Machsupport Forum
Mach Discussion => Mach4 General Discussion => Topic started by: peterbelfanti on January 01, 2016, 10:12:24 PM
-
I've just built a new G0704, using Mach 4.
Here is a link to my backlash test
https://drive.google.com/file/d/0B7IRZwlMOrbxVGVEaHJCZzQ5Ymc/view?usp=sharing (https://drive.google.com/file/d/0B7IRZwlMOrbxVGVEaHJCZzQ5Ymc/view?usp=sharing)
As you can see in the video I can cycle 1 inch off zero and back to zero with no drift. When I approach zero with my indicator zero my X axis and back off zero, there is backlash present during my jog moves. The DRO shows movement, and no table movement. I expect backlash values to be applied to Jog moves as well as MDI commands.
Am I missing something? am I crazy? shouldn't manual Jog moves be backlash compensated as well?
Thank you
Pete
-
I don't think any CNC applies backlash comp for manual moves. When you run a manual machine you always deal with the backlash by approaching a point always from the same direction or by cranking the other way by the backlash amount. In auto backlash comp occurs because the cnc knows to keep the tool on the path. There is no path when jogging, so know way to know what should be done.
-
well that makes me feel better that I'm not crazy or have missed a setting... This explains it..
Just getting started with this mill, and the hobby..
-
What motion device are you using?
FYI - the Mach4 development team decided that backlash compensation would be handled by the plug-ins and devices, not by the Mach4 core (i.e. Mach4 motion planner). So whether your device supports backlash compensation, and how and when it is "enabled" is up to the plug-in.
Bob
-
Warp9 Ethernet smooth stepper.
-
Backlash is not yet implemented in the ESS plugin but will be in the new plugin that should be release shortly (please).
DazTheGas
-
DazTheGas,
so I'm a bit confused then. I thought backlash was compensated in the Motor tuning variable in Mach? or Does the motion controller do the compensation? or do the two work together?
Finally any idea on when the new plug in will be available?
Thank you
pete
-
so I'm a bit confused then. I thought backlash was compensated in the Motor tuning variable in Mach? or Does the motion controller do the compensation? or do the two work together?
There is indeed a field in the motor configuration tab for setting the backlash. While the Mach4 core does nothing with that value, the plug-in is able to read that value** and use it for their backlash algorithms.
** note that plug-ins are also free to disregard that value and force you the enter a value somewhere in the plug-in specific configuration dialog. It all depends on how the plug-in is written, and I have no idea how any plug-ins other than the PMDX plug-ins handle that.
Bob
-
For the record:
Darwin was written prior to Mach4, but to the specs of M4 at that time which didn't call for it.. So backlash was not added
to Darwin. There is a school of thought, which was prevalent during my time in Mach3, that backlash
shouldn't be corrected and that backlash should be mechanically fixed or accepted, personally I tend
be be in that camp on my own machines anyway. Minimizing backlash with springs or anti-backlash nuts etc
is probably the best way to deal with the issue, but when applied to a core application, there is no end to
the misery it can cause to the logical application of motion rules in the engines.
Its like asking your cars software to help with a loose steering wheel, could be done, but probably
shouldn't be..it would affect other safety issues right down to the tires. Just my 2 cents, and I
thought Id explain why Darwin doesn't have backlash..
Art
-
I agree backlash should be removed from all mechanical parts... unfortunately mechanical parts wear and your going to get backlash... after playing with Mach4 and Darwin and knowing what I know now I would have never purchased it.... I'm not knocking it but it doesn't have all the things that mach3 had and probably never will.... so I am now moving to Linuxcnc which has changed alot since I last looked at it and works very well with the parallel port. hopefully I can find another use for Mach4 and Darwin so it won't be a waste of money.
Dave
-
There is a school of thought, which was prevalent during my time in Mach3, that backlash
shouldn't be corrected and that backlash should be mechanically fixed or accepted, ...
Its like asking your cars software to help with a loose steering wheel, could be done, but probably
shouldn't be..it would affect other safety issues right down to the tires.
Doubtless there will be howls of anger, but I have to agree 100% with Art.
For a start, if you build a sloppy car or machine, you get what you have paid for. Don't expect anything more. You won't get it and you don't deserve it. Bubba need not apply.
