Advanced Search

I think it is something the Rpi does at regular intervals, drops the network interface for some reason.
Inspired by some googling can you search through the syslog to see if you find why eth0 is reset?

Edit. Another idea is to configure NO ip address on the interface.
Maybe it can prevent the os from doing ip-related things to the interface.

I tried this ethtool --set-eee eth0 eee off but it did not help at all. No I have tested two independent setups and inserting ethernet switch between Rpi and ETT60n solved the issue. In rpi5 there is no l --set-eee eth0 eee available but same issue when using without switch.

TL;DR: Can I wire these little TMC2226 (or similar) 3d printer stepper drivers to a 7i96?

Before you laugh I know it seems like a silly matchup. I've been out of the cnc game for a few years now but now I have some projects on mind some I'm trying to plan to reuse some old parts. I'm still fairly out of the loop on the advances made in linuxcnc and hardware since.

I want to build another one of these light weight "3d printed" cncs . By default these machines used arduino/grbl based controllers , but I enjoy linuxcnc so I fully intend on on using linuxcnc instead.

I have a NIB Mesa 7i96 that I never put into production .  I have some raspberry pi 4 Bs laying around from old projects. I would need to find new stepper drivers. These tiny stepper drivers are proven to work fine for these machines. I like the idea of cost and especially space savings with these little chips, and how the original design of the above linked machine integrates their electronics into the gantry. 

So can I just wire these TMC2226s into the 7i96 as a drop in replacement for something  like my old TB6600 s?

I don't quite understand yet if these little chips with all their pins, and uart mode, are a different animal than the old stepper drivers I used before.

PS. I'm excited to find out that there is an official release for the rpi. I've already flashed it and have been poking around in it.

Hi there, 
I implemented an DIY Arduino a few weeks back into my automatic injection molding machine. So cool to see that topic popping up again!
For the background, I have two CNC machines running on LinuxCNC as well and I am familiar with some Arduino-Coding too.

In the beginning I evaluated the exact same question as you here and went with the Arduino based PLC-Route as it is easier to implement hardware and manage timers easily, which there are a lot of in injection molding as many events in a cycle are time-based.

A few lessons learned so far were:

• The evaluation of "Readymade" ESP32/Arduino resulted that there is no real all in one provider like Beckhoff Ethercat modules. That drove me away from it as you still need additional "raw PCB boards" for e.g. temperature input. Then "real" PLCs like Ethercat are freaking expensive if you need anything but standard IOs. Could be an option if you base everything on second hand components!
• These modules are really nice: sequentmicrosystems.com/products/eight-t...bj3cF5BPXDmnSfa68zMp
  They are a bit more expensive then a MAX31855 etc. but a very clean solution, easy to install and you can just stack it on top of a Raspberry PI. And are very well documented too + some Libraries exist. Not affiliated with them, just made a really good impression!
• Having temps all in one controller is nice for process debugging
• Used Breakout boards as Optocouplers, SSR Output boards (240V AC 1-2A), external SSRs (40A) and DC Relay boards (24V) and the Temp input board of Sequent Microsystems.
  To be honest I would next time spend a bit more with Sequent Microsystems - they seem to be fairly well priced and offer a lot more IOs and flexibility then the "All in one" Arduino based PLCs. But if money is a big constraint you can go with the single modules too.
• I prefer SSRs just because they are silent, and basically free today and never had a cheap Asian branded fail
• For the IO side Arduino is pretty nice and fast as it is easy to connect components and debug them. E.g. for installing a linear scale I did not need a Mesa card or other boards just connected it to an interrupt and fed it to an existing Library.
• I used BlockNot to create parallel tasks, such as asking for sensor states at fixed intervals, have interrupts for linear encoders, run all the cycle timers... that is a very cool lib to work with!
• Okay now the butttt...
  As soon as you want to build an GUI for it it becomes annoying. Maybe its my lack of experience for it or I am not aware of better solutions. Let me know if anyone is aware of a similar GUI solution to LinuxCNC (in flexibility) but for simple controllers?!
  I tried to use Nextion Displays - which are displays with their own IC. That is cool as they mostly run smooth without Display lags in comparison to other solutions and have a GUI editor. 
  The pain is that you need to manage the whole communication between the display IC and the IO-IC (e.g. Arduino) manually. Eg. read button pushes from the display and feed data back. That is not impossible but a ton of (unnecessary?) work. Additionally the GUI modules come with good limitations in all directions- e.g. you can only have 4 data inputs into a diagram display which is already limiting on e.g. temp tracking. The graph you get is also not really useful in manner of navigation as zoom/ set scale etc. There are many things that need to be build to get something usable.

