I am having the exact same issue. 1.4.7 with SLB. Having fixed the previous issues thought I would try a test job today, roughing cut went fine, change to fine ball nose First I loose x axis zero to the left. If I let job run it shifts left then after a few lines shifts right a few more lines then left again.
I am very confident that my machine mechanically is well setup. I can jog around all day at top speed, no binding noticeable and will return to a set zero without issue. Load a program and will not keep zero.
Changed max speeds for x and y in firmware down to 4000 with no effect on issue. Today I have measured X0 out from 3mm to as much as 95mm.
Other issues experienced today, gsender disconnected and only way to reconnect is to reboot SLB, Jog buttons sometimes not working even though they show active. Or they work incorrectly i.e one press for 10mm movement may give 10mm or may give up to 50mm.
Prior to all these issue I was using regular longboard with 1.2.2 and everything working fine. I upgraded to gsender 1.4.7 a couple weeks ago and only had minor issues related to limit switches. Upgraded to SLB last week and all hell has broken loose.
Okay…the update…The problem has been resolved. Here’s what happened.
As mentioned I did all the maintenance aspects I could on the machine to ensure it wasn’t something simple.
As I did the trial runs, I came to the conclusion that it wasn’t a software issue but a hardware issue. I narrowed it down to the Anti backlash nuts. I did everything possible with the original ones to see if the problem would be resolved. In some aspects it was but it wasn’t 100%. And I need 100% . Some of my carving runs are 14 hours long. I already lost 2 runs with almost 23 hours into them. Talk about frustration!
So I ordered the new upgraded anti backlash nuts from Sienci. Since I live only about 90 minutes away from them, I chose to pick up the parts from their facility, and maybe talk to one of their experts to confirm that indeed this was my problem.
At Sience, I talked with Jason. He has done a number of videos for the machines and his breadth of knowledge is outstanding. He did confirm that my issue was the anti backlash nuts. So I was off to install the parts and get back into production.
He did state that I would probably need to run my x axis back and forth to set the new T12 anti backlash nut, as this part can be a little tight. No problem. I ran the machine in the x axis for several hours.
The y axis and z axis have no issues. However, I was still getting binding issues especially at the travel ends. This was most always in the rapid mode. To say this was frustrating…
I did continue to run in the part and had about 4 hours of run in time. Problems persisted.
I contacted Sienci and Jason responded.
His words below.
A couple of other items to look at in the firmware settings that can help with the rapid movements especially with the SLB.
Some customers have been experiencing motion issues with the faster default speeds with the SLB, we recommend lowering the speeds to the LongBoard default rates. Please take a look at the recommendations below. To update the firmware, make sure it says IDLE in the visualizer.
Open the firmware tool. Scroll down to the following and input the following numbers into your X, Y, and Z maximum rates.
110 x axis rate 4000
111 y axis rate 4000
112 z axis rate 3000
120 x axis acceleration 750
121 y axis acceleration 750
122 z axis acceleration 500
Click and apply new settings
We have had a couple reports about this and believe there is a setting in your Firmware EEPROM page that needs to be updated. Can you navigate there and update your $26 from 25 to 250 ? Once you change the number, you can hit tab or enter or click outside of that box, then hit the Apply New Settings button for the changes to take effect.
If you have one option enabled in the firmware that prevent the ability to jog the machine using the corner buttons.
Open the firmware and scroll to 23 Homing Direction invert . Disable the A-axis.
Me again… Once I made all the changes and ran the machine my binding issues went away. I have run the machine an additional hour watching for any binding issues and none have occurred. That’s good enough for me.
The only negative aspect of the above is that the speed gains that have been made available with the new SLB controller are now back to the original speeds. However, if this means that I can back into production, I’ll take the slower speed.
