Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: Davek0974 on January 30, 2016, 10:23:19 AM
-
Been doing some tuning on the table and got to the Z axis, trying to decrease plunge time etc.
I had it set originally at 1200mm/min and 900mm/s/s acceleration, this seemed a tad low so I pushed it up until it screeched which was at 2000mm/min so i backed off to 1500mm/min. The acceleration went way up to 5000mm/s/s before a screech so backed that down to 4000mm/s/s which gives an estimated 0.3g and 744rpm measured with a tacho on the motor shaft. I think thats ok for a 3.1Nm Nema23 motor. Steps per mm is 1000, 3mm pitch screw direct drive.
This has given a much snappier Z action and normal repeated G0 Z50 / G0 Z0 moves showed no loss of position.
My question is, is there a way to simulate a THC move - these use no acceleration and is usually where the trouble starts??
-
Yes there is a way (thanks TP),
ports and pins - toggle emulation on and my system was already assigned hotkeys U & D
Is it best to set THC rate as high as it will reliably run or can that be too high??
-
As fast as it stays reliable and avoids occilations. ALSO check your setting in Scam for pluge rate for Z After the pierce. THAT helps with avoiding that dimple you may see on pierce. It does not help to have a fast Zdrive unless you actually use it (;-)
(;-) TP
-
Well, using the keyboard to bang the Z up and down I was surprised how lazy it really was!
I had 10% rate set and a max Z of 1300mm/min, Now I have 1500mm/min and upped the THC rate to 20%, It could handle 30% easily but 40% screeched and did nothing so I know I have a bit of slack still from 20 to 30%.
Now, this brings me on to a serious point that is baffling me - I have been bashing the crap out of my Z all morning, hammering it up and down with the keyboard, messing with tuning etc and it has behaved flawlessly all day apart from when being pushed too hard by my setting.
BUT, I put some code on and cut some tests and bugger me if the first time it hit a G00 Z3.8 for the pierce, it just sat and screeched, then next time it did IHS the rapid part of the G28.1 move screeched and was missed then it dropped at the slow rate all the way and carried on as normal.
This appears to be Mach3 not using the acceleration on the G00 move or the G28.1 move - is this a known bug/flaw/defect ??
I really do NOT think it is motor tuning as it has done it before and did not do it once using the keyboard to rapid the hell out of my Z axis.
Any ideas???
-
Well actually JOG uses a different process in Mach3 to move as compared to feed. Always TUNE/Test the axis with G00 or G01.
(;-) TP
-
I did, I made quite a few up/downs with G00 Z5 / G00 Z30 commands from the MDI line and it always worked well, no sign of a screech.
I also used my IHS calibration routine to run repeated sequences and it always worked.
When running code it screeched maybe once every 10 or so runs, now I come to think of it, I'm fairly certain it only did it when doing a G28.1 move.
I always presumed the jog with keyboard and shift simply moved the motors towards max velocity at the stored acceleration curve??
-
Spent the morning ripping my Z axis apart to replace the 10x2 screw & nut with a 10x4 combo.
This seems to have helped but my question is this - it now has the ability to be tuned from the previous 1500mm/min to 3000mm/min easily so should i take advantage of that or will make the problem return?
I currently have it set at 3000mm/min speed and 3000mm/s/s acceleration and it works fine during tests but have not done any cutting yet.
Steps per was a simple change from 1000 to 500.
It is a heck of a lot more snappy though :) I reduced the Z probing rate to 20% for the last portion of the G28.1 routine.
Another fantastic revelation was that because I have a probing subroutine now, I only had to alter the switch offset in Mach and it was back to it's accurate IHS in seconds - no messing with code changing etc :)
-
As fast as it stays reliable and avoids occilations. ALSO check your setting in Scam for pluge rate for Z After the pierce. THAT helps with avoiding that dimple you may see on pierce. It does not help to have a fast Zdrive unless you actually use it (;-)
(;-) TP
.... any suggestions on plunge rate for sheetcam to avoid dimple on pierce?
-
Spent the morning ripping my Z axis apart to replace the 10x2 screw & nut with a 10x4 combo.
This seems to have helped but my question is this - it now has the ability to be tuned from the previous 1500mm/min to 3000mm/min easily so should i take advantage of that or will make the problem return?
... this one is an interesting point that I'd be interested to know the answer to.
Your mechanical advantage has halved (TR10x2D to TR10x4D).... hence if the motor stalled at 1500mm/min..... are you sure that your machine will make 3000mm/min, simply because your leadscrew pitch has doubled .... but your mechanical advantage halved?
-
Sanity says i should cut the rate back down and take advantage of lower motor speed / higher torque etc, but I can ONLY make the fault appear when running code, not when manually jogging or using the MDI, I am fairly certain it has only ever done it when running G28.1.