But second, you can not correct for backlash completely. It just is not possible. Oh sure, in theory you could look ahead and see what is going to happen IN THEORY and how you might correct for it in software, but that won't work when you come to real hardware in the real world.
Any experienced machinist knows that when you are using a manual mill with ACME threads on the X & Y axes (always with backlash), you never climb mill; you always conventional mill. Why? Because if you try to climb mill your mill table is suddenly going to jump by a distance equal to the backlash in the ACME thread. That way lies smashed cutters and wrecked jobs - and possible injury to the machinist.
The same applies to a CNC mill or lathe. For sure, your software can pretend to correct for backlash, but it cannot handle those sudden jumps the table makes when forces reverse and the backlash is taken up in the other direction. Having seen a 20 mm HSS cutter snap and go flying in my early days ...
Basically, backlash compensation is a dream which cannot come true. Get real, and fix your machine. If there is wear, fix it!
Cheers
Roger
-
100% correct rodger
-
I've been looking for something to replace my flashcut pro series cnc box for my G0704, which does have backlash compensation and it works well, it's the other other parts of the software/hardware i'm having problems with. So looking for an alternative too, i saw mach3 had it, now reading up about it and disappointed to see this, if there is a mach3/4 +controller that does support backlash compensation, and we have worked our hardest to get mechanical backlash to a minimum, but there is only so much we can do and i'd like to be able to keep fitting bearings after an endmill and not with a boring bar.
As for the car analogy, that is what a closed loop O2 sensor does, it looks for changes in the fueling so it can feed back to the engine controller and trim for fuel changes, small air leaks, and a whole host of other things that are correctable by software for a small hardware issue which will occur on every car eventually. That seems a lot closer to backlash than a loose steering wheel.
should i just look at mach3 and not 4?
-
Hello Charlix, welcome to the forum.
We all know that allowing misinformation to be repeated is the stuff that turns myths into legends right? Well, this is the only reason I'm going to hit on this topic that has been discussed many, many times already.
As for the car analogy, that is what a closed loop O2 sensor does, it looks for changes in the fueling so it can feed back to the engine controller and trim for fuel changes, small air leaks, and a whole host of other things that are correctable by software for a small hardware issue which will occur on every car eventually. That seems a lot closer to backlash than a loose steering wheel.
This is where you are dead wrong. While your analogy of an O2 sensor is fairly accurate the only way it can work properly is if it has full control of all elements involved. Should the ECM decide the mixture is a little rich and needs to adjust the air/fuel ratio by lowering the pulse width of the injector. It does so but the actual time the injector is open for does not actually change because there is a piece of trash in the injector keeping it open. This is something the ECM assumes it has control of but actually does not. What now? How does the ECM fix something it has no control over? Does it cut off the fuel pump to lean it down resulting in starving the engine for fuel and ultimately shutting it down? Does it inject air to adjust the A/F ratio and in the process raise the RPM of the engine? Do you add a pile of telemetry sensors and mood recognition to read the environment and drivers face so you can take all of that information to make an educated decision about what is the best option? These things and lots of others could be done to compensate but at what cost and what sense does it make to even try to. The best possible outcome would be piss poor at best.
You can avoid reality but you cannot avoid the consequences of avoiding reality.
Mach3 does have backlash comp. that works as good as any but I wouldn't get Mach3 or Mach4 or any other controller expecting it to be able to fix all the woes introduced with backlash. If I knew of one that could manipulate physics I would sure point you to it.
-
sure but the difference is, its a best case scenario versus absolute perfection, it's trimming to overcome slight issues that occur versus fixing something that is a potential disaster. People seem to have difficultly with middle or mean ground. they hear , here is a solution, but it must fix all possible failure modes instead of, here are at the most common failure modes, ie the ones i listed for an O2 sensor, versus what you listed as a failure mode. It's a trim to overcome the slight issues, not a be all and end all, not having the option is worse. My O2 sensor wasn't able to correct for someone who swapped out an injector not realising that the latency for different injectors although with the same flow size type, is different and now my car runs worse etc..
with that logic next time i design an ECU, or work on a car thats got a non 0 trim. i should say,, oh you're not passing emissions because of the slight wear, change in fuel type, location or other things that an O2 sensor does compensate for and you should go and clean, and fix every tiny part that is out of spec. Sure you can make it perfect, but that is what feedback is for. i demand mechanical perfection.
