Machsupport Forum

Mach Discussion => General Mach Discussion => Topic started by: D CUSTOMS on June 12, 2008, 08:54:43 PM

Title: Matching Dros
Post by: D CUSTOMS on June 12, 2008, 08:54:43 PM
How important is it for the Dro on the mill and the Dro in mach to match?
Title: Re: Matching Dros
Post by: jimpinder on June 13, 2008, 02:18:00 AM
I take it your mean that you have some sort of independent scale attached to your miller which gives a read out, as well as the computer screen which also shows a readout.

Since they are both independent, more than likely they will both show different readings. This is not important. The main thing is, if the reading on Mach says an axis has moved 1 inch, then the reading on the other scale should say it has moved 1 inch. Neither reading has any influence on the program - it is just for your information.

By that I mean that the DRO's on Mach 3 do not recieve any signal from your machine. If you want to move 1 inch, Mach looks at the steps per unit you have set, and bangs out the appropriate number of pulses to move the axis. Similarly your other scale has no influence on the machine either. You are dependent on keeping your machine operating within reasonable parameters to maintain accuracy. Start trying to go too fast or too much acceleration, or if your machine binds here and there then you will miss steps. You will find that Mach 3 says the machine has gone further that your other scale reads.

If you are fitting "home" switches, then you could "zero" both DRO's when you home, and this would keep them in step - you would soon see if the were different, However - you would be limited to "machine co-ordinates" since your other DRO would have no way of taking on board any offsets that Mach uses. Having said that, it is easy enough to add a set of DRO's to any of the Mach pages (and you can reduce the size) and say fit a small readout under the main one - and show on this the machine co-ordinates - which you can glance at and check if they are still in line.

There is also an add on board you can get which reconciles the two and puts out an alarm if the two readouts get out of kilter. This will stop your machine, but there is no way at the moment to get Mach 3 to then correct the fault.

Title: Re: Matching Dros
Post by: D CUSTOMS on June 13, 2008, 07:43:55 AM
Thanks for your response.I was setting the steps by inputting G0X1 then zeroing out Dros.Then inputting G0X2 and adjusting to get both to read 1 inch.The Dros are accurate at low numbers but differ at higher ones.
Title: Re: Matching Dros
Post by: jimpinder on June 13, 2008, 02:27:26 PM
You're method is quite alright - BUT -
Don't take this as a criticism, but I don't see what adjustment you can make.

To start with Mach 3, a 1 inch movement is governed by the steps per unit you enter in Motor Tuning. This is not something you measure, or adjust, it is something you have to calculate. I know you can get Mach 3 to do it for you, but that depends on you measuring something, and therefore is open to error.
Steps per unit is a multiple of :- Microsteps (set on the motor driver card e.g Gecko = 10), Steps per revolution of the motor (1.8 degree motor = 200), any gearing gown to drive the leadscrew (mine is 3), and finally turns of the leadscrew to move one inch (mine is 10). The answer for my motors is therefore 60,000 pulses per inch (Not 59,945 or 60,021 - which I might get if I resort to measuring). You can see that I have 60 pulses per 1000th of an inch. I cannot measure that small, therefore I must be innacurate.

Your secondary digital readouts will have an accuracy printed on them, but the digital calipers I have are fixed - I cannot adjust them. Whether you can with yours, I don't know, but of course - if you are adjusting them - what do you use as a measure.

My steps per inch are accurate, and I measure them with by digital calipers and usually get to 1/2 a thou, which is as accurate as I can measure with the equipment I have got.

If you used Mach 3 for your steps per unit, then calculate it out instead (it should be a round figure like 60,000) unless your leadscrew is metric or something, but even then you could set up in metric if that was the case - and still settle on a round figure.

Once you get your step per unit correct, Mach 3 will be accurate, unless your motors miss steps, and this to me is where your second DRO comes in. If your motor miss steps, then Mach 3 will say, put out 60,000 pulses, but you motors might receive only  59,750 meaning that the actual movement is just over 4 thou short - and this should be picked up on your second DRO.