I am guessing that it set at twice the rate and twice the pitch I am actually worse off - it was doing 1500mm/min on 2mm pitch now doing 3000mm/min on 4mm pitch.
I could split the difference and go to 2000mm/min maybe?
But why only stall when doing G28.1 and why only when running code?
-
Have you tried a z axis test script?
Say your travel is 5" (125mm)
set the z at 10mm from the bottom of travel, zeroing the z axis, and give it the following gcode (best done in excel by dragging cells down and then copying and pasting to text file and then saving as a *.tap file)
G00 Z100
G00 Z0
G00 Z100
G00 Z0
etc.
Then change it to
G01 Z100 F3000 (or something of your choosing)
G01 Z0
G01 Z100
etc
Then change it to:
G00 X100 Y100 Z100 (make sure you obviously have the travel distance, and all axis were zero'd at the start)
G00 X0 Y0 Y0
G00 X100 Y100 Z100
etc
This last one will test your power supply by moving all three axis at the same time as well as how well your drives cope....
You obviously want to make sure your machine is setup in its normal cut mode (i.e. torch on torch holder etc.... so the effects of shunting an axis about are correct).
-
Going out to try it now.....
-
Oh..forgot to say.. I sort of give it a motion of say 1000 lines (500 one way or the other).
If nothing trips out or stalls (my drives fault for a few different reasons and give an error output which I have linked to my estop) then I know I'm good to go.
Its quicker that using the MDI, one line at a time (you can use "teach" too.... but excel is quicker for me [drag, copy paste, save etc]
I'd do G02 / G03 as well .... when I feel confident doing MDI circles [one day get my head around doing them] .... after that there is not much else you could throw at a plasma cutter is there given the cutting force should be zero. (bit different on a mill with different feedrates and tools and materials being cut etc).
-
Well I gave it 100 repeats of each test, I only used a travel 0f 55mm on Z as it trips at 65.
At 3000mm/min in motor tuning, it passed 100%, I started out at Z1 with a feeler gauge on a test plate and after each series it was back at Z1 perfectly.
I tried 4000 in tuning but it was not happy and stalled after two repeats.
So using the 75% of stall setting rule I put it back at 3000.
I did try a couple of simple test cuts and it ran fine, no skips in the G28.1 anyway, also left the THC rate where it was before so that is effectively doubled now I guess and still seems happy, or at least I could not see any sign of see-sawing on the Z axis in the cut.
Did not try G02/3 as not enough knowledge there yet :)
Now need some jobs to run.
Does not explain why a different screw made any difference though, the only other change was that the nut is now brass and not delrin, don't know yet if thats good or bad though.
-
delrin coefficient of friction ~0.33
http://www2.dupont.com/Plastics/en_US/assets/downloads/design/DELDGe.pdf (page 33)
brass on steel (no lubrication 0.35.... add lubrication 0.19)
not sure what you would lubricate delrin with.
given it is a polymer, it obviously has creep and also some elasticity + a higher coefficient of expansion etc....
there is brass and then there is brass too.... phosphor bronze is a very "slippery" material.... then there is sintered bronze (I suspect oilite goes into this category... but not sure how good it would be at making "bronze nuts"...)
In short it will probably be better than delrin for slippiness .... but may not last as long before showing backlash (where as the delrin will probably mask this)
I have been looking using one of these for my new Z although probably the delrin nut and not the brass one.
http://www.engineeringtoolbox.com/friction-coefficients-d_778.html
If anyone has any experience with either I'd be interested to know a comparison in use / longevity.
-
luckily backlash is irrelevant as gravity will keep that in check for a plasma.
How long it lasts is anybodies guess :)
-
I don't know about backlash being irrelevant, as acceleration is very high under THC up and dn control, but the degree of slop may be small.
One of those time will tell moments as it probably will have little effect until wear becomes a big issue... Talking 1mm sort of movement not hairs thickness stuff (m8 nut + studding backlash)
How much travel does your z have endstop to endstop?
I've got about 135mm (5.3"), can't make my mind up if I need more with my redesign (175mm / 7")
Plan on putting a 4th axis below my table... Underslung ... So I can take the slats out and cut tube below ( about 1" to 4"). I have converted a 6" rotary table, but in hindsight... At 90:1 reduction ratio... It may not have been a good idea.... As getting 3m/min cutting speed will be hard (0.75 rev / sec on 1" 1.5mm thick stuff)... Anyone got suggestions on z axis travel or a good idea for a plasma 4th axis?
-
not sure what you would lubricate delrin with.
http://www.haydonkerk.com/LinearActuatorProducts/LeadScrewsAndNuts/KerkoteBlackIceTFELubricants/tabid/258/Default.aspx
-
Aghhh, thanks!
