× Forum Header

stepConf and actual running precision issue

More
08 Sep 2016 13:19 #80222 by racedirector
racedirector replied the topic: stepConf and actual running precision issue
What Mod rack are you using? If it is Mod 1 or the imperial equivalent I could plug your values into a spreadsheet I have and give you your scale. If it is the same as I use (Mod1, 30T pinion) give me you pinion teeth, micro step value, any reduction your are using (2:1. 3:1, 4:1) etc and I'll see what it comes up with.

For example, as above, I use Mod 1 rack, 30T pinion, 8 micro step and 4:1 belt drive. My scale is 1724.81516993457

Let me know.
The following user(s) said Thank You: fannlow
More
08 Sep 2016 14:52 #80224 by fannlow
fannlow replied the topic: stepConf and actual running precision issue
I will test your recommendations on Sat.

thanks Tommy
More
08 Sep 2016 14:56 #80225 by fannlow
fannlow replied the topic: stepConf and actual running precision issue
Hi Tommy,
for the same settings in StepConf and hardware.

One motor runs exactly the same, while the other is running way too much.

not sure if that helps?

that means, with the 50mm test, I have one motor (U axis) running exactly that.
while the other (X axis) runs more than that.

but they both have the same settings on stepConf and hardware driver side.
More
08 Sep 2016 14:57 #80226 by fannlow
fannlow replied the topic: stepConf and actual running precision issue
Hi John,
I did check too. previously i was going at 28,000mm/min on both my X and U axis.
now I just travel at 3,000mm/min for testing.

both steps are at 2000/rev.
More
08 Sep 2016 15:00 #80227 by fannlow
fannlow replied the topic: stepConf and actual running precision issue
Hi Race Director,

I will update you shortly.
I am running direct drive, so no ratio for pulley or anything.

I am running 200 and micro is 10 for X and U, while 200 and micro is 4 for Y and V.

Yes, I am running Imperial equivalent, as I get those racks from China, I would think they are imperial.
I will check the teeth per rev and update later.

or you can also send me the excel spreadsheet or where can i download it?

thanks
More
06 May 2017 21:03 #92755 by probamo
probamo replied the topic: stepConf and actual running precision issue
Not to open another Topic, but to use this one for more-less the same issue. In fact, almost finished building CNC machine, this time with rack and pinion mechanism on X - axis (two motors engaged at either sides of the gentry) and on Y- axis (used to move spindle across the gentry) and ball screw nut for Z-axis with the pitch of 5 mm.

LinuxCNC 2.7.8. Setting of Z-axis, based on previously set machines was easy. Motor with 200 steps, 4 micro-steps set the driver, "motor/screw" ratio 1:1, pitch 5 mm, speed set at 100 mm/s, acceleration 300 mm/sec2 working very well, though I did not check for higher speed.

For X- and Y-axis, I used the same size pulleys. Attached to the motor shaft has 20 teeth, driven has 60 teeth, which gives "motor/screw" ration 1:3 in stepconfig. Pinions are attached to the driven pulleys accordingly and have 20 teeth each, mod 1. Calculation says, each motor revolution, lineal movement should be 20*3.14159 = 62.8318 theoretically. This figure I put as pith value in stepconfig. Having measured by tape distance I wanted to get and to be achieved by machine, I have made tuning by taking before-said theoretical value of pitch 62.8318 times achieved movement of machine, divided by desired length. New figures calculated and put as new value for a pitch. I repeated steps as much as I am happy with accuracy. That was not so difficult. In addition I set the velocity as much as 300 mm/s, and acceleration 1200 mm/sec.

Now, here is the issue. Low value for micro-stepping, low resolution, rough movement of machine at the visible high speed. I increased micro-step values from 8 to 20, even to 40 and adjust drivers accordingly. The more micro-stepping the smoother machine run, which is normal. I have no problem with lost steps, momentum of the motors were pushing my body away (motors are 13 Nm). However, the more micro-stepping, though not changes made in velocity settings (300 mm/s), I have filling the machine is getting slower. When I had 40 micro-stepping and come back to 20, I have filling the machine was speeding 2 times higher. Speed of 300 mm/s is 18 m/min, machine should "jump from corner to corner", but that was not happening.

