How Precise is my Haas?

[@johnyoungquist6540]

Years ago I explored machine accuracy with a laser interferometer that measures within a millionth. I had to build a device to calculate the speed of light based on air pressure, temperature and humidity to get to better than 1 ppm. This device was interfaced with the interferometer directly. The speed of light in air is about 280 ppm slower than in a vacuum. I had a high end Matsuura milling machine with box ways. It was spec'd to repeat within 80 millionths, it was often much better often 10 millionths. I was amazed. But how well did it position? That was determined by mostly lead screw temperature and position in the travel. Of course the temperature of everything associated with the leadscrew mount contributed too. At one end of the travel near the beginning of the leadscrew it has little effect where as at the end it has a much larger effect. With a coefficient for steel of something near 12 ppm/deg C a 12 inch screw would change a thou in only 7 degrees of temperature change. what about a 40 inch screw? The lead screw starts at ambient but warms up a lot with use. When it does it is like a scale factor change. It might even not be uniform in temperature along its length. Some exotic machine liquid cool their leadscrews to tame this problem. Hardinge on their high precision lathes handled headstock thermal growth with a linear scale to measure it and put a correction back in to the control. Some mills and many wire EDMs and CNC grinders use linear glass scales to correct leadscrew induced errors. So a mill that repeats to 10 millionths might be out a couple of thou when it warms up. I replaced the Matsuura mill with a new one with ball bearing rails. It wasn't anywhere near as good as the old one. We haven't even talked about leadscrew correction tables which some controls have. I spent days tuning the leadscrews which was a waste of time. While this topic goes well beyond the space here there is one other issue that is everything is elastic. The cutter, spindle, work piece, vice....etc. So the machine only positions correctly at zero load. This what so called spring cuts are for. But light cuts that are under the effective radius of the cutting edge rub more than cut. This means a worn cutter will cut a different size while its sharpness diminishes even though its diameter doesn't really change that much. The more you understand this the less likely you are to believe precise machining could ever be done.

[@colehawkins6623]

Great video. Just ordered my first HAAS. Please keep them coming.

[@ebrewste]

That's a heck of a hobby machine :) and a nice video. Some comments: 2:54 - To be pedantic, the z channel will let you find a home position to 1/2 of an a/b quadrature pulse width (90 degrees of a channel period). The rising and falling edge should always fall in the same 90 degrees of a quadrature cycle (different for each side of the pulse). The z channel can be gated to rise with the quadrature edge, though I never understood why one would actually want that. z channels can be different lengths. The one you have shown in a full quadrature cycle long. I'm used to z channels are that are 1/4 of a quadrature cycle long, though they do the same thing, at the end of the day -- tell you where you are in a full encoder rotation. 5:00 - I'm surprised that you could move it so far. You could hear the servos fighting you. They don't appear to be using an integral at any meaningful bandwidth -- that's honestly pretty soft. For anyone interested, the integral will drive the servo encoder to a zero position error, within its resolution, leaving only mechanical error. Linear / glass scales will drive the linear position error to zero, at the linear encoder, leaving less mechanical error (no screw error, etc.).

[@ChrisSmithCenterline]

Great video John. As you noted towards the end a Renishaw does indeed take it to the next level. Using wear compensation via the Renishaw I'm easley able to keep tolerances of 2-5 tenths. Looking forward to your next video.

[@JohnSL]

Thanks for the info! Given the price for a Renishaw, it's still not on my short list...

[@moki123g]

@@JohnSL I realize that this is a little old but I just wanted to let you know that a probe is super handy not just for the accuracy but being able to pick up locations on parts. Once you have them you will never go back. Totally worth the money.

[@rustedfriend]

Always interesting John. I'm sure this will come in handy. I'm starting to look at machining my own molds but on a much less fancy cnc :P

[@JohnSL]

I started with a Taig CNC and made a bunch of molds with that. You don't need a machine like mine to make good molds.

[@roylucas1027]

Good information John. I'm always happy to attain .001 accuracy.