-
I have a total of 70mm Z travel, I wanted to keep it tight to help prevent wobble etc as it is usually cutting at near full extension.
A 4th axis or tube cutting was not a design consideration so adding one would be hard.
Yes I may have overstated the irrelevant part of backlash but with about 1.7kg on it, the z would have to get very stiff i think before it stopped responding to small moves due to backlash.
-
What you are trying to acheive is lowering teh normal RPM range of teh z to push it back down into teh higher torque range. Doing that you can acheive faster accel before loosing control of the stepper.
If neccessary you can lower teh G28.1 speed in Homing Config.
Always test and tune in G00 mode Then rerun test under Gcode/THC control to verify it all. The first signs you run into with teh Tuning Z is Z will start to loose steps IF you have gone too high. On teh other end IF you get too fast for teh THC you will see Z occilations .
(;-) TP
-
So i should put the speed back down and take advantage of lower motor speed but increase the acceleration rate?
Currently 3000mm/min and 3000mm/s/s
The speed in homing for Z only adjusts the final stage - the slow bit before contact, it was the first rapid stage that had the issues - like it missed to use acceleration sometimes.
-
Rememeber that you are moving the Z under THC control in VERY small moves so having a high VEL is useless as you are NEVER going to get there with a normal cutting move.
On the other hand you are using FAST Accel with EVERY move (;-) so it really becomes the more important element with plasma.
-
Dave,
Remember the THC Up and Dn Acceleration is near infinate (it is shunted / slapped) to whatever your percentage of rapid velocity you have set for the THC
If you look at post #3 on page one...
You listed your maximum (rapid velocity as 1500mm/min) and the THC setting as ~20 to ~30 %...
Say its 20%.
this means your THC slap and go velocity is 300mm/min (5mm/sec) .....
Given acceleration = (Velocity initial - Velocity final ) / Time
In post 6 you gave the steps / mm as 500
hence the time to accelerate from 0 to 5mm/ sec = 1/500 of a second...
You acceleration is 2500mm/s/sec.... for THC
(providing your z axis does not miss a step or two in the process (note 1 step is 1/500 of a mm..... if you can measure that I'll be impressed .... but if it was to miss a step or few the error would be cumulative.... hence if you had a way to toggle the z axis up and down to match the THC UP and THC DN signals.... you shunted it up and down 1000 times (and it missed a step on every cycle) you would accumulate an error of 2mm... which can be a little more visible
But I've read that steppers don't tend to loose one step at a time but the step loss tends to be greater or the motor just stalls.
(can't find the post now, but it was on here regarding encoders, closed loop and steppers missing steps about 5 years ago)
Also you will note that the 2500mm/s/sec acceleration correlates with your Mach3 target smooth curve acceleration nearly of 3000mm/s/sec..
Hence you could probably up the THC % or rapid speed to ~24% (if your rapid speed is still the same as post #3 at 1500mm/min)...... however at 3000mm/min.... your THC speed setting will be at 12%
Try it see if your THC works at 15% / 20% etc if your rapid speed is 3000mm/min.... but according to the calc above it may not.
______________________________________________________
If you had an arduino ..... you could always programme it to toggle two outputs and interface it with two pins on your motion controller, swap your THC pins around and try it.... (I use an ESS so have a load of pins but dod thing about this as a potential THC UP and THC DN testing philosophy)
-
So reduce speed again, increase acceleration and check it under THC control?
Back to 1500 again or compromise at 2000 maybe?
I can use the keyboard as THC inputs again and just flick the U/D keys.
-
After reading Terry's post above....
if the maximum acceleration you can get before stall is 3000mm/sec/sec.... and you are running 500 steps / mm
Should you not be better setting your Z axis rapid velocity at say 500mm/min.... and your acceleration at 3000mm/s/sec
That way your THC Speed setting for THC Up & Dn motion would be ~72% of rapid z axis motion...
(3000 mm/sec x 1/500 ) * 60 / 500 mm/min
Hence you can always tune your setting a little more (72% is an easier number to adjust by 1% up or down than 12% is as a 1% change at 12% is 6 times greater than a 1% change at 72%)
...... Sorry I hope the maths logic makes sense it may take a few times to read and understand what I'm getting at .....
-
I see what you are getting at - reducing Z max speed to increase THC rate ?
I can see that might be worthy of testing, will slow probing down a tad but whats a second here or there???
-
ESS verses teh LPT. NOW you are talking apples and oranges As teh ESS has to handle teh THC internally in its on way.
When you retune for best performance you normally leave teh VEL where it was because it was stable at that point. Then you run teh ACCel up to teh fail point and back it down 25% and retest teh THC function.