The analogy is much more apt. you're just limiting yourself to "properly" as being totally accurate, it's flawed , but it works in some scenarios, not all, its a bandaid, fix your machine, we get it. but sometimes we can only go so far and having the option is nice. It's a trim , not a cure all.
Having an closed loop O2 sensor, is better than nothing.
-
If you are trying to correct for 25 microns of backlash, than maybe 'backlash compensation' could be of some limited value. Fair enough, I guess.
But then someone with 1 mm of backlash hears the magic words 'backlash compensation' and thinks it will fix his problems. And it can't. And a fight ensues.
Cheers
Roger
-
yep thats all i'm looking for not some global panacea. there will always be people looking for something magical, but those people shouldn't be the dividing line.
with backlash comp off on my flashcut i can't fit a bearing without using the boring bar, with it on i can, so obviously backlash comp works in that scenario. if i'm doing some complicated climb milling my expectations are different. i don't think thats all that unreasonable, and i'm certainly not of the ilk of how dare X remove it from product Y, they likely know what they're doing. i'm just hoping for a recommendation on a mach3/4 + controller with existing backlash support. i can only really run to the costs of double nut C7s and abec5 on a chinese mill screwing around making stuff in my garage.
but in the meantime i'm reading up on mach3 + suitable controllers now, thanks.
-
OK.
W7, mach3 .062, ESS
I am running with double-ball-nuts and double races, and with no backlash compensation applied. That is working just fine.
Could it be worth your while checking both the end ball races and the ball nuts to see where any backlash is? Dial indicator and some push and shove. Might pay dividends, even to just know where.
Cheers
Roger
-
Charliex, you are totally missing the point here I was trying to make. I agree that the O2 sensor is a good thing and works well when it has control and overall is a good thing. However, if it does not have control the best it can do is indicate troubles with things that are beyond its control. Backlash is just that, a range of motion in which the controller has no control. This is not a best case, worst case scenario, it is not that complicated. Either the controller has control and can do all sorts of great things of varying degrees with the motion or it does not. There is no middle ground there. Backlash comp is asking the controller to control something it has no control over. By admitting the machine has backlash you are admitting there is nothing the controller can do and it is time to call in the mechanic, no two ways about it. Either the mechanics or your machine is capable of holding tight enough tolerances to do its job or it is not. If it is not asking the controller to control something it has no control over is non-productive.
The only effective way to combat the ill effects of backlash is programming strategy and it is not great. It does not fix backlash but does give some control over it.
-
Backlash comp is asking the controller to control something it has no control over.
What a neat summary! I like ti.
cheers
Roger
-
t, i don't think we're in the realms of breaking physics. as a secondary, a typical arrow band o2 sensor is a closed loop system, it feeds back enough information to allow the engine to operate within parameters and try to keep a predefined air/fuel mix for the most efficient idle burn in terms of emissions output, it does not just indicate an error. this is what fuel trims are. some cars have closed loop all the way to WOT and will trim, as we advance ECU's and use faster cpu's we're moving away from archaic open loop systems with precomputed tables and trims, to directly using the sensors to determine correct fueling an advance etc, the whole engine moves away from being an open loop system to closed loop, the sensors can feed back enough info to do this properly, if a major fault appears that is something else, but to compare backlash to a loose steering wheel is another thing entirely, but i feel like this comparison may have run its course but we're in agreement of what an O2 sensor does i believe, and why its a good thing.
trims vs compensation for a failure, two different things. the ecu will say, hold on we're getting a tad crazy with these trims here and throw a warning code to the operate, backlash compensation to me is the same thing, agreed tolerances and where it can, and can't work, it does work in some scenarios, not in others, not having it means it can never work.
you seem to be looking from the point of view worst case, i'm looking at from the pov of best case, in where it is a useful feature, sure maybe thereis some thing i can learn that i can write gcode or cam up so that i no longer need it, but i'm not there yet, to me i'm still in the realms of i can draw a circle that turns out to be an oval, i can draw on oval that becomes a circle, which of course is a gross oversimplification of it, but thats my pondering over coffee and not being an experienced machinist. I do get that experienced machinists say remove all the mechanical slop and stop dreaming about backlash compensation, but that takes a while to achieve, i'm ok with working within the limitations as i improve the build and race towards 0.
roger, we're doing an incremental build, and are at stage 1 as it were, with 7.62µ of backlash at the moment. the idea is to keep improving the parts we're making with each incremental build of the machine, and improving it, the goal being not to need backlash compensation at all, but for now its extremely handy.
we could be the guys in the pits arguing about splitfires and monster cables, or the guys on track drifting around corners in a beater.
