[DLawrence]

I have started mounting a Mitutoyo 715 linear scale on the x axis of my 1100 thinking it might be usefull ..

Just saw some in Bob Warfields CNCcookbook series here he says:

"Another thing to be aware of is that a lot of machine control software, including Mach 3, has a feature known as “ballscrew mapping”. This feature lets you measure the true position reached at various points along the ballscrew and use it to compensate for errors in the ballscrew. This function works very well, and should be taken advantage of if you have the means to accurately measure the deviations. This alone is a good reason to install an inexpensive DRO at least temporarily on your machine until you can get the compensation tables calibrated."

I searched for "mapping" in the Tormach PCNC and pathpilot sections and didnt find any post son this......Am I missing some posts or is this yet to be considered here?

and does PP presently have provisons available for ballscrew mapping....

Probably later looking at using the linear scale's absolute positioning ability to improve over mechanical homing switches...?

at some point

Thanks
Dave Lawrence

[PCW_MESA]

Linuxcnc has screw error compensation (which includes position relative backlash compensation)
available so I would guess that its still there in PP

INI Configuration
(see the COMP_FILE section)

[MichaelHenry]

What sort of accuracy target are you shooting for?

[DLawrence]

What sort of accuracy target are you shooting for?

Just at one place I need plus/minus .0005

consistantly.

Plan to put a DRO scale also on the Y

also although Y is not nearly as critical.

Possibly a m1 and rehome X right before

this final most close tolerance move would help?

If I can contrive to run this job it will be by far

the best thing I have ever done money wise and an ongoing



endeavor.



Thanks
Dave Lawrence

[SCzEngrgGroup]

I hope you're in a tightly temperature-controlled shop. Thermal expansion alone can easily push you well outside that tolerance...

Regards,
Ray L.

[MichaelHenry]

What Ray said...

[keen]

Of course the big problems with lower spec machines is at the 'turn around' that's when the direction change play in the ballscrew/nut and thrust bearings and other weak links in the chain from command to movement all add up to what we often call backlash - difficult to get below around 0.02 - 0.03 mm total error on machines like this. But once the slide is heading off on its travel, the 1100 and 770 ballscrew specs are high enough to keep the machine quite close given the real world errors of thermal expansion.

Of course not many of us have actually checked that they are on spec - mounting a DRO should give a quick check and I would be keen to here the results. If the ballscrews are to spec though, I doubt you would need to go to the trouble of mapping compensation.

Keep us posted! Keen

[mountaindew]

What Keen, Mike and ray said!
I don't know about adding a DRO to my Tormach. But having a Tormach has me wanting to add a DRO to all the other machines in my shop.

[sansbury]

Whether this is good business depends a lot on how much scrap costs. You can hit super tight tolerances with lots of lather, rinse, repeat and rejecting half the parts you make. If a part is a few minutes of machine time and 6061, no worries. If a scrapped part is a hundred bucks of time and or material, then your yield needs to be a lot higher. Personally, I'd be pretty wary of tolerances like that, but if the money is good enough to allow you to screw up a few times it might be a good learning experience.

[Don Clement]

I hope you're in a tightly temperature-controlled shop. Thermal expansion alone can easily push you well outside that tolerance...

Regards,
Ray L.

The solution would be to monitor the ambient shop temperature and apply a correction factor in software to compensate for temperature changes.

[MichaelHenry]

The solution would be to monitor the ambient shop temperature and apply a correction factor in software to compensate for temperature changes.

Don - Mightn't the work, fixture, and machine temperatures all be sufficiently different from ambient temperature, or even each other to introduce an error that could be greater than the tolerance??

[sansbury]

Here's my take (with a little poetic license for hyperbole):

+/- .005" tolerances you can almost always hit the first time just by not doing anything too stupid. Even if the vise isn't super-square, machining strategy is whatever CAM spits out the first time, if you basically know what you're doing the machine is intrinsically accurate enough that you'll hit this or better.

+/- .001" means your fixtures and tools need to be set up carefully. It might require you to know how part geometry, material, machining strategy, and workholding interact. You are more likely to need to make a few and tweak things to get where you want, though certain things like simple XYZ dimensions you can often get the first time if your machine isn't too loose or out of tram.

+/- .0005" means setting up everything as close to perfect as you can get. It will probably require a number of parts and hand-tweaks to the program to dial the finished part in. Everything matters--switching from a carbide to a HSS endmill can throw you out of tolerance, or how tight you tighten the vise or fixture. And don't use your caliper to inspect the parts--it's likely to be off by 2-5x the tolerance.

+/- .0001" means you need to stop drinking whiskey with breakfast, or if you're not, start. If you're not sure, it probably means you don't even know how to consistently measure parts at that level of precision.

In the context of doing hobby work, there's no reason why someone shouldn't try to make parts to better-than-.001" tolerances if they take pleasure in doing that. It's satisfying to take a part off the machine, mic it, and see it come out dead on the money. However, if you're talking about taking on paid work, especially where someone is offering to pay well for tight tolerances, you want to be careful if doing something for the first time. There's a rule of thumb that each decimal point of precision multiplies the cost by ten, and that's especially true to the right of .010". IMHO with our machines, .0005" is where it starts getting significantly harder, though it is still doable.

If the money can pay for the time and materials to make a half dozen pieces of scrap and get it right on the seventh time and the job is done, then why not. If you need to make two dozen and the first or second needs to be right because it's expensive material or six hours of machining, then you're setting yourself up for a high-wire act.