IF you are chasing teh divot on pierce. You have several things to work with . Normally nothing can move unitl you get teh Arkok signal. Then normally you move DOWN in Z to teh cut height. You can mkae that a GO move instead of a G1 OR speed up teh G1 move to maximum.
OR seeing how you are cutting thin guage metal simple make teh pierce height eh same as cut height and eliminate teh Move to CH move all together. That way as soon as teh Arcok signal is seen XY are moving.
For ME here I override teh Arkok signal ( turn it ON all teh time) so as soon as teh torch gets teh fire signal teh torch is moving. That give teh ark no time to burn a divot. NOW that does not work well with thicker metals as you DO need teh dwell to fully pierce. But with thinner sheets the ark start and pierce completion are almost instant.
-
Dave, yup, that's what I was getting at.
Terry, as always, thanks for the tips and guidance.
-
I forgot about the divot issue - a good reason not to reduce the Z speed i think.
Will have a play with this tomorrow
;)
-
The pierce divot is ONLY an issue cutting open lines where you cannot use a leadin (;-). With a closed shape and a leadin who cares about a divot ? it falls out anyway (;-)
-
Dave, what stepper drives are you using?
At the moment you are using 10ustep resolution...
500 steps per mm... 1u step = 0.002mm
Suggestion to try (if you are up for an experiment and its easy for you to try...)
Change you z axis usteps to 4
This will give you 200 steps /mm (0.005mm per step)
If you can still command 3000mm/s/s acceleration before stalling
Using the above acceleration formula
(3000 / 200 x 60 ) / 3000 (if that is what you want for you rapid velocity), means your target THC % or rapid speed will be 30%
That way you get THC rapid % range (30%)... And Z axis rapid motion speed 3000mm/min
I'll work out the time difference later dropping from say 3mm pierce height to 1.5mm cut height... But the difference should be very small
What is your retract height before touch off (25mm?)
-
Using CNC4YOU's digital stepper drives on this one, yes all axes are set to 10micro-steps.
I can easily change the Z as very accessible.
I take it steps per would be 200 then?? ((500/10) x 4)
Yes Z retract is 25mm, Z slow limit for G28.1 is 8mm
-
Yup, 200 steps at 4usteps.
4mm pitch / 200 full steps / 4 usteps = 0.005mm/step
1/0.005 = 200 steps per mm.
Make sure its all off electrically before flicking dip switches.
First try it at 3000mm/s/s acceleration...then try increasing it to see if the stall point has gone up (or down).
If down, your fallback is just to go back to what you had.
Nothing lost but a bit of playing around time.
You can then try to find your THC up / dn sweetspot.
Try thcspeed at 30% if Ok try at 40%, if fail, try 35% if Ok, try 37%, if Ok try 39%......sort of split the difference ( must have a name for it ), then split again etc.
You at least know the maximum setting and can adjust down to prevent overshoot (if you end up with that)
-
First try it at 3000mm/s/s acceleration...then try increasing it to see if the stall point has gone up (or down).
What speed though?
-
That is an interesting thought and worth testing BUT normally you are up against the Physical limits of the motor not teh electronics that drive it.
You can normally adjust teh feedrate of teh screw itself to a point where you can easily outrun the THC responce rate. It will be somewhere around .5 - 1"motion per rev of the motor
(;-) TP
-
Well it works for you at 3000mm/min... And you are happy with that?.... Start with that (I.e just change the ustep dip switch, plus z axis steps per mm from 500 to 200 and see how it runs... Then increase your acceleration (and see what you get shunting the z using the keyboard THCUp and THCDn buttons you've setup.
If 1500mm/min is what you had before changing the leadscrew and you know what divot you ended up with that... Then that is your ultimate cannot go worse than that point.
Terry had an interesting point about the THCOK before motion.... Especially on thin stuff and high amps as some of the hyperthem cut tables have zero or near zero for pierce delay....hence not sure if you can decend to cut height on the move (never tried it.... Still waiting for my new machine to arrive)
-
Interesting, flipping to 4micro-step gave some odd results.
200steps per, 3000speed, 5000acc = G00 moves pass, G01 F3000 fail
200steps per, 3000speed, 4500acc = G00 moves pass, G01 F3000 fail
200steps per, 2500speed, 3000acc = G00 moves pass, G01 F3000 pass, THCrate at 25% pass
Not sure really what its telling me, it seems stable at the final settings so i've left it there for now.
Not sure why G00 worked but G01 failed??
I hammered the THC inputs and it returned home every time so no lost steps there at THC rate 30% it did lose a few steps.
The motor seems to whine a bit more at 4micro steps.
-
Well, that has confused me if g00 passes and g01 fails.
Interesting you got the acceleration higher up to 5000mm/s/s