Comparing Z-axis movement at 100mm/s and X- or Y-axis with 300mm/s, Z is faster.

Is there anyone that had the same or similar issue. What I made wrong, any advise?
Thank you.
More
07 May 2017 19:11 #92784 by probamo
probamo replied the topic: stepConf and actual running precision issue
Found issue. Though I have filling machine would go rocket speed, I am forced to reduce the speed.

If am not right hereafter, please correct me, not to confuse community, but this is what correlation I've found today.

Setting Jitter at 12000ns (based on Latency), gives ~35 kHz available, that computer could communicate with external, e.g. machine (the first screen in Stepconfig). As low jitter, the more Hz are available. No one single axis should exceed available Hz, which is in relation with the maximum speed of the axis and chosen micro-stepping. In my case I set the speed for all axes 120 mm/s, and for X- and Y-axis set 32 micro-steps which gives ~33 kHz, and for Z-axis at 4 micro-steps gives ~20 kHz.

By figures combination for micro-steps and speed I could increase/decrease overall velocity of axes as well roughness/smoothness. If I set more speed for certain micro-steps, e.g. exceed available Hz, for instance 32&150mm/s gives beyond 40 kHz. When I started machine, and make jogging at full velocity by holding down Shift key, movement stopped, saying the limit switch for the axis is being activated. Most probably was lack of Hz and communication between comp and machine, causing stoppage of machine.

If all above said is right, then question is what computer to be used to provide more available Hz, hence to get more speed at more micro-steps, hence has smooth running over the table?

Thank you for any comments.
Dejan
More
07 May 2017 20:28 #92788 by tommylight
tommylight replied the topic: stepConf and actual running precision issue
I have a Fujitsu Siemens workstation that can do 65kHz under ubuntu 10.04 Lcnc, also have an Asus P5B Deluxe with core2quad that does 50kHz. All others can do 20 to 35 and no more.
All in all it is a hit and miss.
One advice if you vant better latency from the existing computer, start debian wheezy iso from usb, usualy i get 20 to 30% better latency from usb than an installed system.
More
07 May 2017 21:02 #92789 by rodw
rodw replied the topic: stepConf and actual running precision issue
The other alternative to consider is to use Mesa hardware like the 7i76 or 7i76e. This eliminates the base thread which is required for your current step generation. Step generation is offloaded to the 7i76.

I use the 7i76e ethernet version but had to upgrade my PC to run the mandatory Linux preemptive real time kernel. I have tested it through to 90 khz with entry level steppers and a Gigabyte Brix J1900 USFF PC. I think I could have pushed it further if I had played with the step timing parameters. The 7i76e is capable of stepping at 10 Mhz but there are other limits that will come into play before you get to 10 Mhz.
More
07 May 2017 22:48 #92793 by tommylight
tommylight replied the topic: stepConf and actual running precision issue
Just did a test on the afore mentioned Asus with an PCI addon parallel port card (has no onboard), it can do consistently 66.4 kHz and throw a joint error at 66.6 as it should, base thread is at 15000.
Occasionaly geting the real time error, but this is on a Lubuntu 16.04 with Linuxcnc 2.8 RTAI ( thanx to CncNoob for his great tutorial on installing it on Mint, works for Lubuntu also ).
I did test on a 7i76E, pushed it to 0.96 MHz and called it a day, had to go to a weding. I am sure it can do much more, but i can not think of anything attached to it that could follow, the best stepper drives i ever had could go to 300kHz, and they are the best drives money can buy, namely Lam Technologies, and they do have even better than the ones i had.
www.lamtechnologies.com/Product.aspx?lng=EN&idp=DS10
P.S.
I have nothing to do with them, but if i intentionaly try to kill their drives by doing everything that should not be done to drives, like reversing the polarity of the power supply, cross wiring the motors etc, and they did not fail, well i will gladly pay for them, install them and forget about them ( over 4 years of everyday use for the first set, 24H a day ).
Time to create page: 0.073 seconds
Powered by Kunena Forum