-
roger, we're doing an incremental build, and are at stage 1 as it were, with 7.62µ of backlash at the moment. the idea is to keep improving the parts we're making with each incremental build of the machine, and improving it, the goal being not to need backlash compensation at all, but for now its extremely handy.
Sounds good to me.
Cheers
Roger
-
charliex, we do agree on somethings but you are clearly not getting what I am saying. That does not bother me in the least. What does bother me is others may see this and some will hear what they want to in your post and believe it.
if a major fault appears that is something else
This is almost funny. To a machine controller backlash is this something major. It does not get much more major than this. There is no degree of backlash only varying amounts. It either exist or it does not. Backlash is the distance the driver went around the course with the steering wheel disconnected from the shaft. No control! ;D Backlash is the additional length of wire needed so that the O2 sensor you keep talking about can be connected. Backlash is the length of fuel line needed to connect the fitting on the tank to the fuel pump. Backlash is the distance between the connecting rod and the crank when there is no rod bearing installed. Backlash is that tick you hear when a pushrod gets bent. Believe me charlie.............. backlash can be some serious juju.
I really am on your side, you just haven't figured that out yet. Do not underestimate the implications a "little backlash" can have.
-
So if you could show me at what point i said backlash doesn't matter i'd be happy to be on your "side" ? i'm pretty sure i've said exactly the opposite, and gave an example of it, just as i gave the example as to where it works for me. You cannot argue against that no matter how much you try, since all i'm saying is, in my specific circumstances/setup the backlash compensation is something i'd like to have, whilst also understanding why its a bad thing, perhaps not to the extent that you do, but thats ok since i know its bad and am trying to get rid of it. I also like the idea of screw pitch adjustment over the length of the axis.
again i'll repeat, i'm looking for a recommendation on a mach3/4+controller that supports backlash compensation, not if backlash is good or bad, people who read this thread and all the others will come up with their own line of thought regardless of whatever either of us says, thats just how it is, neither of us will live long enough to change that aspect of humanity.
and i quote
"but it works in some scenarios, not all, its a bandaid, fix your machine,we get it. but sometimes we can only go so far and having the option is nice."
and that option is when i'm end milling out circles. it does work for me, on my setup, in that case that is undeniable, i'm not going to assume that other people aren't also able to use it to their advantage in some way, maybe i'm wrong to have such an opinion.
cheers
-
backlash is a monster, backlash comp is a big monster get it wrong and it will bite you hard, A person on the yahoo forum spent some time on this, he got it in the end but it was hours and hours of hard work. he fixed the problem mechanically as best as he could first, before backlash comp was even tryed what was hours and hours of work.
is it worth spending hours and hours on it, and not cutting anything for days or weeks hell no.
-
charliex,
i'd like to be able to keep fitting bearings after an endmill
I used BC for a long time with MACH 3. Yes you can use the mill that has backlash and be able to just slip that bearing into what is cut,
but, you need to adjust both the BC settings and your code, and even with dong that,.....it's only good for a particular operation.
I don't know where BC implementation is relative to MACH4 or current external motion controllers. So can't help you there.
Take this advice:
Your machine is what it is, you either accept it's limitations or make improvements. Do yourself a favor and get rid of the
BC ( oh, BTW, there is no such thing as "0" backlash ).
RICH
-
RICH, "roger, we're doing an incremental build, and are at stage 1 as it were, with 7.62µ of backlash at the moment. the idea is to keep improving the parts we're making with each incremental build of the machine, and improving it, the goal being not to need backlash compensation at all, but for now its extremely handy. "
-
There needs to be a mention of what backlash comp can and can't do. Like a manual machinist it can take out the backlash that happens when an axis reverses. Like a manual machinist it can't allow you to climb mill no matter how careful it is. We had an acme screw Spindle Wizard knee mill that used a spring to turn a second nut slightly eliminating the backlash. It climb milled with no problem ever. The down side was that the considerable friction this introduced meant it was limited to 40 ipm.
