Smoothieboard - 2 steppers/axis homing

As suggested, I've started a new thread specific to this problem.

Firstly, I have a chinese compatible MKS sBase v1.3 and not the genuine article. In my defense, I wasn't aware of smoothieboard's nature of open source until after I bought this board that was recommended by a local supplier. They appear to my amateur eye functionally the same. Onwards....

I've decided to use 2 steppers for the x axis on my laser etcher/cutter build.

It's been suggested that a simple way of homing the axis is to push it to the hard limit, so it's square, prior to homing. Sounds like a good workaround to me.

What I'd like to know, does anyone know how to home the 2 steppers independently using the smoothie? I can't experiment with methods because my steppers are in the mail. Will try a few things when I can.

My thoughts - The drivers on board are slaved using jumpers from the input signals for one of the drivers (actual X) to the input signals of the slaved driver (E1). So that any stepper connected to E1 will be driven in exactly the same way as X. Therefore no independent homing can occur.

Suggested method 1: Post process the file and convert arcs to short line segments. Add E1 commands to all the G1s (duplicate the x axis dimension but call it E1), so that E1 is tied to X axis by g-code alone. No need to jumper the axis. Potential for much larger file size and arcs not as smooth.

Suggested method 2: Use a switching method, associated with an M-code, that would break the link that's jumpered on board. Then home, then re-activate link. Some sort of relay, relay board, etc.... Allows arcs to be interpreted correctly, but the config file says you need to remove certain references to the slaved driver/stepper and this may preclude it being homed independently.

Anyone have any ideas or thoughts on the matter?

I think that with independent homing you will double your homing error, technically you would be still out of square. I really doubt that independent homing for two steppers would give your more accuracy than simple manual move of gantry into hard stops.

The idea of individual homing, is that each stepper won't stop until it hits it's corresponding limit switch. Providing the limit switches have been placed correctly, each stepper will find it's zero point independently, even if they started off out of square. Hope that makes sense.

Each limit switch will have positional error(repeatability). For example inductive proximity switches has error about 5% of sense distance, so one side of gantry could be triggered on top limit and other on bottom limit, i.e. 0.2mm difference detween sides for quite common swithes with 4mm sense distance.

Quote:

---

Originally Posted by kfmut

Each limit switch will have positional error(repeatability). For example inductive proximity switches has error about 5% of sense distance, so one side of gantry could be triggered on top limit and other on bottom limit, i.e. 0.2mm difference detween sides for quite common swithes with 4mm sense distance.

---

I have tested very cheap inductive switches and find their repeatability to be much better than you indicate. On my mill over hundreds of homing cycle tests, the repeatability differed no more tha a few ten-thousandths of an inch. (0.0002")

My switches are mechanical in nature. I'll have to check their accuracy with multiple homing sequences and measure any errors to see if that's an issue. But being a DIY job I can change them and their mounting hardware if needed. I don't expect it to be a problem but certainly can't hurt to test them to be sure.

109jb, 5% of Sn error is straight from datasheet, it's includes variations of supply voltage and ambient temperature. It's considerably better for branded sensors.

Looking at the datasheet, all I see is Hysteresis, which is not the same as an error. Repeatability of activation will be nowhere near the hysteresis value. The sensor triggering distance is affected by the size and electrical characteristics of the target, but unless the target is dynamically changing properties the triggering distance will be very near absolutely the same.

Hysteresis can be compared to the operation of a household thermostat. A thermostat that comes on at say 70 degrees F will remain on until say 72 degrees F. That 2 degree spread is the hysteresis and prevents the furnace from going on-off-on-off constantly. The same is true with the switch. There is a hysteresis designed in to prevent the switch from :"bouncing" causing repeated on-off-on activations.

Regardless, my TESTING has shown that the switch activation of a cheap inductive switch is repeatable to 0.0002" over hundreds of homing cycles. The number on a sheet of paper means nothing when there is actual on-machine testing performed.

109jb, of course those numbers are for worst case scenario, in one of datasheets i've seen that this numbers were obtained after 8 hours of testing with temperature swing from 10 to 30 degreees celsius and supply voltage varition of +/-5% from nominal.

https://www.is-com.ru/files/im12.pdf , page 4, repeatability is 2% with constant voltage and ambient temperature.
http://literature.rockwellautomation...d001_-en-p.pdf , all kinds of inductive sensors, repeatability goes up to 10%,

Regardless of what any datasheet says, the demonstrated repeatability on an actual machine is far better than the 0.2mm that you cited in your post. You stated that the 0.2mm error is "quite common", but I would argue that 0.2mm is the absolute worst case and therefore would be quite uncommon..If you want to trust the datasheet then that is up to you, but real world is what I put more stock in. Also, the datasheets don't say the repeatability =2%, the say the repeatability is <2% of the target distance.

Here is a video I put together a while ago with a hodgepodge setup that showed 0.0002" repeatability and this was only a few of the cycles where I performed testing. I actually performed hundreds of homing cycle tests and never had more than 0.0002" difference.



https://www.youtube.com/watch?v=IGO47l62T2o

109jb, "quite common" were only referenced to sense distance of 4mm, which is usually default for no-name barrel-style proximity switch.

