Advanced Search

I have been trying to install ethercat using the step-by-step instructions for a month and nothing is working. I have erased my partition and reloaded at least 20 times, but there is always a roadblock: broken links or unmet dependencies or it can't find the repositories.

My downloaded iso file installs debian 10 and linuxcnc 2.8.4

My current error is:

The following packages have unmet dependencies:
linuxcnc-ethercat : Depends: libc6 (>= 2.34) but 2.28-10+deb10u4 is to be installed

I am either getting closer or just circling the drain, any help would be appreciated.

when I try to update libc6 it tells me the most recent version is 2.28-10+deb10u4 which is clearly older than what ethercat installation wants.  How do I get to success?

blazini36 wrote:

I haven't played with the qss builder but I did update to 1.1.1 this morning so it sounds like you are saying I should be seeing bold text with the dark-touch theme?
 

Yes you should check out the themes thread I uploaded fresh screen shots.

JT

I haven't played with the qss builder but I did update to 1.1.1 this morning so it sounds like you are saying I should be seeing bold text with the dark-touch theme?

I've recently discovered that you have to set the font-weight to 700 in the style sheet to get bold which is what I did this morning in the touch themes. The Flex QSS builder creates the correct stylesheet code for bold.

JT

@royka I can confirm that you fix works on Raspberry Pi 5. See this thread:

forum.linuxcnc.org/media/kunena/attachme...ernel-realtime.patch

So far so good. I now get "Note: Using POSIX realtime" after applying the patch from that post I linked. Thanks a lot rodw.

I've updated my gist for anyone that lands here in the future.

I decided that I was setting a bad example by running my own lathe on LinuxCNC 2.8.0~pre on Debian Wheezy. So I installed the latest LinuxCNC ISO on the DN2800MT that I built it with in 2015.
This consumed the entire 8GB SATA DOM SSD and was glacially slow. LinuxCNC and latency was fine, but UI response horrible, and I was having to press the cycle-start and (more worryingly) cycle stop buttons for a couple of seconds to get a response. Possibly a graphics issue, as Linux has never properly supported that chipset. It's also debatable whether it's a fully 64-but motherboard. (The specs and manual simply don't say...)

I could try playing about with graphics drivers and such. In fact I did, a bit, but then I got bored and bought a new motherboard. 

Asrock N100DC  www.mini-itx.com/~N100DC-ITX
The shopping list was:
PCIe slot (for the existing Mesa 6i25)
DC Power input ( to run off the controller 12V)
M2 SSD slot (so that there is no need to sort out SSD brackets, and to keep everything compact and low-power)
VGA connector (but it also has HDMI) so that my existing touchscreen (and touchscreen wiring, through the conduits and machine base) will work

Other possible plus points:
It has a parallel port header on the motherboard, in case you want to go old-school
It has a serial port, and a motherboard UART port too. 

Latency is "fine". It runs at 11k most of the time, but can go up to 25k when opening web browsers, GLX gears, that sort of thing. Perfectly OK for a Mesa card. This is with an un-tweaked LinuxCNC 2.9.3 ISO install. 

It's very snappy (possibly the SSD?) Touchy opens in under 2 seconds from pressing the button, and its 30 seconds from power-on to a responsive LinuxCNC GUI once autologin and autostart are set up. (and several seconds of that startup are spent waiting for grub to timeout and load the default). 

I added an 8GB DIMM (the smallest DDR4 I could find) and a 240GB M2 SSD (again, the smallest and cheapest I could find)
Motherboard £106 + VAT
Memory      £12.99
SSD           £16.29
Total   =     £156

More than a Pi, but it will slot right in to the existing space. 

Can't edit that post or the quotes turn to mush but I meant that that is a bold font, it shows in designer but not when run. Noto Sans Bold appears the same as Noto Sans Regular

JT wrote:

 

What bold text thing?

JT
 

There's a note in your docs:

NoteThere is an documented issue with Qt5 Designer and bold fonts not appearing properly.

Not sure if that's referring to the same thing but normal fonts set to bold appear fine in Designer. When run the bold text is missing. I've only tried the dark-tough theme so far but this:
 

Is turning into this for me
 
 

fully_defined wrote:

Okay. I am looking for known good values for Clearpath SDSK servos on an Avid router, specifically a Benchtop Pro. There is no way that I am the first person to set up this specific configuration.

[/quote]

Its rare to find people with identical setups. Start at about 750-1500 and see how you go once you are moving. The mass you have to move will affect the maximum value. Assume for now 3000 will be absolute maximum. Remember its a trapezoidal accelleration profile which means the deceleration is also affected by this setting. I have had to reduce accelleration to prevent the motor decelleration being overrun by the inertia of a moving mass which caused following errors.

Its better to get reliable motion instead of shooting for the stars immediately and failing.

Its likely you have only 1 encoder enabled in your hal file.

To check the pinout (and number of encoders):

mesaflash --device 7i96s --addr 10.10.10.10 --readhmid

The 200XT may be a HF start device. hyperthem machines that size are. This might mean you need to connect the thcad upstream of the HF generator. If not, It's likely to have damaged the THCAD or the 7l96 encoder. Tommy is the expert in this area.

But as PCW says, seeing 0 volt and 45 volts indicates it is working.

Rather than look at the reported voltage, check the encoder velocity pin in halshow. When not cutting, it should hover at or near the 0 volt frequency. Also it should be bouncing around. If it isn't and just a fixed value either there is an error on your encoder config in HAL or there is an issue on the high voltage side.

Using the plasma bit file, you have 3 encoders so damaging one encoder might have an easy  work around.

 

PCW wrote:

Looks like halcmd was run without running LinuxCNC

mesaflash --device 7I96s --addr 10.10.10.10 --write 7i96s_p1encd.bin

should work

mesaflash --device 7I96s --addr 10.10.10.10 --write 7i96s_p1encd.bin --reload

is probably better

 

Seemed to upgrade, but I don't see additional encoders.
Before I did the upgrade (Linux running) I captured all the hm2 stuff see: hm2names.txt
Then LinuxCNC closed,  I did the upgrade and followed up with reload, that seemed to work see upgradetxt.txt
Then captured hm2 names again see:  up_hm2names.txt
No apparent changes, and when I look in HalShow I only see encoder0, I was expecting to see a few more.

Not sure what I did wrong, unless I was supposed to do the write and reload in one pass instead of two?
     

OK, went through those, and i have to agree with PCW on number 2 here
forum.linuxcnc.org/10-advanced-configura...fb-mesa-7i76e#112557
It is pretty much useless as it does a lot of things but the thing it should do, it does not, namely amplifier fault!
I would not use it.

johnbl wrote:

I don't have the time nor money to sue Tormach to get it.

I am sure they did provide it several years back, but only on request. Did you try shoveling and e-mail in their direction?
-
Personally, i would think they should provide it publicly... but still, they were selling the full PathPilot on a USB stick for 35$, so i can not be angry at them.