Other than that, I would have thought that the two DRO's should match - within the tolerances of your second readout.

Title: Re: Matching Dros
Post by: D CUSTOMS on June 13, 2008, 09:35:11 PM
Thanks Jim and Rich for your responces.What i ended up doing is setting up a dial indicator for each axis.Then i did a G0X1 move.Reset mach dro and dial indicator to 0.Then put in G0X2.If it moved more or less than 1 inch i adjusted the pulses so the dial indicator read 1 inch.However all of my final settings ended up being allot of different numbers-example 20,976037 instead of 20,000.Now in theory if i then put in G0X1 and the axis doesn't fully return to 0 that number would be my backlash.Am i correct??
Title: Re: Matching Dros
Post by: jimpinder on June 14, 2008, 04:37:11 AM
I am sorry - but I must disagree with Rich.

I sound pompous - BUT - the only way  to set your steps per unit is to calculate it. Think about it !! It cannot vary !! The number of microsteps does not alter, the number of steps per rev does not alter, the gearing does not alter, and the leadscrew does not alter.
All these parts of the system are, in themselves accurate to a tolerance (AND THAT ERROR IS NOT REPEATABLE) so for any step there will be an accumulation of errors in the system - lets say 10% - but this is 10% of 1/20th of 1 thou. The machine moves to the next step, but the error does not go with it, it starts again from zero. So for any step you can get a minute positioning error - but it is only for that step - the distance you move has no bearing.

If you get the answer 20,000 as someone said, which is similar to mine without the gearing, then that is the figure. It is no good then arbitarily selecting one part of the table travel and then checking that figure over that distance. As Rich said, there might be some slight imperfection in the leadscrew that has moved the table more or less in the right direction - and this shows up when you measure (the trouble is your measurement is not accurate either - you cannot measure to a 20th of a 1000th of an inch) - so you adjust the steps per unit. The problem is then, you move to some other part of the table, when the leadscrew has a slight imperfection the other way - and your calculations are well and truely out this time. What do you do then.

NO - the steps per unit are a fixed mathmatical number - If you had 10 machines with the same motors, drives and leadscrews and gearing, their number of steps per unit would be the same.

There are many reasons why DRO's don't match - but the biggest problem, as with all things digital, is that we can see them too easily. - Ah, you say, the error is 123 - 123 what? and when you calculate it, it is minute, and we wouldn't have bothered with it if we didn't have something telling us it was there. The second reason is that there is a tolerance on each thing we use - there has to be or everything would seize solid. If the tolerances add up the same way then they are not noticed, if they add up against each other they are noticed.

That is enough diatribe - If your anwer is 20,000 - use 20,000 - it is the correct figure for the entire length of your bed.

As far as backlash is concerned, this is entirely different, and what I call the non-movemrnt of the axis. It is simple to check.
Make sure backlash is turned OFF first.
Move x0 to x1. Set your measurement DTI up and set to zero. Move x1 to x2 then back to x1 (using g0 or G1 commands (not jogging).
In theory your DTI should read zero, but it will not. Your table will not have returned the full 1 inch. The shortfall is backlash, and can be entered in the backlash table.
The reason is that when the axis change direction, all the parts of the drive train have to settle down and put their shoulders in to shoving the table the other way. It is chiefly the leadscrew and the gearing. It is unavoidable and is present on all mechanical set-ups. There are those who say they don't have any backlash - principally the ball screw fraternity (which I hope to join eventually) - but they are mistaken - what they mean is they can't measure it.

Mach 3 deals with backlash very well - and applies it everytime the axis changes direction - and in my experience it is very accurate (and you should see my backlash)

I think that enough to be going on with now.