So today I installed regular longboard and 1.2.2 same speed as I had been running yesterday with SLB, 4000 for x and y, only difference was acceleration on SLB was 1000 and on regular longboard today 750. Ran the file at same speed I was attempting to run yesterday albeit with lower acceleration and it ran flawlessly. Total runtime was 5hr 23min, estimated yesterday with SLB was 6hr 15min, but I suspect it would probably have been quicker with higher acceleration rate, I have noted times in version 1.4.7 bear no resemblance to reality. So now I am not sure if it makes any sense to me to try the SLB with 1.4.7 again or stay as I am as if I cannot get an increase in speed I see no upside at the moment.
Good follow up, glad you’re back up and running. I too have gone back to slower speeds. I find that I seem to have better cut quality and less tear out 
Thought I would reinstall 1.4.7 and SLB to investigate speeds. Started running same file as yesterday in air and was getting multiple grbl error 24. I was using same laptop usb c port and connected to slb usb, tried different cables and different laptop ports then noticed every so often gsender would flash “reconnected to server”. I reconnected with a usb c dongle with ethernet and issue went away. Also jog buttons in 1.4.7 as noted yesterday do not work correctly in any direction. I then proceeded to run same file as yesterday in air starting with x y speed 4000 and acceleration 700 ran fine. After 2 hrs made adjustments to speed and acceleration eventually getting to 5000 speed and 1000 acceleration without loosing zero. Decided to try in pine as yesterday using 1/16 TBN cutting at a depth of 3/4" same as yesterday. Found increasing speed from 4000 to 5000 was OK without change to zero. Increasing acceleration to 800 causes a stall on the x axis. It is like the motor does not have enough torque/power. Are there settings for the drivers that can be checked?
@Mickus @Skyboltguy1 @RasterTheGrain I have created a new topic in the SuperLongBoard category and moved your posts here. The Sienci SLB team monitors this category quite closely and I believe having your posts here will bring the issues to their attention.
Usual preface, I’m with PreciseBits. So while I try to only post general information take everything I say with the understanding that I have a bias.
I don’t know anything specific related to the SLB or the rest of this hardware. However, there are physics issues. It’s more stress on the machine, tool, and steppers/servos the higher the acceleration. There’s also possible issues with “smooth” motion at a feed/voltage/current/controller ranges in some steppers. You WILL see effects on your cuts by changing the acceleration. This is enough of an issue that we only recommend a 20in/s/s ( 508mm) for use with nano or micro tools (rule of thumb, changes with machine/tool specifics). In systems with belt drives or low relative rigidity we’ve often even had to lower this to prevent tool breakage.
Just to give you a rough idea here’s what you are looking at for 500mm, 750mm, and 1000mm.
-
500
- Feed reached in 0.039"/1mm of travel: 75IPM/1897mm/m**
- Feed at a 10th of a second: 118IPM/3000mm/m**
- Distance moved in a 10th of a second: 0.098"/2.5mm
- Feed reached in 0.039"/1mm of travel: 92IPM/2324mm/m**
- Feed at a 10th of a second: 177IPM/4500mm/m**
- Distance moved in a 10th of a second: 0.148"/3.75mm
- Feed reached in 0.039"/1mm of travel: 106IPM/2683mm/m**
- Feed at a 10th of a second: 236IPM/6000mm/m**
- Distance moved in a 10th of a second: 0.197"/5mm
750
1000
**Assuming feed isn’t limited by something else. Doesn’t include a deceleration as that can vary depending on what you are doing and the controller/settings.
Not saying to not try different settings. Just keep in mind that everything else attached to those steppers down to the tool are dealing with different force dynamics.
Even just from the machine standpoint think about disconnecting the steppers then grabbing your router/spindle and moving it by hand at different speeds. It’s more resistance and vibration the faster you go.
Hope that’s useful. Let me know if there’s something I can help with.
