Advanced Search

I think I may have found the issue:

There's a bad 7i97t_7i74d.bit firmware file around

If your file is dated 2023, it's bad

The current one, fetch-able from Mesa's  web store works.
(dated May 2025)
 

Thank you sir for a detailed response. It confirms what others have said, it wasn't a DNS issue. It wasn't a routing issue, it had nothing to do with ROOT servers. It's the response time from the server hosting the forum.

Should we change the tittle to G83 ?

From here it sure doesn't look like a DNS issue. 
It looks like a TLS handshake issue. As I mentioned earlier when comparing times to download the logo with and without http/https:

Here is a screendump of wireshark. The delay happens after the client (my Firefox) sends TLS Client hello. It takes sometimes a long time to get back a Server TLS hello up to a very very long time and even timeout. Right there. This time it took 17 seconds.
So the client has long ago gotten an IP address, any DNS queries was a long time ago. It's not DNS.
 
Even if I use the IP-address itself I get the same response. Later on firefox start to complain about certificate, but that's another story.
The network traffic and network delays are exactly the same with and without FQDN. So it's not DNS.

It must be that the traffic goes via cloudflare to the webserver instead of directly to the forum web server.
 
That was a surprise and rendered a bit of googling. QUIC protocol instead of TCP.
Of course, with so different network transport, no wonder one is faster than the other.
But why is the original TCP traffic so slow. Especially the TLS handshake?
 

@PCW
I actually only talked about one problem, or rather, for me it was only one problem, namely that no setting leads to the index signal being recognized or responded to.
I tried changing the jumper settings to TTL signal to see if that would trigger a response, but unfortunately that wasn't the case.
For example, I went through all possible combinations from 000 to 111 to see if anything changed:

setp    hm2_7i92.0.encoder.00.index-invert 0
setp    hm2_7i92.0.encoder.00.index-mask 0
setp    hm2_7i92.0.encoder.00.index-mask-invert 0

No difference was noticeable with either the RS-422 or the TTL signal.

@tommylight
That's right, I'm not logged in as root; I just checked again, but you shouldn't do that under Linux anyway. Anyone working on the machine could destroy the entire system. Besides, everything is running smoothly except for the problem with the index, and it's not due to permissions.
I don't use a parallel port; a 7i92t + 7i76 + 7i85 is built in.

However, I have since discovered that there are often problems with the DC11 GMS and the index signal; it is either very weak or not output at all.
I probably have two such units. But since I don't want to waste any more time on this problem, I'll leave it alone and work with it as it is now.
Thank you very much for your support, and I'm glad this forum exists.

Greetings
Olli

 

@langdons
The bird nest is quite a issue,sometimes it catchs on the workpiece causing a spike on spindle load, it can even break small diameter drill. I dont think add magnet should helps since the swarf was tighten on to the drill when pulled.
Are you sure about steel wool? Mine are sharp af, got numerous cuts just to pull it out with a plier.

Standing infront of the machine for hours just to remove chips defeats the purpose of a CNC machine.i tried manually adding M4 wait for 3s then M3 it works like a charm, but also when camming a hundred holes, that shit gets old real quick.
There is no locknut in the spindle so spinning it backward shouldn't be a problem. I'mma add braking resistor to the drive so that should be fine too.
@unknow
Remaping is way way over my head and messing around with it can crash my machine (I broken a spindle shaft once, it was a sad month). I may take a look on subroutine.
i drilled some decent steel, and the chip break nicely, just that I have shit steel. Good steel is too far to get and not every part required one.
@tommylight & Aciera
It's a typo on the tittle, I always use peck drill

Hey Guys,

I´ve bought a probe for my mill and installed the versa probescreen v1 in gmoccapy. I´ve used this source (forum.linuxcnc.org/48-gladevcp/46467-pro...-v1-for-python3-gtk3) because I´m using LinuxCNC 2.9.1. I don´t want to use v2 cause I have a lot of things programmed for the toolchange in my .hal file I don´t want to change. I just like to probe xyz and auto zero. 

After probing the probe is not at the corner of x and y and the zero point is offsettet. 
Similar issue for probing in Z direction. Zero is not set at the surface.

Is there a missunderstanding of function or is it wrong? 
I´ve uploaded a video of the issue. Maybe you´ll see something 
Edit: Updated Video (cause of wrong video cut)


 

Okay, thanks, I will try it out and keep you updated

I will post the full config when I am done

The second image shows seven tabs playing YouTube videos.

I will use the laptop solely and exclusively for the CNC. I will not run anything else at the same time.

Is my laptop suitable? It is a small CNC. I use a feed rate of 1,000 mm/min, and I will use the 5 motors/axes to machine a thin plastic plate (little effort). I will machine molds on 3 axes.

First picture is OK, the second is not, so:
forum.linuxcnc.org/18-computer/40285-bios-settings-pictures
Se what you can disable there, test, if the result is still bad, try this but try this at your own risk as it requires messing with kernel parameters and might leave you with a non bootable system, not terrible if you are still inn the testing phase as you can reinstall in about 5 to 15 minutes depending on the PC/USB/HDD/SDD/NVME speed.

Bumped into this today, goes in great detail on how to tweak and tune for good latency:
dantalion.nl/2024/09/29/linuxcnc-latency...arameter-tuning.html
Not my article.

Your picture here
forum.linuxcnc.org/42-deutsch/56803-glas...85-s?start=10#334968
shows several warnings about permissions, it is not just index, so if you are running as root user, switch over or reinstall as LinuxCNC will vigorously refuse to run properly as root, as will most of the Linux "user space" software.
Also, are you using a parallel port? Why does it say that in the error report?

Ok

When, you say  "unfortunately the result is the same" , it would be helpful if you
specified the actual results as there have been at least three separate issues discussed here.

What's going on with this site? It takes forever to load, or it's not even accessible. I've been trying to reply here for 35 minutes.

@tommylight
All other changes are executed, so why should it be any different with the index signal? Yes, it should be executed as ROOT user.

@PCW
 I also set the jumpers to TTL and tried it out, but unfortunately the result is the same.
Here are the HAL and INI: