Advanced Search

Thank everyone for the given responses. I tried to find a hint, but was not able to find the mentioned threds, I will read and try to make some sense out of it.

I dId not have the hal file at hand, But it is the one that Pncconf generated after clicking through the basic menues quickly. Hal file I need to copy from the machine to the Internet, when I am in "the cave".

p.s. solved see next posts


 

spumco wrote:

Is your servo drive in position or velocity (or torque) mode internally?  I suspect this makes a big difference... but I also don't know of any servo drives that accept S&D command in velocity or torque/current control.
 

Pretty much what you said, it runs S/D in position mode. When I set it up I didn't use any sort of PID on the linuxCNCs ide, for milling it works perfectly fine. Rigid tapping works pretty well too except it switches directions very rapidly which sometimes gets me in trouble with smaller taps. Been planning on tuning it up one day but since I don't have any real issues it'll probably be later than sooner.

PCW wrote:

A simple fix would be to simply tie the spindle  stepgen enable true.
Then the stepgen accel settings can be used  to limit  the
ramp-up/ramp-down rates.
 

I think that's what Andy and NoJo essentially came up with in the thread I linked for NoJo's C-axis lathe.  Basically tied the stepgen enable to motion.machine-on.

But the issue going to keep cropping up as long as PnCconf outputs configs for newcomers with the 'gotcha' built-in.

A simple fix would be to simply tie the spindle  stepgen enable true.
Then the stepgen accel settings can be used  to limit  the
ramp-up/ramp-down rates.

blazini36 wrote:

This is something I never actually noticed. I've been using a 5krpm servo with S/D on my spindle for years. It probably does this but I never actually noticed because it's such a low inertia spindle, even with the tool. I'm sure this would be a much bigger problem on a lathe swinging a part.
 

Is your servo drive in position or velocity (or torque) mode internally?  I suspect this makes a big difference... but I also don't know of any servo drives that accept S&D command in velocity or torque/current control.
 

This is something I never actually noticed. I've been using a 5krpm servo with S/D on my spindle for years. It probably does this but I never actually noticed because it's such a low inertia spindle, even with the tool. I'm sure this would be a much bigger problem on a lathe swinging a part.

This seems to be an issue that's cropped up over the years for stepgen-based spindle control.  I think it's because the stepgen simply turns off at an M5, and if the motor drive is in position mode (typical stepper, and sometimes servo) the drive sees that as a command to stop instantly.  If the drive is in velocity mode, then stopping the stepgen will (likely) result in the drive's internal decel settings taking over.

I've been doing some reasearch on this, and I noticed that both PnCconf and Mesact both create configs where the spindle stepgen control is treated like an axis stepgen.  I brought this up a few months ago, but didn't receive a response.

forum.linuxcnc.org/39-pncconf/49366-pncc...ep-dir-spindle-issue

JT responded to me on github, but I've not checked to see if mesact has been changed.

I've come across two solutions... one mentioned in the post above, and the other was a HAL time-off relay combined with a limit.comp that hijacked the M5 command and ramped down the spindle.

I have a stepgen spindle about a day away from commisioning and I'll have to sort this out as well.  I'll probably go with the method worked out in the link above.

Should probably post your hal file.

M3 S0 may be treated the same as M5. It may be setting a motion hal pin like spindle.M.on = false. Maybe that's connected to the stepgen enable signal or something. hard to guess without seeing it.

Hello,

I am playing around with some motors and a lathe configuration. (no physical load on the motor shafts)

I used Pncconf to set up a lathe example with x-z axis and a step dir pulse spindle. 

I put values in for RAMP_RATE, MAX_ACCELERATION and STEPGEN_MAXACCEL

• M3 S3000 speeds the motor slowly up
• Then M4 S3000 slows it down and speeds slowly up in reverse direction
• M3 S3000 and then M3 S0 stops the (stepper motor) immediately, without slowing down
• M3 S3000 and then M3 S1 slows it down
• M5 also stops immediately without slowing down first
  
  Is that normal? Why is M3 S0 behaving totally different from M3 S1?
  Usually the spindle is ramping down over a certain time when M5 is in the program. But the machine is either "freewheeling" or a VFD does its "slowing down procedure" without the Linux computer.

I will be happy about some pointer in the right direction.
Greetings
 

Hi All

I got my XHC-WH04B-4 working recently. I can control all axis and homing, got the MDI macros to set zero and spindle start.
I have a few things that I really need to improve and functions to get working. Hopefully this topic will provide useful development for others going forward.