Title: Re: Matching Dros
Post by: jimpinder on June 14, 2008, 12:18:50 PM
Yes - pardon me for saying this - I take it English is not your first language - congratulations because it is better than my whatever language it is.
I understand what you are saying.
1. Yes - there may be an error in your ball screw. You are told that it is 10 turns to the inch, it may be 10.05. This means that your calculation will be incorrect. A screw thread is one thing where there could be a continuous error - where the pitch is too great and the error accumulates over the distance.  With modern production methods this is not too likely, but possible. I see no problem with making adjustments to your steps per inch in those circumstances.
2. If your machine is such that there is such an error over a particular part of the travel that you are using, and you want to be very accurate (over that part) then - yes- adjust the steps per unit to take account. I cannot think of another way to do it. But you MUST be aware that to use such a solution throws out your accuracy over the rest of the travel of the table.

D Customs - I hope you are understanding all this, because we are getting very theoretical, which is not really to the advantage of someone trying to learn. My own qualifications are in Maths and there will be many on this forum who are far more experienced in CNC that I am, I only came to it about a year ago, when I started to convert my lathe/miller to CNC. I am retired so my use of the machine is only as a hobby, although I do build miniture railway locomotives and equipment and use the lathe/mill for that.

I think we are agreed anyway. Use the 20,000 figure for your  steps per unit, and how did you go on with the backlash. Now backlash is one figure on which you can measure and use an average error if you want  ;D
Title: Re: Matching Dros
Post by: D CUSTOMS on June 14, 2008, 07:27:03 PM
Sometimes i feel like i am no further ahead than i was last month.The correct pulses for 1 inch i got was around 20,967.You say just use 20,000 instead of 20,967.When i changed it to 20,000 the axis wouldn't travel a complete inch.I thought the idea was to have mach command the table to move to the exact number.Do i use the 20,967 because i get the complete inch out of it or do i change it to 20,000 like you said???
Title: Re: Matching Dros
Post by: Chip on June 14, 2008, 11:21:37 PM

Use the 20,967 that gives you a 1 inch move in the Pos direction ( x0 to x1 pre-load, zero the indicator, then x1 to x2 (20,976) 1 inch) OK.

Now move back to x1, Does the indicator move back to Zero, If not the distance to go is "Backlash".

You'll need to Enable Backlash, Config, "Backlash" or find out where the slop is.

Not to Bad 17 post's, 2 fwd, 1 back, 2 fwd, 1 back, Won't be long now.

Let us Now, Chip
Title: Re: Matching Dros
Post by: jimpinder on June 15, 2008, 02:38:09 AM
I will only say this once again, then I shall have to withdraw from this post.

If your calculated number of steps per inch is 20,000 - and that calculation is correct, then, despite what you say, measure or otherwise arrive at, that is the number of pulses that will give you the correct one inch of travel. Your figure of 20,000 would seem to be correct. Mine is 60,000 but I have 3 to 1 gearing on my leadscrews. It would seem you have Gecko drives (or similar) with 10 microsteps, you have 1.8 degree motors, and a 1/10 inch pitch leadscrew. Now - the only one of those that can change is the leadscrew, and I have never heard a lathe manufacturer who produces a leadscrew of 0.1048 pitch. Even if the leadscrew was a 25mm metric masquarading as an imperial screw, the pitch would be 0.1016.

You have arrived at your 20,967 by some sort of measuring - and it is incorrect. It may be you have run the table left, set everyting to zero, then run the table to the right for one inch measured by one of your  measures and gradually increased the number of steps till you got an accurate reading. The problem there is backlash - and you appear to have about 50 thou (same as me).

If you are taking measurements, you must always move the table in the same direction. Enter 20,000 as your steps per inch. Move the table 1/2inch right to get rid of backlash, then set up your DTI and zero it. Now move the table to the right again. If you use a G1 command, you can specify the feedrate f say at 4 to keep speeds low so you dont miss steps, and when Mach tells you it has moved 1 inch check your DTI It should be correct to within 1 thou - maybe 2. Move the table another inch to the right - now two inches, and check with the DTI this should also be accurate to within a thou or 2, because errors do not accumulate.I used digital calipers to check mine, and could do up to 5 inches. I don't know what measuring equipment you have. I am a bit reluctant, but I suppose you could use your secondary scale.