I've seen this video previously, it's just irrelevant cause you are stating that Grizzly G0704(made in China?) mill has same repeatability as HAAS VF and ten times better than something like Tormach PCNC...

First, The video never "claimed" that the machine was as accurate as any machine whether it be a Haas, a Tormach or whatever.

Second, The video only claimed to show the homing repeatability of the switch and as such it only requires a small amount of motion. It doesn't even really require any overall machine accuracy to show repeatability. Something that you choose to ignore because it apparently doesn't fit your opinion.

Third, the machine in the video has double nut ballscrews with measured backlash of 0.0008". It is not an out of the box G0704.

Fourth, Where a machine is made has IS TOTALLY IRRELEVANT to how repeatable it is or could be. BTW, direct from the tormach website:
1. Where are Tormach products made?
Tormach machine tools are designed and tested in the USA, and manufactured to our specifications in China.

Fifth, The video clearly shows a dial indicator set to read the final position after the homing cycle runs. It clearly shows only about 0.0002" difference between any of the homing cycles performed. This absolutely means that this Chinese made machine converted to CNC came to its final position after each homing cycle within 0.0002". You can call it irrelevant, but if it didn't repeat that accurately then you will need to explain why the dial indicator read the table position the way it did.

You can sling BS all you want about datasheets and how clearly relevant testing is irrelevant all you want. How about you show us testing on your Haas that shows how non-repeatable a proximity switch is for homing purposes.

Too many words, too little meaning. So without backlash compensation and with measured ballscrew backlash of 0.0008'' you have accuratly measured value of 0.0002'', I bet that you also have less than 1000 steps per mm in your GRBL-config...I think I'll skip deep dive into metrology here.

Backlash is irrelevant because the homing cycle completes by moving in the same direction every time and the backlash is taken out. So yes I can measure that the table came to a stop each time moving in the same direction within 0.0002". Also, how do you know that I don't have backlash compensation turned on? Is it by chance because you are assuming things you know nothing about? The machine was used for testing Grbl, but it does not run on Grbl normally, and in the video presented it is running on LinuxCNC. The steps per inch at the time of the video were set to 32000 steps/in (1259 step/mm), using a 0.2" lead ballscrew and 1/32 microstepping. Even if you want to argue that 1/32 microstepping isn't accurate enough positioning, each 1//2 step is accurate and would equate to 0.0005". Even if I conced that my test can't show any better than that 0.0005", it is still more than an order of magnitude better than what you claim the switches are capable of.

Quote:

---

Originally Posted by kfmut

Too many words, too little meaning. So without backlash compensation and with measured ballscrew backlash of 0.0008'' you have accuratly measured value of 0.0002'', I bet that you also have less than 1000 steps per mm in your GRBL-config...I think I'll skip deep dive into metrology here.

---

Kfmut,
Your posts are not enhancing or contributing in any positive fashion to this thread, do you have a specific reason for attempting to undermine a member that has a thousand times as many constructive posts as your manipulative rhetoric?
Live from Burbank, home of the endless Cup-Of-Joe
JoeyB

Lets keep it civil folks.

Mark

Quote:

---

Originally Posted by Blackened

As suggested, I've started a new thread specific to this problem.

Firstly, I have a chinese compatible MKS sBase v1.3 and not the genuine article. In my defense, I wasn't aware of smoothieboard's nature of open source until after I bought this board that was recommended by a local supplier. They appear to my amateur eye functionally the same. Onwards....

I've decided to use 2 steppers for the x axis on my laser etcher/cutter build.

It's been suggested that a simple way of homing the axis is to push it to the hard limit, so it's square, prior to homing. Sounds like a good workaround to me.

What I'd like to know, does anyone know how to home the 2 steppers independently using the smoothie? I can't experiment with methods because my steppers are in the mail. Will try a few things when I can.

My thoughts - The drivers on board are slaved using jumpers from the input signals for one of the drivers (actual X) to the input signals of the slaved driver (E1). So that any stepper connected to E1 will be driven in exactly the same way as X. Therefore no independent homing can occur.

Suggested method 1: Post process the file and convert arcs to short line segments. Add E1 commands to all the G1s (duplicate the x axis dimension but call it E1), so that E1 is tied to X axis by g-code alone. No need to jumper the axis. Potential for much larger file size and arcs not as smooth.

Suggested method 2: Use a switching method, associated with an M-code, that would break the link that's jumpered on board. Then home, then re-activate link. Some sort of relay, relay board, etc.... Allows arcs to be interpreted correctly, but the config file says you need to remove certain references to the slaved driver/stepper and this may preclude it being homed independently.

Anyone have any ideas or thoughts on the matter?

---

Blackened: Did you ever figure out how to do this? I am trying to configure my own CNC with dual endstops and steppers on the y-axis using a BigtreeTech SKR V1.3 board.