Advanced Search

your log say's: wrong header: 116, 105, 114, 119, 0,...

this is the tx header, it looks like you get the data back you have send (UDP-loop test or so).

the right rx header is: 97, 116, 97, 100,

meister post=306847 userid=19472
do you run halcompile on the linuxcnc host ?
or at least you have to copy the rio.so (from halcompile) file to /usr/lib/linuxcnc/modules/ on the lcnc host

yes of course, or else it wouln'd have connected to the fpga. From the host computer: $ sudo halcompile --install rio.c 

Can I bother you more and ask to see how you configure your network interface on your lcnc linux? Below is what mine look like. 

do you run halcompile on the linuxcnc host ?
or at least you have to copy the rio.so (from halcompile) file to /usr/lib/linuxcnc/modules/ on the lcnc host

Yes I did.

I've seen you made changes to your repo and so I've deleted everything locally and started fresh from scratch. Git re-cloned everything, recompiled the gateware, flashed to fpga, halcompile --install rio.c, then copied over the Linuxcnc config files. E v e r y t h i n g brand new!!

That way I ensured everything was fresh as you left it after your mods and tests yesterday. Even went and get the prod computer to test fresh onto.

Saturday's Evils I tell ya lol Trust it's a good day to panic!

Funniest thing is it works(ish) onto the PC I dev onto. That PC can't be used with Lcnc because of the crappy ethernet adapter... one of those affected realtek nic. At least I'm not getting this bad header error onto this PC. So weird... totally lost

no panik !!!

do you have recompile your hal-component ?

Tests carried. Tested onto my production Lcnc-uspace computer which host a high performance r8169 ethernet adapter. Same I use for my gigabit based lathe (LiteXcnc). Yet I'm unable to have Rio working on this one

Strangely enough, I get zero pkt losses and a fast ping response from the wizzy part on that computer. As if it likes it very much! Therefore fair to assume I'm connected solid.

So upsetting to see Rio working for others, but not for myself. I smell badluck arrggh

HELP, calling the Godz!!

If anyone would still look for pinouts, please see attached 

Hi 
I've got some weird issue with my configuration. 
When I home I should be on X0Y0. But I'm always 0.003 - 0.005 from this point on both axes. 

Same when I go to X50, it is stopping at 49.995. DTG shows 0. 
I've noticed in hal config that j0pos-cmd and j0pos-fb are different. 
It happens on all of the axes 
​​​Can someone advise anything? 

Aciera wrote:

You EtherCAT guys are doing a great job at making things more accessible but it will be some time before I'll send a newcomer looking for an easy entry down that path.

+1

Not yet, there are things ti disable in BIOS, like hyperthreading, power saving, TPM, sleep states, turbo mode, speed step, etc.
And get the older Wheezy ISO as it uses the RTAI kernel and test with it, should do much better.
Also, repasting the CPU cooler will often make the PC usable.

My new project is a control panel. Always wanted to make one, but something kept getting in the way. Now I finally managed to realize it. Nothing special, the case itself is bent and welded from 2mm sheet metal, 12" touch screen monitor, industrial keyboard and all the buttons, computer also fits. All the buttons and MPG are connected via adruino. As usual, in many things had to figure out from scratch, the config turned out perhaps the most complex of those that I have ever done. But in the end, everything works as it should, thanks to this forum, I was able to find answers to all my questions here.

Moutomation wrote:

If it weren't for you it wouldn't have been solved so quickly. Updating Ethercat may not be under Linuxcnc's control, but I think that such important files, versions and packages that were running in the past should be preserved or backed up. I appreciate your support and you serve the purpose of the Forum's existence.

This isn't possible. Due to licencing requirements, Linuxcnc is unable to keep a copy of Ethercat because we are not a company. So linking to an external repository is the best we can do.

For the record, I had my part fixed in 24 hours but held back announcing the update because Ethercat didn't install. Once they fixed their bug, there was no further action required by me.

Well, the OP could consider Ethercat which is flexible and expandable but adds some complexity, reduces wiring and probably adds more cost than mesa cards

I'm not ignoring ethercat but since the OP wants a solution that doesn't involve a lot of research I opted to point him to the widest and most trodden route. Which is also the route with the most support along the way.
You EtherCAT guys are doing a great job at making things more accessible but it will be some time before I'll send a newcomer looking for an easy entry down that path.

@cornholio post=306822 userid=35653
Can we append a Kb of great zero-dummy nothingness to the rx and tx buffers to cancel this throttling misery thing? How silly.
This needs testing on a 20~30bytes load to confirm. I'll see if I can rig something up before cracking on something gigabit-like. 

@meister post=306785 userid=19472
That's awesome news, knew you'd sort it eventually (ha!). Let me test this out after dinner when my bunch here is quiet (saturday!).

I installed linuxcnc and checked the latency.

It scored 12 million nano second 

back to dumpster