Backlash compensation?

yurirage
Posts: 13
Joined: Wed Sep 19, 2018 7:25 am

Backlash compensation?

Postby yurirage » Fri Sep 28, 2018 6:18 am

I've measured my machine's backlash very accurately using a dial test indicator, however, the "Cuttion Backlash" settings in LaserSoft 7.21 do not appear to actually apply backlash compensation. When the file is sent to the T9 controller, the geometry is simply either expanded or shrunk by the amount entered in the backlash menu under "Crafts Parameters." This does not effectively mitigate backlash at all.

Is there another setting either in the T9 controller itself, or hidden elsewhere in the menus that will apply more effective backlash compensation?

Or, is there a plan to update the LaserSoft software to apply true software backlash compensation?

--Yuri

Tech_Marco
Posts: 4350
Joined: Mon Jun 15, 2009 3:00 pm

Re: Backlash compensation?

Postby Tech_Marco » Fri Sep 28, 2018 11:40 am

I'm trying to figure out what is the "Cuttion Backlash". Waiting for reply from the programmer
As I was told, you're using "Carving backlash' to make compensation. The other way to improve the backlash is to change the belt and tighten belt pulley a little bit.

Marco

yurirage
Posts: 13
Joined: Wed Sep 19, 2018 7:25 am

Re: Backlash compensation?

Postby yurirage » Fri Sep 28, 2018 3:33 pm

Marco,

It appears the "Carving Backlash" is a single axis, speed oriented offset to be used with raster engraving (which appears to be a common setting in most laser cutting/engraving software). If I understand it correctly, its intent is to minimize double imaging, tuned to a given speed.

There should still be a way to account for 2-axis backlash compensation during cutting operations. I assumed that "Cuttion Backlash" was intended for that purpose.

I did tension the belts during my backlash measurement. In fact, I tightened them incrementally until backlash ceased improving, then backed off the tension to the minimum required to maintain that minimum backlash measurement. As such, I should be able to achieve maximum speed (no missed steps) with minimal backlash. Still, I'm stuck with 0.075mm on the X axis, and 0.25mm on the Y without software compensation.

Definitely looking forward to further response.
--Yuri

Tech_Marco
Posts: 4350
Joined: Mon Jun 15, 2009 3:00 pm

Re: Backlash compensation?

Postby Tech_Marco » Fri Sep 28, 2018 3:47 pm

I never head that backlash happened to the Y axis. Usually it happened to the X axis during high speed engraving with high acceleration. How fast did you run on your job? What kind of machine do you have? Send pictures to show your gantry and motors. 0.075mm is really small on X axis. But 0.25mm is quite a bit, though. It seems that you may have 'escaping' rather than backlash. You may need to drop the speed because software backlash isn't an ideal solution. In this case, you may want to pay attention to the motor & drivers. Plus, a better quality of belt should be in your shopping list.

I'll post an update once I found out that what is the "Cuttion Backlash". But again, backlash should not happen to the Y axis and it is abnormal to see the value on Y is greater than the X. Strange. :ugeek:

Marco

yurirage
Posts: 13
Joined: Wed Sep 19, 2018 7:25 am

Re: Backlash compensation?

Postby yurirage » Fri Sep 28, 2018 6:11 pm

Marco,

I'm using your PR530 XY stage, as delivered.

During a raster job scanning in the X direction, Y backlash would never be a factor, as it always moves in the same direction. However, during a cutting job, the laser head has the potential to reverse direction many times in each axis, each time inducing a backlash error in the reversing axis.

I'm measuring static backlash, just like I would on a mill or lathe. In other words, I move the laser head toward a fixed dial test indicator to ensure that I've taken up all slack and zero the indicator. I then move another 0.3mm in the same direction, followed by a commanded move of 0.3mm in the opposite direction. The difference between the 0 reference and the machine's position after the completion of the moves is the measured backlash.

The attached file shows the effect of uncorrected backlash during a cutting job. The circles are 3, 5, and 10mm wide, each cut at a fairly slow 50mm/s. If you look closely, you can see the offset at each axis extreme, much more pronounced in the Y axis (as I would expect based on measurements).

I've used several CNC controllers (FANUC, Mach3, LinuxCNC, and even Arduino-based), all of which have the ability to apply software backlash compensation. I have yet to discover the equivalent feature in LaserSoft.

--Yuri
Attachments
backlash.jpg

Techgraphix
Posts: 484
Joined: Fri Jan 20, 2012 1:39 pm
Location: Appelscha, the Netherlands
Contact:

Re: Backlash compensation?

Postby Techgraphix » Sat Sep 29, 2018 2:53 am

Backlash is always a shortcoming of mechanics, IMHO.. You should first solve that.. I made several lasers with absolutely no visible backlash.
Even engraving a vertical line of 0.05mm is straight and continuous. Only under a microscope of 40x magnifier I can see in which direction was engraved (bidirectional engraving 500mm/sec)
- Check the belt-tension. Not much slack but also not too tight
- Check the motor to pulley(s) if they are fitted tight. Absolutely no freeplay is allowed
- Check the linear guiderails for dents, "sticktion" or even loose bolts

If the belt is too tight, your motor can have problems to move it. If you don't use closed-loop systems you can have steplosses then.

Also check if the head is fixed rigid (i.e. you can not move it in any direction. The same for the bridge. You can imagine that when you pull on one side it might tilt a bit to one direction. When you pull to the other side it will tilt to the other direction.