-
with 7.62µ of backlash at the moment
That's 0.00000762" so how did you measure that?
Maybe your units are incorrect since that's almost impossible to accurately measure.
Using a mounted 500x microscope with fical micrometer as compared to a calibrated u line scale I had trouble
measuring to down to 0.000040".
If you preload axis thrust bearings to that you won't have movement / stall. There is no such thing as "0" backlash,
maybe min BC would be better term.
But improving to a practical min BC is certainly desirable.
FWIW,
RICH
-
Hey Rich
7.62 microns is very likely metric.
Cheers
Roger
-
that's nasa speck
-
As completely off topic as this thread went, does anyone know if they ever implemented Backlash comp for the ESS??? and as far as properly built machines not needing it, I work on $3 million dollar, Fanuc controlled multi spindle CNC machines for a living...and every single axis has Backlash compensation.
-
Fanuc controlled multi spindle CNC machines
Yeah, well, old machines with Fanuc controls probably do have backlash ... >:D
Cheers
Roger
-
Lol, They're brand new, Installed 5 of them this year already.
-
Banr yes it can be done it is a hell off a lot of work yes it does work but for a single person it will be from 10 to 100 hours, there is so much involved in getting it working the average person it's in the hard basket
-
huh? Getting the software to work, or setting back lash comp, cause that takes 30 seconds.
-
getting it to work as intended at any speed at any position on the table, sure flick a switch what is the backlash amount in the first places say it is 0.001 at x100 and 0.005 at x300 what is it
-
Backlash comp is there to take up 0.0001" to 0.002" of slop in the mechanics, ie leadscrew and nut, ballscrews. bearing preload, belt stretch, motor coupler tolerances. these things do not vary over the course of axis travel, Any mechanical mechanism has to have some varying degree of tolerance, "backlash" if it didn't, it would be welded tight and not move. If you have different amounts of backlash over different portions of axis travel, then you have have different problems, that can only be corrected by replacing the defective component. generally your leadscrew or ballscrew, or worst case scenario, your ways, or linear rails are wore out.
-
Any mechanical mechanism has to have some varying degree of tolerance, "backlash" if it didn't, it would be welded tight and not move.
Strictly speaking, that is not really true. You could have ball bearings with zero backlash. They might wear a bit more quickly of course. Such an idea is readily available, as double opposing angular contact bearings. As featured in many good lathe headstocks for instance.
Now, elasticity - that's a different matter. You could have zero backlash but some compliance due to elasticity, and it would then be a question of machining forces affecting tolerances. However, you could not use conventional 'backlash correction' for this case. In most cases such effects might not matter in practice, but when diamond machining large optics to lamba/10, it does.
Such fun.
Cheers
Roger
-
touche sir, touche
-
na hes someone who knows more than most phd and all.
you can get differences in travel does a 20 year old haas have the same backlash amount as it was new if not all of the table was used no it wont it will have a wear patch same as any machine
-
na hes someone who knows more than most phd and all.
you can get differences in travel does a 20 year old haas have the same backlash amount as it was new if not all of the table was used no it wont it will have a wear patch same as any machine
That's not acceptable backlash, that's a slopped out machine that needs to be rebuilt. there is a difference. and excessive wear patch is exactly that, worn out. it needs a new ballscrew and or rails. but apparently you need a phd to know that.
-
these is hobby not 2 million dollar kit there are barely any hobby machine or sub 100,000 machine with nasa speck that have a perfect surfaces over the length of the skew, rail, bed. compering a million plus machine from a billion dollar corp with teams working on one part of the machine, to M3 is compering lead to gold.
the amount of math to do backlash comp is massive and that is just working it out not the code then there's the code it's a undertaking in a half.
from a brand new machine sure backlash wont be that hard to work out in the + and - direction but if it's .1 or 0.01 that would be the parts, if it's 0.009 or lower sure why not get it lower.
a converted machine yer why not.
getting the + backlash and the - backlash you can use the average on that day 30 degrees ever side it will be different so easy no everything plays a part on what the backlash is at any given day depending on a lot of thing includeing the weather.
what the fanic guys would of worked out that's why they say warm the machine up for x amount of time before use. the specks will be at a temp and humidity
-
to M3 is compering lead to gold.