I see 2 approaches to resolve the issues -
1. modifying HAL configurations
2. modifying the source code behind xhc-whb04b-6

Issues so far:
FEED OVERRIDE - MPG
On start up the pendant is not in Continuous or Step Mode - I note that feed override is changed when using the handwheel
Once homing is complete or Continuous or Step is selected then feed override cannot be affected by the hand wheel
I presume because unlike the 6 axis pendant this 4 axis  version does not have a button to select MPG mode, so we cannot return to MPG mode - which i have read is combined with MPG mode for the the feed and spindle override.

Regardless of the above, the operation I really want is simple - Once homed - Select LEAD position and have the hand wheel control feed override - Then I can safely work thorough machine runs using this as a way to quickly halt the machine, or slow feed through areas of programs.  

LEAD Position
The rotary position LEAD puts out a different code to the 6 AXIS VERSION - so I cannot select LEAD mode as far as the driver is concerned. This i think requires a change to the source code. I found the offending line in pendant .cc

Line 383 : Needs changing to 0x09 for the 4 axis version (is 0x1C for 6 axis version)

Continuous vs step mode jog.speed :
Continuous operation gives a jog speed far too slow. Step mode gives a step far to large at the higher settings 60% and 100%.
I used halmeter to verify the jog speed scale in the 2 modes and found :
selector Position             Jog.scale cont  Mode          Jog.scale - Step Mode
0.001                                         0.0008                            0.001
0.01                                              0.002                             0.01
0.1                                                 0.004                             0.1
1.0                                                 0.012                           1.0
60%                                               0.024                           5.0
100%                                              0.04                            10.0
Noting also that:    continuous mode - axis.n.jog-vel-mode = 0            In Step Mode         - axis.n.jog-vel-mode = 1

In Step its Ok from position 0.001 to 1.0, but step of 5.0 and 10.0 is crazy and will result in smashing tool into something at somepoint - Note removing the option for big step - B from  loadusr -W xhc-whb04b-6 -HsfB  Does not help the large steps remain active.

As the option B is ineffective , I'd be quite happy if the 60% and 100% settings just gave no output or capped it at 1.0 - so need a way to invoke this.

I see a few possibilities here between a hal solution or a source code change.

Regards my HAL file I have implemented a lowpass filter and also to scale the jog.counts - this improves the continuous mode speeds but then, of course,  undesirably scales the step mode distance. I figure I then need develop the HAL code to change the scale applied depending on the mode (cont or step) - first though is to use axis.n.jog-vel-mode with a mux to apply a different scale - but I haven't experimented yet.

But I also wonder why the driver reduces the jog speed scale so much in continuous mode? Perhaps a change to source code giving higher jog.scale in continuous mode is more elegant than complicating the HAL file. 

PENDANT LOCKS UP

Not sure how but a couple of times i have the pendant in a state where nothing responds - the usb is fine and the pendant shows activity but no response to hand wheel or buttons. possibly after applying MDI commands in axis then returning to manual mode. - needs further investigation.

Machine: MESA 7i76  + 5i95  - stepper motor axis mill and lathe in mm mode

My gut feel is some source code modification with an extra option for the 4 axis pendant would be great - then the lead mode issue could be fixed. Possibly a change to the cont mode jog.scale values - i'm sure they could be much greater as the machine simply stops when the pulse stop so its much safer scenario than large steps in step mode.

I have some degree of confidence in investigating and modifying the source code - but I have no idea how to go about compiling and implementing changes to xhc-whb04b-6. However I have a spare machine running a sim and i'm wilting to try - any pointers on where to get started are greatly appreciated. (I have some experience with developing software in windows, but I'm a linux newbie)

Otherwise if you think HAL offers a solution please feel free to suggest things. I may have missed a simple method to resolve some of this and happy to consider other ideas.

If your still reading at this point  - well done, thanks for taking the time.

Bally

Hello everyone, I own the SPINDELMOTOR ISA 900 W 477009 33245 ( www.isel.com/en/spindle-motor-isa-900.html ) along with its FREQUENCY CONVERTERS ( www.isel.com/en/frequency-converter.html ). I'm interested in controlling it using LinuxCNC. I possess two boards of a 5 Axis CNC Breakout Board with Optical Coupler for Stepper Motor Driver MACH3 ( www.aliexpress.com/item/32796933346.html ). My query is whether it is possible to control the spindle motor speed to achieve a precise number of revolutions and, if so, how can I achieve this? Thank you.