If it is done all right, i'm sure you won't have any visible backlash..
To compensate for backlash also has its disadvantages, better get the mechanics right..

Kees

yurirage
Posts: 13
Joined: Wed Sep 19, 2018 7:25 am

Re: Backlash compensation?

Postby yurirage » Sat Sep 29, 2018 5:22 am

Kees, Marco,

Thank you guys so much for your input. I did readdress the mechanical backlash in the machine this morning and achieved much better results. I have sub-0.1mm backlash in both axes now, which, if uncorrected, is hardly detectable by eye except in the most minute of details.

Excellent tip, Kees, about engraving a line. I drew three parallel vertical rectangles (LaserSoft requires closed geometry for engraving), 10mm high by 0.1mm wide and then engraved at various speeds to test backlash and X-axis belt tension. The alignment of the engraved dots in each line is an outstanding visual indication of backlash. I did the same for the Y-axis, using horizontal lines and Y-axis scanning.

HOWEVER, some amount of backlash is nearly inevitable in any mechanical application of power. While I agree that software compensation is a last resort, it should be an available tool, and I look forward to Marco's response from the programmer.

--Yuri

Tech_Marco
Posts: 4350
Joined: Mon Jun 15, 2009 3:00 pm

Re: Backlash compensation?

Postby Tech_Marco » Sat Sep 29, 2018 2:58 pm

The PR530 machine or XY stage can do better than image you posted without backlash compensation. I need some info from you
1) Motor driver micro-stepping

Assuming you're using one of DSP controller card we're selling:
2) Manufacturer parameters
3) User parameters

4) Speed of doing cutting/engraving

Other things you need to pay attention
Do not put too much or too less tension on the belt. Check pulley or axis coupling to make sure it is secure tight. Check belt balance between left and right belts on Y axis.

I don't know too much about CNC but so far I don't see Y backlash compensation on any laser controller software that I have deal with; Ruda card, Anywell, PH... If you're talking about precision, well I admitted that laser is not as precise as a CNC router because laser in general is belt driven while a CNC machine is ball screw driven. In general, belt drive machine the most can reach is roughly 0.05 ~0.08mm while a CNC can be as precise as 0.01mm. Of course, for high end industry CNC, it can reach even better precision but only if you're willing to spend $mil :lol:

By the way, I talked to the tech in China and he told me what was "Cuttion Backlash". In fact, it is like what you said, an offset feature (expand or shrunk) in geometry with X&Y in proportional. Or simply say that it is just a 'scale'

Regards,
Marco

yurirage
Posts: 13
Joined: Wed Sep 19, 2018 7:25 am

Re: Backlash compensation?

Postby yurirage » Sat Sep 29, 2018 3:43 pm

Marco,

I have the drivers set to microstep at 3200.

I'm using a DSP T9 controller (hence the post in this forum). The attached images show the parameters you requested. I admit that I may have the acceleration turned up a bit high, and I'm certainly willing to accept feedback on the parameters. It took me quite a bit of time to achieve what you see there, and I'm sure the settings aren't quite optimized since I started from scratch on unfamiliar software. I just noticed that the X um value is cut off - it's 15.939.

The previous picture showing the circles was cut at 50mm/s. As I mentioned in my last post, I did manage to significantly improve the backlash by paying some more attention to the belt tension and doing some raster testing at various speeds.

Thank you so much for the info on the "Cuttion backlash." I assume it was an attempt at software backlash compensation, but it fails miserably in execution. I'll leave it at zero unless I can find a very good reason to use it for scaling.

Thanks again for your attention - I really appreciate the help,
--Yuri
Attachments
user-param.png
user-param.png (40.77 KiB) Viewed 178 times
y-factory.png
y-factory.png (17.44 KiB) Viewed 178 times
x-factory.png
x-factory.png (14.34 KiB) Viewed 178 times

yurirage
Posts: 13
Joined: Wed Sep 19, 2018 7:25 am

Re: Backlash compensation?

Postby yurirage » Sat Oct 06, 2018 11:56 pm

Marco, any feedback?

Tech_Marco
Posts: 4350
Joined: Mon Jun 15, 2009 3:00 pm

Re: Backlash compensation?

Postby Tech_Marco » Sun Oct 07, 2018 11:55 am

Your settings of acceleration for X and Y were way too high!
X: 10,000
Y: 5,000

Also, try to use 5000 for micro stepping


Marco

yurirage
Posts: 13
Joined: Wed Sep 19, 2018 7:25 am

Re: Backlash compensation?

Postby yurirage » Mon Oct 08, 2018 9:44 am

Marco,

Thank you. I appreciate you taking a look. I was playing with the accel settings to see if I could find one high enough to skip steps. Even with these settings, I’m not seeing skipped steps, though I don’t think the controller even accepts these values and uses some default value instead.

5000 for um? The Leadshine drivers are limited to 1600, 3200, and 6400. I used 3200 based on the video about the PR530 table posted by LO’s YouTube account.

—Nate

Tech_Marco
Posts: 4350
Joined: Mon Jun 15, 2009 3:00 pm

Re: Backlash compensation?

Postby Tech_Marco » Mon Oct 08, 2018 10:03 am

5000 or 6400 for the micro stepping is just fine

Marco


Return to “DSP T9”

Who is online

Users browsing this forum: No registered users and 1 guest