[@JohnSL]

I was always curious. Given that this is a machine design specifically for "precise" work, I was wondering how precise.

[@scottwatrous]

I would say the Renishaw system as you mention at the end, is good, but in practice (for me, as a glorified Hobbyist) it's not really much better than about .0005. It'll certainly feed the machine outputs down to the tenth, but in practice there's some wiggle on that. You can probe the same surface 5 times and get different results within .0003 or so. For tool probing it's the same. Theoretically it can tell me the average diameter of a tool between the flutes, but, when I actually put cutter comp into practice it's still liable to be a little generous with the diameter. I'm sure half of that comes down to runout and the other half the probing, so using a much higher grade of tool holding than my mid-range ER collets would probably bump it tighter and then in production knowing when to check a part and add/subtract a few tenths to the probed diameter would get you to where the inherent qualities of the machine is a bigger concern.

[@JohnSL]

It's good to know. I think the main advantage for me of a probe would be tool breakage detection.

[@michaelkelly3158]

I've always found the probe to be within microns of repeatability - usually 1-3 microns! It does need some setting up though.

[@TroubledTimes2024]

That was interesting John, have never done that, now that spring is here and I can get out in the shop (not much colder then a uninsulated steel bldg) will test how my CNC fair's.

[@JohnSL]

I'm so glad my shop is completely insulated with a mini split. So it's comfortable all year around. Plus, because it's attached to the back of our attached garage, I can get there without getting wet or cold. Hope you get to use your machine now that it's warming up.

[@Ujeb08]

I would replace that Fowler mic with either a Starrett or Mitstuyo for good accuracy and repeatability.

[@DragonflyEngineering]

A very nice machine John

[@JohnSL]

Thanks Dave.

[@kenarij]

Just a suggestion, don't leave tools in the spindle unless cutting. No need to keep load on the pull stud/drawbar springs. Also, put a dial indicator on your "probe". I self calibrate/adjust mine using a dial indicator and a gage block. Definitely always put your probe in the spind oriented position. Thanks for posting your video, are those BT 30 holders?

[@JohnSL]

Thanks. I've now started to do just this.

[@ВладимирАлекс-г3с]

Спасибо!

[@Badge124]

I'm curious what the spindle runout is on this machine. Have you checked this at all?

[@mrechbreger]

Is this machine belt driven or is there a direct ATC spindle installed? Can you show some pictures from the top inside? For such small machines I'm still in favour of doing them myself (since I already built a relatively small one 40x20 which can also cut S45C steel I'm aware of the effort.

[@JohnSL]

It's belt driven, and showing a photo of the insides would require me to take some panels off. This is a Haas OM-2A, which is basically the same as the newer CM-1.

[@MrKurdishFreak]

I need to do something like this with my emco and compare to your haas.

[@max_eley]

Hello, good video, however please remove all tools from the spindle when not in use, they can break the toolchanger plate right in half, I've seen it many times, it happens when the tool is left in there for a period of time long enough for it to lock up in the spindle. Temperature and drawbar force ect.

[@JohnSL]

Thank. I am now doing that. I guess I've been lucky so far, but don't want to push my luck.

[@dass1333]

All I know is your machines are far more precise than mine. Glad my work is just art work with tolerance of whateverths.

[@JohnSL]

One thing I keep hearing is that precision is more about the machinist than the machine. You can get really accurate parts with a manual machine that has huge backlash if you know what you're doing. I was able to do some really detailed work with my Taig CNC (using end mills down to 0.01 inches in diameter). That being said, it's nice having a "tight" machine.

[@dass1333]

@@JohnSL the thing I found out was lash is real bad for climb milling. Conventional mill keep your bits. You can file just fine.

[@ianpatton8903]

Your taking a lot of your measurements on the upper edge of the stock where there could be a burr. The machine may be even more accurate then you think.

[@Steelcrafted]

I said the same thing...each time he touched off the X axis, the haimer was at a different Z location....def introducing uncertainty into your data....