thanks but not help anythink, germany not good translate normal human language all go wrong, and peoples who use first time all life linux and linuxcnc and not has newer before read or write any code not understand code anythink, and i need only help, were i add example Ztouchplate all codes, i not has newer before use code,linux or liuxcnc. need know how make button, how button answer to bush and how do what g-code want doing. were all this code and parts need add, wery simple answer if know. button code need add to folder xxx. parallerport pin 15 need add to xxx file. xml code need add to xxx file or folder. ngc file need add this folder xxxxx. but any people not has answer to me were need add all file parts.
touch plate need only build button first, then hall meter test pin 15 working, my machine pin 15 working ok, whit hallmeter.
then build how answer to button move spindle to right place and do G-code 38.2 procedure. were this need add folderxxxx. and what other need add. wery simple answer if have linuxcnc user .
but i ask thats because i not has newer before make code, write code, add code to linuxcnc, used linuxcnc before newer and need know were i add all file parts.
i has try search thats last 2 month in google but not found any help, all have only other mach3 sites and videos, or expensive mesa card systems, i not have mesa card i not know what this doing and not newer can buy expensive mesa card, i use only paraller controller board, and first time all my life linux and linuxcnc software all have new and not newer has before all my life write any code, im not have computer guru pro im old 65 year men who build cnc machine my young son garage,help him build own garage woodworking shop. my machine have normal sterppermotor cnc size 1200mm x 1200 x 300. only woodworking hobby in garage. simple cheap chinese paraller port board 5 axis board and use 4 stepper motors nema 23 470oz. 4A, spindle have better than router have air cooled 800W ER16 0-30 000rpm. and use first time all my life linux. and want use linuxcnc because have FREE not need expensive licence. need only know how add touchplate working can set up tool height after change bit. thats is normal set up ewery cnc after change bit. basic job but linux have lot different than windows use, not understand linux and code anythink.hardware have easy simple, but this code system not have simple, need help were all code need add. what folders and what path need add button code i has copy to internet, G-code example i has copy, but were all need add.

carson wrote:

So would I have any reason to expect better performance for threading operations with the Mesa FPGA board if the spindle synchronization servo is still being done by LinuxCNC? 

Yes. LinuxCNC is not the limiting factor, parallel port is.

carson wrote:

  With my current setup (using the parallel port) threading messes up once every few passes (seems to get slightly out of phase with previous passes

Most probably parallel port exceeded the ability to count encoder pulses.
You can always try a slower feed rate till you find a reliable value.

carson wrote:

Also, for closing the loop with linear encoders and steppers, I've read some people on this forum have had good success, but is it even worth trying for a setup (like mine) that has regular acme threaded lead screws with a fair bit of backlash?  That seems tricky for the feedback controller to handle.

Linear scales would help a lot, the machine will get to where it is supposed to despite backlash.
With simple steppers you can still use backlash compensation, same as with encoders attached to the motors, so not much of improvement, but with encoders it would alert if it skipped steps, and than can be used by LinuxCNC to stop the machine so you can check what went wrong.

So would I have any reason to expect better performance for threading operations with the Mesa FPGA board if the spindle synchronization servo is still being done by LinuxCNC?  Maybe it depends on whether it's the base thread or the servo thread that is limiting performance with my current setup.  With my current setup (using the parallel port) threading messes up once every few passes (seems to get slightly out of phase with previous passes: forum.linuxcnc.org/26-turning/45294-thre...zation-issues#236574).  I never was able to fully diagnose what's going wrong.  I guess it could also be that the software encoder counting is too close to the limits of its performance.  But I'm really hoping things will work better with the Mesa board.

Also, for closing the loop with linear encoders and steppers, I've read some people on this forum have had good success, but is it even worth trying for a setup (like mine) that has regular acme threaded lead screws with a fair bit of backlash?  That seems tricky for the feedback controller to handle.

Yes, a 7I96S basically incorporates a breakout board so is intended
to connect directly to motor drives, limit switches etc. A 7I92T needs
an external breakout so a 7I92T could work with your existing parallel
port breakout.

All higher level motion control is done in LinuxCNC including all
spindle synchronized motion (threading, rigid tapping etc)
Any feedback/PID loops are also done in LinuxCNC.

The nominal 1 KHz servo thread rate is usually sufficient for this
kind of motion control. This is because the controlled device is usually
under velocity control and has a local velocity loop.

There are some threads here ( on the LinuxCNC forum ) that describe
using linear encoders with step/dir drives to increase the accuracy
and which were successful. Unfortunately these were a couple of years ago
and  I could not find them with a quick search.

EDIT: Here is a thread about using linear encoders with step/dir drives:

forum.linuxcnc.org/49-basic-configuratio...le-closed-loop-setup