John, little too technical for this time in the morning but I understand what you are saying. Yesterday was the first time I actually got to see what happens when the machine looses zero. The bit slows way down as it travels along the x axis. The x axis motor is still turning but extremely slowly, then frees up as it traverses in the oposite direction. This suggests to me that either there is binding of the backlash nut or v wheels, or motor driver output is an issue. If it is binding then it is a cumulative issue because it is not immediately occurring in that the machine will run correctly for a while before it happens. I am fine with running with the slower speed and in fact have more concerns with the comms isses with the SLB and the jogging that is either gsender or SLB.
Yeah, I hear you. Not nearly enough coffee in me yet for proper thinking (keep that in mind for the below).
If I was going to try and tune this I would work my way first from a no load stand point. Reason being that the cutting forces are directly related to cubic material removed per flute, tool geometry, and material being cut. So unless you tune for every combination of materials and tool, there’s always the risk that you are going to lose steps changing the tool or material at the same feed.
With that in mind I’d setup a test like our sweetspot test. I just ripped this from our site. Keep in mind it’s only an example. You will need to possibly change header and units depending on your machine’s setup. I also can’t remember if grbl likes line numbers off the top of my head.
N000 G90
N001 G00
N002 G20
N003 G94
N004 Z1.000
N005 X0.000Y0.000
N006 M03
N007 G01
N008 F5
N009 Z-0.0313
N010 G91
N011 ( ** START ZIGZAG ** )
N012 F10
N013 X1.000Y0.125
N014 X-1.000Y0.125
N015 F15
N016 X1.000Y0.125
N017 X-1.000Y0.125
N018 F20
N019 X1.000Y0.125
N020 X-1.000Y0.125
N021 F25
N022 X1.000Y0.125
N023 X-1.000Y0.125
N024 F30
N025 X1.000Y0.125
N026 X-1.000Y0.125
N027 F35
N028 X1.000Y0.125
N029 X-1.000Y0.125
N030 F40
N031 X1.000Y0.125
N032 X-1.000Y0.125
N033 F45
N034 X1.000Y0.125
N035 X-1.000Y0.125
N036 F50
N037 X1.000Y0.125
N038 X-1.000Y0.125
N039 F55
N040 X1.000Y0.125
N041 X-1.000Y0.125
N042 F60
N043 X1.000Y0.125
N044 X-1.000Y0.125
N045 G90
N046 G00
N047 Z1.000
N048 X0.000Y0.000
N049 M05
N050 M02
You don’t need the Z moves here but what this does is run a zip-zag that’s 1" long and 0.25" high per zig-zag at increasing feeds. That should be the worst case scenario and you should be able to hear and see if the movement is too violent for your taste. Be sure to set your zero at some point you can remeasure but not too close to the edge of an axis (in case you lose steps). If you don’t return to zero then you lost steps. You would need to run this per acceleration setting as you are unlikely to get as high of a feed at higher acceleration. I would also run it at each corner and center to check for variation.
Keep in mind that the higher your $11 (Junction deviation) setting, the more feed you will maintain between direction changes but more violent they can be. $12 ( Arc tolerance) will also effect it in arcs as the more sub divided segments, the more slowdown based on acceleration and deviation.
This should give you a max for no load rapids but I’d still drop it by at least 10% for some margin. Again, there’s no good way to test it for actual cutting but to cut. And as soon as you move to higher hardness/density/friction material or larger tool diameter/lower rake/different helix tool the forces go up (like for like).
All the above aside the best settings for you are almost certainly not just the highest you can maintain for max feed, acceleration, and junction deviation. As we all have to deal with machines with physical limits it’s down to what you plan to do with it. e.g. lots of big straight moves (large segments/lines) will do better with lower acceleration, deviation, and higher max feeds. Small engravings or 3d carvings with lots of small direction changes will probably want the opposite. Both to within the limits of the machine. This is also ignoring tool limits like micro tooling, long reach, or larger diameter. Trying to use those will change a lot of what settings you actually want to run.
Hope that helps.
Thanks John, I had thought of doing something like you say but had not thought of doing the zig-zags, a better way to test I think. I really mainly do relief carvings with the odd sign thrown in now and again. I will give this a try sometime next week and report back.