Machining Formulas

snipped for brevity

they forgot the most important formulas

+-.03 =$ +_.005=$$ +-.001=$$$$ +-.0001=$$$$$$$$$$$$$$$$$

and of course haas wouldn't print these formulas

+-.01=Fadal +-.005=Haas +-.001=Mazak (cheap series) +-.0005=OKK, Makino, insert your favorite here (into a temp controlled room)

controlled

:>)

jon

Switch your CNC machine into Metric and all most tripled your machine accuracy. That's how encoders are calibrated. JS

"Szaki" wrote in news:BI-dnW4rh7KQ4cXfRVn- snipped-for-privacy@comcast.com:

That's so wrong I don't even know where to begin.

machine

At the beginning ? :>)

jon

"jon_banquer" wrote in news:1113106548.649280.87690 @l41g2000cwc.googlegroups.com:

It's simple. The encoder and servo motor rotational position is in fractions of a degree. This is translated into linear motion via the ball screw. It makes no difference whether you program a move of 1" or 25.4 millimeters, the motor moves the same number of degrees rotationaly. To the interpolator it's all just numbers. The resolution of a decent servo system far exceeds the machines mechanical ability to position to the same resolution.

Some "encoders" are linear scales, most, if not all, of which are actually metric. If the ballscrews are metric then a rotary encoder outputs X pulses per metric unit distance (per full rev).

In addition, Y pulses is probably never an *exact* decimal inch distance. The control has to convert from internal metric distance to inch units, losing precision due to truncation & rounding, not to mention that .0001 inch is larger than .001 mm (typical programmable word address resolutions).

For all I know if the control does not have to convert units it may be faster too. There may be some machine "error" in some cases when the conversion takes too much time ... I don't kow.

Cliff wrote in news: snipped-for-privacy@4ax.com:

The ones that are on our machines that are equipped with scales are either .1 or.01 micron resolution. Which means exactly nothing when the machine is programmed in inch. The accuracy of interpolation and positioning have little to do with the resolution of the feedback anymore. The servos are so precise that there really is not a lot of need to monitor actual position. There are several companies employing open loop systems systems in their machines these days. The program is compiled into binary data and is run synchronized to the processor clock. These systems bypass the interpolator and have no feedback to the control. The advantage is that processing is eliminated so the machine runs at very high speed. An infinite number of axes can be simultaneously interpolated with a low cost control as well. The big disadvantage is that editing the program is a bitch. You have to edit the G-code then recompile the program and load the data tables whenever you want to make a change to the program.

Which again means exactly nothing when programming in inches.

Nor in metric. Think about it.

I can only speak from experience. We did a turnkey and SPC acceptance run off on a very tight tolerance part. They never told us that they wanted the machine programmed in metric. We passed the run off in inch mode, then repeated it upon installation in metric. There was no difference in either cycle time or precision.

D Murphy wrote in news:Xns9634C543E1026BW12BU20MU38SY@130.133.1.4:

Although the control DISPLAYS 0.0001" or 0.001mm, the actual values it would use internally to calculate positions would be many more decimal places. The margin for error in any machine with less than a few miles in axis length would be so slight as to be negligable, and the error would be almost infinitely smaller than the mechanical capabilities of any machine.

Anthony wrote in news:Xns9634EAC068AC9acziparle3sp835@216.77.188.18:

On the newer Fanuc controls it's a nano meter.

Kind of like the resolution on those metric scales & screws?

But you want inches? What does 1 namometer equal in inches? To how many decimal places?

Does it matter? One micro inch is 25.4 nano meters to however decimal points you want. Build the machine to inch standards internally and your conversions will be exact. However, the scales are not really 1 nM, and the screws have neither a uniform nor an exact pitch. Ultra-precision machines are all error compensated so nothing is gained being in either inch or metric. My machines have a feedback resolution of approximatly

34 pico meters. I say approximatly because even though the scales themselves are Zerodur, and the mounting spar is super invar, and I have manufacturesrs certifications and compensation maps, ultimatly the scale must be compensated against some known traceable master.

FWIW, the math in the parser is carried out at double precision floating point (56 bit mantissa, about 15 decimal places of accuracy)

Believe it or not, most (if not all) CNC controls are integer math position loop and feedback systems. Why? Because scales and resolvers when they are decoded in hardware are integer counters and registers. The control actually uses feedback units (FBU) that are integral representations of the interval between rollover points, as in the case of resolvers, or a relative count of the position along a scale. Since we are counting pulses, the units are integers and the pitch of the scale or the gearbox/pulley drive determines what distance those counts represent. The servo system then maintains internal units that represent the current position of the system in FBU and these are then converted to inch or metric units for display purposes. Absolute encoders are merely extended relative encoders with rollover points that are greater than the expected length of travel of the slide. Absolute encoders do all the counting internally and present the control with a set of bits that represent the position of the encoder within its travel range. The control uses predefined integral multiplication factors to calculate the integral slide position based on that count. Distance coded scales are incremental scales with check-points along the way that make it easier to home or reference the system.

The question is not how many decimal points do you want, but how many decimal points of precision are meaningful. In the case of most "real world" CNC positioners in the metalworking industry, the answer is no more than .0002 mm or .00002 inch (yes 20 micro inches) per FBU. The general rule of thumb is that the servo system must resolve 4 to 10 times better than the expected position accuracy. So if you want to hold and program +/-0.001 mm you better be able to resolve 0.00025 mm or better. The selection of scale pitch, interpolation electronics and maximum slide velocities is the realm of the design trade-offs and limitations in your control technologies and electronics. In other words, if you want high speeds, expect to use higher performance CPU and feedback systems, if you want high precision, expect to use slower speeds to realize them.

