Advanced Search

Those latency values should work, perfectly normal for linuxcnc.

What I see is in your syslogThe slaves were not able to synchronize their time in 5 sec.
I have had that problem myself and didn't solve it, it just went away one day.
The ethercat master sends a synchronize request every 1 msec, so 5000 requests are sent.
What I understand is that this initial synchronization is a slow process and sometimes 5000 requests is not enough.
What I read is that TwinCAT can send tens of thousands of synchronization requests when needed.

As an experiment you can increase this value to 50000, 10 times more, why hold back when experimenting?
It is this value in ethercat/master/fsm_slave_config.cChange toCompile the ethercat master, install it and make sure the new version runs.
And try again.
Interesting to hear if that works.
 

yes, i used this 'mesa-smartserial-device-template'
to build my relay card:
github.com/multigcs/SmartSerial-ESP32RelayX8Board

but you need an esp32 or something faster than an uno/nano, because the higher serial baud-rates

Great thanks! can you please point out to the correct github? I found this but I don't know if it is what you intended:

github.com/fdarling/mesa-smartserial-device-template

Hi
i just ran into your post about the DKC drives
Did you manage to solve your problem with the sercos protocol or did you replace the driveline?
I am in the process of retrofitting a bof411 Homag with the same driveline as yours and  Ihave successfully modified the DKC drives so they now run analog by replacing the firmware
Feel free to reach out if interested.

 

Ok Peter, so that is resetting my original understanding. But I have been unable to get the index pulse, which is clearly present in halscope to trigger the position reset.

The way index works on LinuxCNC with quadrature encoders is that when LinuxCNC (or some hal component)
wants a encoder reset at the next index occurrence, it sets the index-enable pin. This arms the encoder counter
hardware so at the next index, two things happen, the encoder position is reset to 0, and the index enable pin
is set false, so LinuxCNC or the controlling hal component knows that the index has been detected (and position zeroed)

I will install LinuxCNC onto a random (not-super-ancient) HDD.

I hope it goes well.

Weird.

FreeBSD seems faster.

Why would they switch to Debian?

landoneilers wrote:

Why TrueNAS over FreeBSD?

(I know TrueNAS is based on FreeBSD, but still)

It's not, new version is Debian as far as i can gather from the boot time info.

landoneilers wrote:

I have never seen a motherboard with more than 4 SATA ports.

I've always had with more than 4 as they also had a separate SATA controller with RAID, still have 2 with 6 SATA ports, and all Enterpise towers also have 6 or 8 ports (HP Z420 and Z600)

Yes I will, I did take some photos and know what's the system PLC/System they use, I'll have to come back I hope later today to upload some photos about it.

Does HDD cache size have any impact?

HDD read/write has no impact on latency.which measures hardware performance, network latency is important for mesa/ethercat systems.
The buffering of gcode will be sufficient to cover HDD speed.
Using a SSD is  a great boost in performance and I usually install one when refurbing a machine. Recently I didn't on a 8 core i7 & I really noticed how much longer Debian took to install. 

Use the one you say is ideal

Why TrueNAS over FreeBSD?

(I know TrueNAS is based on FreeBSD, but still)

I have never seen a motherboard with more than 4 SATA ports.

I don't really need NAS, though that is a cool idea, I suppose.

This is a snapshot of latency of that PC when I tested servomotors.

I'll try exact same former procedures ( I have written something important about disabling eoe, and recalling everything as much as possible ) with former PC today.

Thank you so much for your every skillful reply.