[Don Clement]

Don - Mightn't the work, fixture, and machine temperatures all be sufficiently different from ambient temperature, or even each other to introduce an error that could be greater than the tolerance??

That could be true. I know the method of correcting for dimensional changes due to temperature works for telescope focal length changes by monitoring some parameter (temperature, strain gauge output) measuring the changes in focal length over a parameter range and then automatically applying a correction to the focuser to compensate.

[Spinnetti]

if your parts are small, you only care about the local region in your machine envelope that you are cutting in and the mapping won't do much unless you need super accuracy for a given part over the whole travel of the machine, and you also control temp and humidity as well as have a machine warm up cycle that is identical for each run, along with identical setups, minimum tool stickout and head height among other things. There's lots of variables to control and if you get all the ones that are possible, you can get pretty high precision (not to be confused with accuracy which is a whole 'nuther animal)

[adamvs]

What Ray said ...: Ferrous stuff changes at about 6 millionths/ degree F / per inch ( aluminum about twice that ). So if you've got a 20 degree change over the day it adds up about .0022 over the 18 inches of travel on X. If you have two or three vices on your table, that could mess with things. What would be cool tho, is to use your DRO to create a closed loop. Don't know if Linux supports.

[PCW_MESA]

LinuxCNC does support closing the loop of a stepper motor or step/dir controlled
servo motor with a linear scale.

Basically what you do is set the step generation hardware
(or software for parallel port systems) to run in velocity mode
rather than position mode and then setup a PID component
in HAL to close the position loop with feedback from the scale.

Very low backlash is a requirement for loop stability.

[DLawrence]

From CNCCookbook ultimate build section 3

" Glass Scale functions: High end machines will often add glass scales, the same kind that are found on manual machine DRO’s, to the rotary encoders. Such scales tell the machine exactly what the actual axis position is. This is often an expensive option, but it can greatly increase the accuracy of a machine. Because the glass scale is measuring the actual position of the axis, and not just the ballscrew shaft, it can compensate for thermal expansion, errors in the leadscrew thread, errors in the timing belt system driving the ballscrew, backlash, and so on. There are complications associated with doing this. For example, these scales typically cannot be read fast enough that they can substitute for the rotary encoders, so the control system needs both kinds of sensors and the glass scales are only used at various test points during a commanded move. Since there are two sensors, sophisticated software is needed to be sure the two don’t “fight” with each other over the motor control causing oscillations and other undesirable behavior. The vast majority of commercial CNC machines do not have glass scales due to the extra cost and complexity, so we’ll rule them out of a hobby class system, even an “ultimate” hobby class system up front. That’s not to say it is impossible, just that it’s starting to get into some very rarified air and there are likely other problems to be solved before the weak link in a machine is any lack of glass scales. As an aside, you can install a set of glass scales and use them to map your leadscrews and then provide a compensation map back to your control software. This will improve your accuracy and can be helpful, if a lot of work to deal with on a one-time basis."

When I Just casually watch both DRO's.......... Mitutoyo and Pathpilot..... without yet trying to "map" indicates the PP system falls behind 2 or 3 or 5 tenths often, and occasionally1.5 thou but

later catches up.

It seems like even using the add on scale somehow to replace the mechanical homing switches would be a really nice start start for a mapping project,



For the mapping to be meaningfull , wouldnt the "map " would have to be repeatable...several times...... .

Thanks
Dave Lawrence

[rowbare]

What are the customer's acceptance criteria? How are they going to measure the finished part? If you are dealing in sub-thou feature placement, you really need to know this otherwise you might as well just take your new stock and put it straight into the scrap bin and save yourself the time and cost of machining it. At the very least, you need to know at what temperature the part will be when it is measured and you should have a sense of the conditions under which the part will be measured.

As well, does the drawing even make sense? How does the tolerance on the spacing of the features compare to the tolerance of the features themselves? For example, if a drawing calls out two 1/2" drilled holes 4.000 +- .0005" apart, you can probably hit the spec with your drill press since the tolerance of a drilled hole that size is probably in the .005" range. In this case, the spacing tolerance is lost inside the feature tolerance.

The devil is in the details.

bob

[DLawrence]

This is a vertical lathe style turned part.........., a plain spur gear blank .712 dia +- .001 and .230 thick also +- .001 Its from 1015 steel

With 2.5 inches of stock protruding from the collet I was getting about .0015 taper which I could correct for.

Hitting The cutoff thickness might be solved by adjusting the vertical distance between the finishing tool and cutoff tool to .230 ..... so the Z axis is not moved between the facing cut and the cutoff.

Getting the x axis DRO mounted took a lot longer than planned for......first had mounted it on the column side then started over and mounted on

operator side, spaced outside to leave all origonal limit switch setup.

Although the Mitutoyo 715 "Absolute " scale isn't yet set up for directly replacing the mechanical homing switch , I plan to use it for establishing a

more consistant homing reference, by repeatedly referencing(homing) the axis ,,,,,perhaps cleaning the roller, or turning it... or by tapping the mechanical stop

block till the refrencing/ homing routine is right on with the liner scale zero.

Although this procedure itself wouldnt usually be called "mapping" a consistant zero reference isthe starting point for perhaps later map


Yes I know this part would run great on a new Tormach Slant Bed Lathe.....

Thanks

Dave Lawrence

[Don Clement]

Don - Mightn't the work, fixture, and machine temperatures all be sufficiently different from ambient temperature, or even each other to introduce an error that could be greater than the tolerance??

Am I to refuse to eat because I do not fully understand the mechanism of digestion? ~Oliver Heaviside