. I've been trying to tell them those numbers represent metric unit distances when you have metric unit feedback/ballscrews. Some seem to want to first convert/round/truncate them to inch units .....

Cliff wrote in news:r4lo5197is7qtm1lao1qt9ak662i5l947e@

4ax.com:

Really? That's not what the thread is about, is it? The original claim was that CNC machines were more accurate when run in metric mode. So far none of your clip and snip bullshit has convinced me otherwise. Feel free to keep on trying to belittle those who know better.

news:r4lo5197is7qtm1lao1qt9ak662i5l947e@

Dan,

Since you know his game and understand how he uses belittling tactics to make himself feel superior, why do you bother to respond to him ? Is responding to him good for this newsgroup ? Is it good for you ? Don't you think that most of us recognize that your very talented and highly skilled with advanced CNC machines ?

Have you noticed how many of the regulars no longer respond to him. Note how many of the regulars he pings and they *never* respond. Note how many have asked him to stop what you label as "belittle" tactics and yet he ignores them all and makes a joke of it. The FACT is the joke is on him and it's a sad one at that.

Why not consider debating those who don't feel the need to belittle you ? I don't agree with some of what you post. An example would be the Haas control vs. the Fanuc control but I don't feel the need to belittle you or ridicule you just as I know you don't feel the need to do that to me.

Most of us here know he has very little actual machining experience... at one time he was a contract APT programmer at a big company and in the latter stages of his career, before he was forcibly retired, he was a Unix system administrator. No actual machining experience is listed in his resume. Bill Roberto has pointed out several times that he hates machinists and the he needs to make himself feel superior to them.

He obviously feels he's an expert on everything and he can never admit when he is wrong and takes great pride in this, so why not make better use of your time here by doing things like providing more info on the very advanced machines you sell, etc. Start you own threads if necessary.

Your time is valuable. Why waste it on him... he's simply not worth it and just about everyone here now knows and understands this.

He's forced many people that I like to give up this newsgroup because they can't stand him. I don't wish to see yet another disappear.

It's a losing game. Opt out. I have for I think 2 years and it's been great ! :>) Wish I had done it much, much sooner.

jon

"jon_banquer" wrote in news:1113371804.723144.20160 @z14g2000cwz.googlegroups.com:

I've actually thought about this some. I think that Cliff contributes some interesting stuff now and then. He often sees what some overlook, and has a different perspective that adds to the fabric of the group. I think that he's a smart guy and he certainly knows quite a bit about specific topics. It's just unfortunate that you have to wade through 150 posts of utter nonesense per week to get to the relevant good stuff. Then there is the problem of him not being able to let go of an issue that he didn't feel got resolved to his liking. This thread on machine resolution is an example of that. There was a metric vs. inch thread a little while back and instead of letting it go, he tries to drag me back into it on this thread. He knows very well that programming the machine in metric vs. inch makes no discernable difference in the accuracy of the parts being machined. As the resolution of the servo system on a properly designed machine tool is several orders of magnitude higher than the minimum input resolution. I'll tell you what, I'll try to let some of what he writes roll off my back from here on out. But if people stop debating and others stop chiming in, nobody learns anything. Then what would be the point?

news:1113371804.723144.20160

tactics

example of

His behavior of belittling others *far outweighs* anything he does contribute. He's done more damage to this newsgroup than anyone ever has by driving away numerous posters because he has made the environment here an unenjoyable one for many current and former posters. I'd be more than happy to start naming names if need be.

"I think that he's a smart guy and he certainly knows quite a bit about specific topics"

How much longer are you going to keep making excuses for his behavior ? Yes he is smart on some subjects and on others ignorant beyond belief.

issue...

That's part of the mental illness that he suffers from and it shows itself in many, many ways. You gave just one example. I can probably list 5 examples right off the top of my head. If I really thought about it, perhaps 10 examples.

back >from here on out. But if people stop debating and others stop chiming in, >nobody learns anything. Then what would be the point

Raise the bar, Dan. When you raise the bar he loses and the newsgroup wins.

I'll give you an example of what I consider to be raising the bar:

I feel that Haas is a far better thought out control than a FANUC. From how you can whip through a program with like 900 percent override, to the built in engraving, to not needing a laptop to do extensive editing very quickly.. you can can uses the Haas electronic hand wheel for editing as well.

However, I tend to think your right about FANUC being a control that is capable of higher accuracy on better machines. What I would like from you is a simple guide to understanding the technologies that FANUC and others uses to achieve it. Even though you don't deal with Haas, I would appreciate it if you looked at a HAAS equipt machine and compared it to something you sell and pointed out where whatever technology is superior to what Haas is using. That's the kind of useful info I'm looking for. I think others are as well.

I have no doubt your right about a Fanuc being capable of things a Haas control isn't but I need simple explanations of why other technologies are superior to what Haas puts in their machines.

Most machinists are not electrical engineers so lets keep it on a basic level.

Thanks for taking the time to respond to what I had to say, Dan.

jon

Curious as to how a handwheel helps edit.

Regards

Daveb