While this takes a long time, in my opinion
alignments are the most important aspect of metrology. I personally don't care if a part takes 5 minutes or 5 hours, as long as I'm measuring it right!
I hired a new guy and he insists upon doing one DCC alignment and never doing another alignment, he lets the feature control frame control the alignment for him. I don't think that is a good practice, does anyone have an opinion on this?
depends on the part. I assume he is using a readpoint at the beginning, then doing a dcc alignment and then continuing with the rest of the part and using the features picked up along the program to create his datums.
it's how I roll. as long as the features for the dcc are cleanly hit, and the machine knows the orientation of the part on the table, I haven't had any issues doing it this way.
the only time I ever do more than one alignment is if I'm using iterative, or if the part is so horribly off it doesn't come close to the model.
If
all dimensions are FCF (no Legacy), and the part is
reasonably correct, you only
need one DCC alignment to drive around and collect the hits. You might need/want more if you use Legacy dimensions (to match the DRF), if the part is bad (local alignments to be able to collect the points - relative measure might be an alternative), if the tolerances are really tight (to minimize cosine error), if you like to see coordinates matching a certain alignment, and so on.
