Hi All,

Let's say you have any sort of cnc table with one of the axis dual drive. Just for arguments sake lets say it was a gear rack on each side and due to terrible manufacturing one gear rack was a few mm longer than the other (please don't get caught up in the aforementioned sentence, it is just hypothetical and the next paragraph is the question I'm trying to find an answer for). To compensate for this you'd need to have different steps/mm for the master and slave motors. Now here's my question.

In the Mach manual it says that a slaved axis recieves the same steps pulses as the master (under normal conditions !?). Is it possible for a slaved axis to have different steps/mm as the master axis. This of course would mean that the slaved axis could not get exactly the same pulses as the master. Rather the slave would recieve the correct number of pulses to move it the same distance as the master.

A member on Cnczone said he tried programming different steps/mm in his slaved axis but Mach ignored it and sent the same pulses as the master. Anyone know of any workarounds for this one.

What is meant in the Mach manual when it says "under normal conditions". Does this imply there are non-normal conditions where they don't get equal steps.

Thanks,

Keith.

Keith

As far as I know the only "Non Normal Conditions" is when the slaved axis homing is done.

During this process the slaved axis is temporarily decoupled from the master.  And this is when different settings between master and slave will cause problems.  Even homing speed.

The rest of the time slave axis settings are ignored and master settings are used.

In answer to your real question I suppose it should be possible to have your CAM output individual position calls.

This,

G1 X100 Y50 Z20

Would be output as,

G1 X100 A100 Y50 Z20

So every time an X position was called, it would be duplicated as an A position.  Your CAM post processor should be editable to do so.

This way different steps per unit settings should be possible.  Acceleration rates would still need to match.

However not having the X and A slaved together would mean other complications.  There are button codes to jog X and A together.  Not sure how or if you could handle homing.

Any way this is all theoretical.   Not sure at all if it would work. 

Greg

Not sure how or if you could handle homing.

Greg - Using your scheme above (very creative) I'm wondering if maybe a refCombination for X and A would do it.

Ian

During this process the slaved axis is temporarily decoupled from the master.  And this is when different settings between master and slave will cause problems.  Even homing speed.

Especially Homing Speed. Learned that one first hand.


I think you're making a big issue out of nothing. In over 5 years, you're the second person I've seen worrying about this prior to actually using it.
But I've never seen one complaint from the hundreds of people using it on a daily basis. Almost all the routers you see being built on CNC Zone use slaved gantries, and a lot of them do very precise work.

KEITH here is what you keep getting caught up in. YOU DON"T BUILD THE MACHINE THAT WAY.

You must build the machine to be accurate. IF not nothing in MACH is going to save you from yourself.

There is tuning in mach to compensate for manf tolerances (screw mapping and formulas) Yours is a design /build/ tolerance problem

(;-) TP

Terry, 

I totally agree BUT as a beginner and this being my first table I built it the way it is. I know a heck of a lot more now than when I first designed my table (thanks to helpful guys like yourself).

I WILL change it when money and time allow, but for now I just wanted to fully check out that there is no workaround for it. I'm not "caught up" in it, simply exhausting any possiblity of a software workaround which if it had been possible, would have been fast and cheap. How do I know if I don't ask. It is the first time I've asked on the Mach forum where if there was a workaround it would most likely be found. Regarding this matter on the other forum, this question was not addressed a great deal. The post ended up more about the issues of my chain design rather than the original question of a software workaround. Hence the way I started this topic.

Hey by the way you are totally right about my chain design. A lot more bouncing with mine. I've got quite good acceleration but the sudden direction changes cause a small but fast bounce in my chains which show up in the cuts as a "hacksaw blade" edge. The way my table is at the moment it will be smallest tips possible to get the lowest speed and looping corners wherever I can, that's where my "sawtooth" cut edge comes after.

Cheers,

Keith

I have a similar problem with my router gantry.  It is built on 25mm ballscrews that are 1500mm in length.  One side travels just slightly more than the theoretical 5mm pitch.  The error is very small but accumulates over a long cut.  At one end of travel the gantry is slightly out of square.  I have reduced the error by making sure it is square at the center and splitting the difference seen in large parts. Or by setting the material up close to the machine zero to minimize travel.  It would be a very nice addition to Mach3 if separate steps/unit travel could be assigned to accomodate small mechanical differences while retaining the other properties needed with a slaved axis.  Post processing editing to assign movements to each axis is out of my league right now and may not be supported by the G540 controller I'm using.  If anyone knows of a patch to enable such a feature that would be really helpful.  Right now I am trying to calibrate each ballscrew on the machine to see if the z axis screw is a closer match.  Wish I had thought of that prior to building!

A possibility is to do the correction mechanically, rather than in software. In other words, physically change the gearing between the motor and screw.

The most accessible reference on this would be Cleeve's 'Screwcutting in the lathe' where he gives various compound gear combinations that do such ratios as 999/1000 or 1001/1000, etc. They are as simple as 27/26 x 24/25 (0.997).

Similarly, if the leadscrew varies along its length, do the correction in the same way as jig borers: the leadscrew nut is allowed to rotate and is guided up or down a ramped profle, thus advancing or retarding the axis movement

Tpetmed,

I'm so happy I'm not the only one who has this problem. Since actually theorising about this situation I measured the travels on both sides of my gantry and got a difference of 1mm. So by doing the same thing as you and squaring the gantry in the middle of it's travel I'll get +0.5mm out of square at one end and -0.5mm out of square the other end.

Your post has actually been very valuable in another sense. From my previous posts (other forums too) on this subject it was made obvious that many members just assumed chain lengths would be exactly identical in length (over 2.4 metres) and what surprised me more was that they believed they would wear exactly equally too. Your post has shown that even leadscrews can have slightly different lengths, so perhaps so can gear racks. Definitely worthwhile checking pairs of items to be bought to make sure they are exacly the same length.

Heavy industry knows about this issue and they don't rely on hope that gear racks are made perfectly. They travel lengths are much greater of course. They use some precision means of measuring the travel bit by bit and their computers store the REAL travel distance in memory so that even over runs of many metres their gantries always remain square. Each side of their gantry is individually calibrated using this method.

I'm glad that my theory turned into reality, primarilly because a few members (over various forums) were telling me I don't need to worry about it and that new chains SHOULD be an equal length and that they SHOULD wear exactly equally over time. Nobody who said this mentioned if they had ever even measured their chains. My chains are pretty much brand new, same make, good quality, and I have a 1mm difference from the start.

Keith

I was thinking about this the other day.

In most CAM programs, it's pretty easy to modify the post to output both X and A commands for X movement.

And there was a thread here the other day about jogging multiple axis' with one button. I think you need to use physical external buttons to do it though.

But that would let you guys do what you want. MDI would be tricky, and unforgiving of typos, though.
But it's an option when the standard slaving won't work.