I'm going to disagree with you on that one. 40, 30, even maybe 20 years ago, sure, control comp was viable based on the slow typical machines and simple programming of the day. These days IMHO wear comp is mandatory; if you're doing anything else you're leaving money on the table and asking for problems. I won't go into detail, as this has been beat to death in other threads.

The way I do it is I program for the nominal sized tool (.1875 in your case), even if the tool measures a little small (which it usually does). While proving out the first part, If it's for a tight tolerance feature, I'll put +.002" in the diameter wear offset, run the program, then measure the feature. If it comes out .0025" over, for example, I subtract that amount (so wear offset shows -.0005") and rerun the tool. Usually comes out dead nuts, or maybe a couple tenths under since deflection force was lighter. Next part will usually be a couple tenths larger than the first prove-out part, since it has the full deflection force. Using skim passes (0.0000" stepover) takes out most deflection and improves surface finish, and I find modern, good quality cutters don't mind taking as little a cut as you like; they'll make shavings on a skim cut.

1. As mentioned, you MUST learn to use wear offsets. Figure the time to edit, repost, reload the program, vs hitting a couple buttons on the control. That time is money. Also, I bet since you're not using offsets, a lot of your parts are using much of their allowed tolerance. My clients have mentioned how happy they are that when given +/- .005" I'm generally within +/- .001 or better. In most cases that doesn't cost me any more, because I'm using wear offsets to get there. 2. Use template files. Instead of programming each part from scratch, especially if your parts are substantially similar, pull up a copy of the file for a similar part, import the new geometry, and reattach the toolpaths. Saves me a lot of time. 3. Use HSMAdvisor or similar for feeds and speeds; it will give you different numbers for various cutting conditions and stickouts, with the same tool in the same material. So you can do full width slotting safely, but also minimize cycle time on narrower cuts. 4. Develop a setup sheet template that all your setup guys can agree on. That way one guy can program, and hand the setup to another guy. That'll pay off the first sick day. After a while you'll learn which couple guys should be mostly dedicated programmers, and which should be mostly setup guys. 5. Look at the ratios of time utilization: machine idle while planning and programming, machine being setup, and machine running. If you have a lot of the first, that's a bottleneck that can be eliminated with dedicated programmers. If you have a lot of the second, you can improve that with standard setup types and fixturing, and keeping standard tools in the machine. Maybe some quick-point or zero-point mounting plates or vacuum fixturing depending on the parts. When you get to having a lot of the third, focus on cycle time reduction with tooling and programming style changes, and cutting parameters.

If you're going to spend in the ballpark of $350k on a mill, you can afford to build an airconditioned room around it.

How about a Hermle C250?

Depends on the scale of the work. I recently went to 64 for headroom, but was doing fine with 32, even with an array of 98 parts in a sheet and stock models at .01mm tolerance. But each part is small and simple compared to what many of you guys do.

For many controls, "79" = .0079 in inch mode, or .079 in metric mode. "79." = 79.0000 or 79.000 respectively.

The machine depends on the work. Are you going to do oil field parts and battleship cannons, or miniature medical device parts? Inconel and Titanium, or military grade billet aluminum? Flat plates with holes, turbines, or bone plates?

You could always use a macro to check for offset values outside of a certain range.

If the Okuma MU-4000V is too rich for you blood, consider the Okuma Genos M460V-5AX. Same quality and precision, but with fewer options, and should be head and shoulders above a Haas UMC.

I manage this kind of thing in ER collets all the time, but my holders and collets are in very good shape, and I carefully clean them every single use. Harvey's suggested parameters are for cutting in a straight line; you'll be making concave scoops in those slots, which increases the engagement angle.

I've always defined them as a bull, with the theoretical radius in the tool definition, but also using the accurate geometry as a custom profile. That way it calculates the toolpath as a bull, but makes the stock model accurate. Yes, the stock model takes a little longer, but not terribly.

Most people charge hourly, but your client may want a quote ahead of time. The rate will, of course, depend on your skill level.

We got the math chip on the old 8088 in 1981. Expanded memory card for 640K, external 20MB HDD and tape drive combo, CGA color graphics... Xerox Memory Writer was hooked up as a printer. The molds I'm doing lately are around 200 ops and pushing toward 32GB RAM on the stock models, but I'm putting all the parts for a mold in one file. Went to 64GB for some headroom.

So many ways to skin that cat. Face it with a large high feed mill, side-mill it, dynamic with a smaller cutter, etc. It'll depend on the material, your machine, and the tooling you have available.

Just for future reference, I cut both right and left hand threads in Ti and 17-4 H900 with standard right hand threadmills. Just need a slow entry feed if you're entering at the bottom. Also, three passes with approximately equal material cross-section, plus a skim pass.

Can you give a ballpark? Are we talking $50k or a million?

Yeah, the devil is in the details. Do you need .0005" true position from one end to the other?

I don't think it would fit in my CM-1... What other MTBs is the system compatible with?

https://www.youtube.com/watch?v=7hwhDz6Sdg4

You can get an AR-10 with optics for less than that.

That's going to be a pretty expensive bow.

Use a full radius slotting cutter and do it on a 3 axis mill. There could technically be a tiny bit of deviation from the model by the sweep of the tooth, but not enough to notice for this application. 4 axis if you want to tip over and get that sidehole in the first operation.

Far too many third party solutions duct taped together. It severely hampers control over the product, and customization capabilities. It's time for a clean sheet ground up build. But that would take money and effort, so we'll never see it.

Not sure the top end, but he wasn't squirming at the talk of a $550K machine.