Advanced Search

Hallo PCW, 

vielen Dank für deine Hilfe

der Befehl hat folgenden Output 

cnc@cnc:~$ sudo chrt 99 ping -i .001 -c 6000 -q 10.10.10.10
[sudo] Passwort für cnc: 
PING 10.10.10.10 (10.10.10.10) 56(84) bytes of data.

--- 10.10.10.10 ping statistics ---
6000 packets transmitted, 6000 received, 0% packet loss, time 5999ms
rtt min/avg/max/mdev = 0.066/0.078/0.143/0.007 ms

Bezogen auf den Takt von 1ms sind die Ping Zeiten ja eher gering. 

Was sonst kann den Watchdog Fehler auslösen?
Aus dem halcmd show param *.tmax sehe ich die Werte für read.tmax und write.tmax sind ein vielfaches höher als alle anderen Werte zusammen. Deshalb die Vermutung die Ursache liegt in der Verbindung zur Mesa Karte.

Grüße 
Erwin

 

rodw live view. Hopefully Alfred turns out to be a whole lot of not much.
Live View

what i found out in the meanwhile that at the time where the cyclic of do_homing() seem to be stopped, one of the synchronized joints is still in status "PK_PEAxisState_axHOMING_BACKING_OFF". but it allway seems to stop when the status changed to PK_PEAxisState_axHOME.

in that cas it's legit that one of them is still not homed, but for me is still unclear why the cyclic calls of do_homing() are being stopped.
the other miracle is why the flags of both joints are set to:
.homed = 1
.homing =0

edit 7 new findings:
i now tried to workaround the problemb changing the home Sequences so that the Y will get Home sequence 3. and X&X2 get home_sequence -2: that endet up that now the joint for Y doesnt get homed (i mean halui.joint.2.is-homed still shows 0)
   
that seems that it is not related to the synchronized homing sequence of joint.0 and joint.1, but more likely to the last sequence that is being homed.
and also logging showed that the cyclic calls of do_homing() (or all calls to home_comp) are being stopped right at the moment when the PulseEngine Status changed to 10 (= homed). 

as i'm quite sure that Linuxcnc motion or Halui know nothing about PE_AxisState, i think it is not the status change itself but something related to it, but i didnt see the debug message message. that do_homing() returned 1. so it is most likely something else

Yeh, I saw you had been there. Its still very normal here and Alfred has decided to be a day late.
There is some pretty rough weather to the south of us even though we are closer to the crossing point

M4MazakUser wrote:

Have you checked squareness of the part, using a micrometer measure the height in the 4 corners. It dosent look like a machine problem more like a clamping/squareness or misalignment issue. Unless you are loosing steps....

Indeed, today I checked the perpendicularity of the clamp and the truth is that it was not right, that is why there is this misalignment in the piece. I have aligned it and now it works perfectly.

Thanks! 

ugh. haha

if i know it was gonna be this disorganised i would have not bothered. I though there was a pretty clear instruction path.

I've been developing a new GUI that was originally based off of QtAxis as a starting point.  I'd like to contribute it back as an example UI in official LinuxCNC QtVcp, but I wanted to ask first about whether that was generally encouraged or not.  My GUI may not be 100% ready for prime time yet, especially in terms of being completely generalized for any type of machine config/axis count/etc.  If new GUIs are indeed welcome, where is the bar set in terms of being fully tested and wrung out before inclusion?

 

hey rodw is it ok to post the link of your youtube live view?

Yes, a sserial card can add real time 24V I/O I would suggest a 7I84U which
adds 16x 5 to 28V 500 mA sinking or sourcing outputs and 32x 5-28V inputs.

The sserial devices are RS-422 so do not allow more than one device
per interface. This so the remote interface is real time.

The 7I95T also supports Modbus devices (mainly for VFD use).
Modbus I/O devices are common and inexpensive but do suffer
from a couple issues:

1. They are not real time
2. The inexpensive I/O devices I have tested do not have watchdogs
so leave outputs stuck on after LinuxCNC exits or there is a fault.
 

Good find there. That should fix merging, but not when there is an incomplete circle and there comes a line.

More on the philosophy around this - I don't know enough of the background to this - but when in center punch mode, would it be ok to center punch the center of any circle segment? Doesn't have to be full circle, like the center of a fillet. I mean, you go into center punch mode deliberately and mark circles. The intention is to make a center punch for all marked circles/segments. So only that should be done, no straight lines or half circles or anything. It is only making spots while in center punch mode. If so, one should rearrange the code so separate out this action, mark the center of the circle segments and ignore everything else.

I wonder if this with Smoothing off should be in the description, the PDF where center punch is described.

Hello Jpevner
What AMC drives do you have? 
First, you may need to re-tune your AMC drives. Software may still be available at AMC. I would set the drives to torque mode if possible.
After this is done then adjust the PID etc in Linuxcnc.
I converted a Denford Novaturn. My set up is Pathpilot, IBOB2, MESA 7i77, 6i25, Yaskawa SGDA 01AS servos in torque mode.  I first tuned the servos for my machine's inertia. After this was done then the PID settings.
Some of the work is located here: 
DENFORD:  NovaTurn Servo retrofit AMD Gizmosphere iBOB2 PCIe x1

Hope this helps,
Iron-Man

So that is connected with one of those network cables on the 7i84 and then clamped to the screw terminals of the 9i95?
Is it possible to connect more than one of the cards to the same serial bus?

I must admit that I mixed up smart serial with modbus, because the RS485 tricked me into believing it is similar. But I do not have a clue about all this (who can guess?). I wonder if the 9i95t still could be used with the modbus RTU, similar to the video and omiting the USB adapter. I would probably not try to research it and just spend some extra money.



Seems there are many option to add I/Os to that LinuxCNC

 

hahaha Kam

It's intriguing how people will try to point the finger at the hardware or software rather than themselves......Yes I've been guilty of that myself, but I'm less likely to do that now, at least I hope.

I did some additional testing yesterday and found that I could never get the last hole to be a full circle unless I turned smoothing off. MergeCircles or changing tolerances did not seem to influence the end result. The sketched geometry was identical for all 4 holes. I did find a thread over on the Autodesk forum that seemed to be related to the development of this post processor.
forums.autodesk.com/t5/fusion-manufactur...le-arcs/td-p/9326368

I may reach out there if this keeps happening on other parts, but for now, I think I'll just use as is and test with other parts. If anyone else would like to use the post processor, I could remove the debug statements I added and repost it here.