Ah ....
And Mach3 is the gold I presume? ;D
Mind you, I can't help feeling that the Mach3 (UCCNC, LinuxCNC) direction is the way of the future. Proprietary HW/SW combos (such as Fanuc) always lose out to 'Open' SW ime. Yes, it may take the 'Open' SW a little while to get there (budgets, budgets), but just look at every supercomputer and every web server out there. And I find those massive (Fanuc-style) front panels quite boggling compared to the simple GUI of the more modern systems.
I am told that if you want to add a new g-code command or a macro facility to one of the older systems, it can cost you thousands. With the PC versions you usally get everything in the first package, with no extras needed.
I'm raving, and diverting the thread. Sorry. To the best of my knowledge, the ESS does not support backlash correction itself. It would be a silly idea to try imho.
Cheers
Roger
-
to M3 is compering lead to gold.
Ah ....
And Mach3 is the gold I presume? ;D
Mind you, I can't help feeling that the Mach3 (UCCNC, LinuxCNC) direction is the way of the future. Proprietary HW/SW combos (such as Fanuc) always lose out to 'Open' SW ime. Yes, it may take the 'Open' SW a little while to get there (budgets, budgets), but just look at every supercomputer and every web server out there. And I find those massive (Fanuc-style) front panels quite boggling compared to the simple GUI of the more modern systems.
I am told that if you want to add a new g-code command or a macro facility to one of the older systems, it can cost you thousands. With the PC versions you usally get everything in the first package, with no extras needed.
I'm raving, and diverting the thread. Sorry. To the best of my knowledge, the ESS does not support backlash correction itself. It would be a silly idea to try imho.
Cheers
Roger
I agree with it being the way of the future, now if only some one cough..artsoft...cough,cough would make a specific motion control software package for controlling industrial robots. Used robots are dirt cheap, or scrapped once the proprietary controllers take a crap. being that they are the future, a guy could make quite a few extra bucks retro fitting new controls to them. and yes you could probably do it with mach, but it would be a pain, and no one wants to be programming robots in gcode in 2016.
-
being that they are the future, a guy could make quite a few extra bucks retro fitting new controls to them.
Reckon.
But not with Mach. For robotics you would need to take the mass/inertia of the arms into account to control the dynamics, and Mach can NOT do that. And will not, to the best of my knowledge.
Mind you, the computers used in some of the early (I think) Pumas were DEC PDP-11s, and they were a pleasure to program. Infinitely better than an X86. Support for them would be more difficult today, although there is a huge community to call on.
Cheers
Roger
-
M3 can't do that the what ever it's called function it does not have it there is a plug a dude in south america is doing that can do that multi axis at a time stuff.
the new motion control M4 is getting one day will be able to do that sort of stuff. it's one day budget and all. apparently Linux can do that robot stuff never seen it tho
we will have Auggie to added in soon to Roger it works quite well doing 3D stuff on my little router.
I had a go at that backlash idea years ago I read up on it and that was all that happened it was to much work getting the number's correct, I just got a expensive ballscrew, nut and bearings 0.001 is good enough for me over 100 mm
-
Hi Dude
Mach4 - someday, but not yet.
Auggie - interesting stuff, especially on the corners, but a long way to go too.
Q: is g-code really the right way to go? I believe there is Open Source Robotics SW around whcih handles this.
Cheers
Roger
-
yep your correct fanic I robot does it to.
Auggie is a cool little program slapping a chunk of C++ code in a G code is fun, it's almost a IOT machine controller, I could get it to stop at smoko and get it to make the coffee, then start again after smoko.
it does corners on the little machine a bit better than M3 could on 3D files.
Mach4 is about the same as M3 now just some bits missing it is a little bit smother you can hear it, and it is faster, I had to up the settings on the ESS and the look ahead, the backlash comp will be out soon on the ESS. what I don't need.
I will be going to M4 when the plug for the pendent I have is done it is good enough for me now but I can't use the pendent so I won't.
I think you will be able to use backlash comp with M4 as it is faster so it can keep up
-
Mach4 is about the same as M3 now just some bits missing
I'll wait. M3 does what I need at present. But I will watch.
Cheers
Roger