If the error using such a method DOES NOT accumulate - i,e, you were 1 or 2 thou out every time over the length you checked, then your pulses per inch are correct.
If the error KEPT INCREASING  each inch of travel, then your pulses per inch was incorrect - and I will stand corrected.

That is all I can say. I will read the post with interest to find out what the answer is.

Title: Re: Matching Dros
Post by: jimpinder on June 15, 2008, 11:39:55 PM
This is exactly what I was saying (even with the correct numbers in). In actual fact there are not that many combinations of steps per inch.

Micro steps might change - depending on your cards ( some can do 1/2 steps right up to 1/16th steps) - motors seem to generally be 1.8degree these days - fixed at 200 steps per rev.

The difficulty might lie in the leadscrew. In Britain, although in theory we went metric many years ago, we never did, and all we have now, instead of one system, we have a metric/imperial mash. We have things that are sold as 8 feet by 4 feet in the building trade - and they turn out to be 2400 by 1200mm - i.e. short. I regulary buy 10mm bolts, the head on which should be 17 mill hex - the head turns out to be 5/8 inches.

I can see leadscrews in that sort of thing - marketed by some as 1/10 inch pitch, but being 2.5mm - it might not be as bad in America.

All I can say to Mr Cousins is - I hope you are following all this - it has certainly got a bit off the matching DRO's we started with. I think the pictorial by Rick spells out what I have been saying. Check your leadscrew with your manufacturer and see exactly what pitch it is if you are in doubt, and work from there.
Title: Re: Matching Dros
Post by: Chip on June 16, 2008, 05:38:40 PM

I'm sorry this has turned into a Mathmatical/Threory Class.

For Sure You Can Calibrate any Axis without knowing anything about the BallScrew Pitch or Reduction/Gearing.

Mach shows on it's DRO's to .0001 ackuracy, But in fact it retaines a .000000000000001 "15" dec. point's or so I Think, Precision Accuracy (The Remainder is "Applied" to the "Closest Step Per" Unit as setup in Motor Tuning.

In my last post I was trying to help you find out If you have Backlash Issues Also.


Using the 20,967 that's Giving you a 1 inch move.

Move the axis "pos" 1 inch "Pre-Load", (It takes backlash out, If there is any.

Then zero the indicator and your linear scale DRO, Move in the pos dir 1 inch.

Dose the indicator show a 1 inch move. ? _____________

Dose the linear scale "DRO" show a 1 inch move. ? _______________

Dose the Mach show a 1 inch move. ? ______________

Now move the other direction 1 inch.

Does the indicator move back to Zero, ? _________

Dose the linear scale "DRO" show a 1 inch move. ? __________

Dose the Mach show a 1 inch move. ? ________

If the indicator dosen't return to 0, The distance to go is "Backlash".

Thanks, Chip
Title: Re: Matching Dros
Post by: DAlgie on June 16, 2008, 06:06:00 PM
Hate to jump into this, but you might also check your table gibs out. Most smaller machines, and certainly the cheaper chinese machines use set screws for gib loads, these make up for a lot of manufacturing problems but aren't very rigid. My chinese lathe has the more rigid tapered gibs, BUT; they were very poorly fitted, had big clearance on one end while the other was tight, and weren't even STRAIGHT!, had some curvature to them. I had to surface grind the X axis one to the correct taper and to grind the bow out, then refit it. All this can make a table or crosslide walk around and if you have digital readout scales on one side this might account for your differences in readings and even backlash. Mount a dial indicator on the table side and grab it and load it by hand, you might be surprised to see one end move and the other be tight, etc.
Title: Re: Matching Dros
Post by: D CUSTOMS on June 16, 2008, 08:20:18 PM
Thanks for the advise guys.Chip i did what you said and everything came out great!!!THANKS.