How long does it take you to program?

Let's say I'm in charge of this project
1: Make sure that the drawing is the latest version;
2: Request the latest version of 3D files;
3: Confirm the product process and prepare the corresponding measurement program according to the "DFM" file provided by the project team;
4: Design measurement fixtures for the corresponding measurement procedures (familiar with the operation of 3D design software, e.g. SW, NX, etc.);
5: Prepare the measurement program offline; (please execute "collision detection" after the measurement program is completed).
The above is the preparation work before the development of this project. Of course, after the "DFM" file is modified, the measurement program should be updated simultaneously.
The project of 150 sizes can be completed in about 2 days, but this does not include the time for modifying the measurement program.

Simple adjustments to the measurement program may take about 4 hours, depending on the extent of the process changes.

hi,

Example 1:
for a rotor shaft of a wind turbine (About 3m long). Simple A-B pattern, including drilling pattern, grooves and other details, around 75 positions I need about 2 hours depending on whether a 3D model is available or not. Runs smoothly the first time

Example 2:
High pressure pump housing with approx. 300 positions, 3D model available, 7 drawings, datums from A to G, a cube with approx. 500mm x 500mm. including test run on the machine About 8-12 hours, based on how often the boss looks over your shoulder.

Example 3:
Prototype drill bit for an oil drill, 1.2m long, approx. 200 positions, complicated freeform contour, 12 different sized pockets in a variety of angles, 3D model available, about 6-8 hours including test run.

---------------
I have very good experience with collision detection. I almost never have any problems during the test run.

I don't use the clearance cube, just clearance plane

I use prepared standard programs for all axes so I just have to see which direction I want to start from

We have prepared enough Probes, I don't have to build any new ones in the examples

We have enough standard prisms and fixtures, including the 3D models for them, which saves a lot of time when collision detection knows where the fixture is

---------------

Unfortunately, it happens all too often that a drawing is not clear or the required geometric tolerances simply do not make sense.

Asking the customer or making the associated changes that need to be implemented take a lot of additional time

Immensely helpful, thank you. Especially the last comment. Our experiences are pretty similar.

Sounds like you’re a pretty fast programmer! Legacy or GeoTol? Hmmm maybe I should give clearance planes a shot again. I use move points. Not a fan of clearance cubes at all. Do clearance planes give less false collisions (compared to clearance cubes) when running collision detection?

Very few people can really concentrate on the same thing for hours, when I say 8 hours that's usually 2 shifts for one person.

So actually not really faster than Ocean Zhu example. As I said, we have enough standard fixtures, programs and probes.

All the header data is copied from a database into the measurement program. I can start with the first details in minutes, which saves a lot of time

I use 99.9% GeoTol

clearance planes work great. Of course, there are a few things you have to take into account, but I personally find it very reliable

I can hardly remember whether I ever received a false collision,

Except for one thing: star probs always give false collisions when you change the tip without clearence move near a wall,

You learn this exception pretty quickly, so it doesn't matter

Getting the question of how long it will take to program a measuring routine is always rough.  People often have the impression that CMM programming should be faster than it actually is.  It is especially rough when you are the only programmer at a company, and you are getting pressure from all directions to have an unreasonable turnaround.  Hopefully you are in a good work environment and people are giving you the benefit of the doubt. 

I would say a reasonable size part with 150 dimensions to report would take somewhere between 4 to 24 work hours for me to program.  I will often give someone a wide range like that and then tell them I’ll look over the print and get back to them with a more precise time estimate.  The wide range in that estimate depends completely on how complicated the part is to fixture and how tricky it is to get at things with the probe(s).  I have had some parts that looked relatively simple end up taking a whole shift to fully take in the inspection requirements and figure out how the heck to fixture the part and organize a measuring approach that accounts for as many potential problems as I can predict. 

As for your other question.  More often than not my programs run perfectly the first time.  In real time crunches I have even started long 8-10 hour measuring routines before I left for the day and came back the next morning to find that the program successfully ran through the night – It’s pretty sweet feeling when that actually works out!  As Henniger123 says, collision detection is quite reliable, and I use it often.  However, I also have the luxury of being one of many CMM programmers where I work, and I often get to write programs a month or two before the part ever arrives in the QC lab – I basically get as much time as I want to check and double check my programs.  It is not a matter of being perfect, it is just a matter of being diligent in finding and fixing my mistakes beforehand. 

I have also worked in more fast-paced environments where I would come into work and find a new part waiting on my desk for a new CMM program.  In that case I would cut more corners when creating a program and fix issues at the CMM during the prove out.  It is nice to have it run perfectly the first time, but the extra care it takes to make that happen can sometimes be more trouble than it is worth. 

Henniger123 said:

Very few people can really concentrate on the same thing for hours, when I say 8 hours that's usually 2 shifts for one person.

So true.  I know I can't run on all cylinders all day every day.  And, it seems like there are always interruptions.  

Let's assume that the part has been machined on all sides too, and that you won't be interrupted whilst programming. (oh how I miss machined parts!)
I guess also assume you can check it all with one setup... anyhoo... 

My typical process: (I program exclusively offline)
1. Look over the drawing and start figuring out if I can rest against any datums. Consult feature sizes to find the largest probe that will check the most features.
2. Open up the model and spin it around on the computer screen for about 30 minutes thinking about how to fixture it (probably makes me look like very silly, but has saved me enough times I keep doing it)
3. Load up the model of my CMM's table and import the part model, then open my tooling library and build it all in the computer. Since the tooling is already "on the table" I know everything will fit perfectly.
4. Armed with my part/fixture/table combo model I start it up in PCDMIS and get to programming. Figure out the alignment, then slam out a bunch of features, run collision detection, fix anything that needs it, repeat and repeat until its finished. Then start dimensioning. Game soundtracks and caffeine help with this process.

150 dimensions, probably 1.5x that many features would probably take 2-3 days with how I do things. Due to starting the program to include all the tooling and table I am very confident when I run the first time that there won't be any issues proving out the program. I keep a note pad and jot down any adjustments that need made, normally adjusting prehit/retracts to increase speed. 

When I worked at a machine shop and most of my parts could fit in your hand, I used clearance planes and avoidance moves exclusively.
Now I work mostly with large fabricated parts, my machine is 26' long so clearance planes have been replaced with move points and avoidance moves. Speed is less of a concern, the DEA isn't going anywhere very fast, at least compared to your average sized CMM. (also takes longer to prove out programs now, have to get up and walk around inside the machine to follow the probe around)

Having a library that includes all your available tooling really helps out the process, they do not need to be fancy to be effective. For my current machine we use a tooling table, for my old machine I used the machine to create a DXF file of the actual stone with tooling holes and travel limits, then used it to model a plate perfectly matched the machine.

This is one that I did today to update a program to this decade. Everything including the C-clamps I may or may not use are included and pass collision detection. (part is model hidden)

We aren't allowed to listen to any music at my job. You are so lucky you have that option.

You can't use headphones while sitting at a computer?