So I got stuck and ran it from basically a serial console + manually edition code for changes. That kind of works for home/DIY but is not a "solution" I would say as I run this machine for work. Okay that is a bit funny, but its actually some kind of special setup that does not exist on the market.

So my position is that I have a working machine, but no real interface.
I consider now to mange all the IO's by Arduino as nothing is really time critical and the linear encoders in my app don't need to be updated to the GUI in real time. You can also mix both options and use a motion card like Mesa or a simple Breakout card for motor control and safety IOs and do the rest with an external "PLC". 

Some strong points for LinuxCNC in this application:

• Easy to build GUIs
• Real time IO's for motors+safety
• ShowHAL+HalScope is priceless for real-time debugging of states, that would need to be programmed as special IO's otherwise
• Settings can easily saved in files - I would miss-use a .nc file to save projects, temps and timings?

Weak points for LinuxCNC in this application:

• Calculations and conditions are just painful as you need to do everything with components, which would be a simple calc sign otherwise :/
• Not sure how to implement a nice cycles with timings? Could be several timers connected to each other! Would be cool to be able to activate/deactivate these in real time on GUI input to control auxiliary devices.
• Is there an option to use e.g. the HalScope inside a GUI? Or have another graph implemented?

Let me know your thoughts and considerations, happy to exchange ideas here!

Mesa is pretty much the standard, firmware sources are open source, as you have found. Pete is rather open with his hardware designs, there are 7i92 clones on Ali express.
If you really want to have a look at well evolved ecosystem I suggest you check out this topic
forum.linuxcnc.org/18-computer/49142-lin...d-on-fpga-ice40-ecp5
This has evolved into the riocore project.
It may be an idea to contact Ollie and discuss enclosures and such. Whilst he has designed some boards his energies are mainly taken up by the firmware and software side.
So rather than another project from scratch, I honestly think offering your hardware skills would be awesome to take that project to the next level.
I'd even like to offer my help, I've been working on a custom RPi5\4 image based on the RPi build system, rather than another build system, that is basically exactly the same to setup on Raspberry Pi as a standard image.

Hello, I have a few questions.I am building a complete computer with fpga and all other boards, and I have some questions:
1. What firmware should I choose, mesa, remora or, write my own?
2. I want a universal solution, so what input/output boards should I design?
3. Should I design my own case or choose a ready-made one?
Here is one of the prototypes:
-32 outputs 24v 0.5A, 32 inputs 24V, four boards
-4 analog servo -10V.. 10V outputs, 4 encoders input, one board
-Fpga 96 IO board with buffers, one board ()
-Firmware - ported and modified mesa for SPI interface.
-software driver - modified hm2 driver from source code.
-3d printed case.
-Computer Raspberry pi 5 aluminum case

all works fine and stable 
   

Ihr habt natürlich recht! 
Der Stift war doch wackeliger eingespannt als gedacht.
Eine Fräsung in einem Reststück ergibt ein schönes Ergebnis - damit bin ich zufrieden!

 

Die Mesa-Karte ist immer noch am Weg, damit spiele ich mich vermutlich im Winter mal.

Ich bin jetzt noch am überlegen ob ich die Ansteuerung der Spindel-Drehzahl auch über LinuxCNC machen möchte, einfach um unabhängig von der eingebauten Platine zu sein.

Laut dem Blog von Entropia brauche ich die folgende Ansteuerung:
Über SPI auf den Pins 14,16 und 17 wird die Drehzahl eingestellt.
Die Datenrate ist 30 kHz; die Daten werden in 16bit-Worten (Big Endian) übertragen.
Bei einem ATMega328p mit 128 kHz Clock haben die folgenden SPI-Settings funktioniert:
SPCR = _BV(SPE) | _BV(MSTR) | _BV(SPR0) ;
SPSR = 1 << SPI2X;

Die gesendeten Werte sind linear Interpoliert zwischen:
14000 RPM in Editasc -> 2866
20000 RPM in Editasc -> 4094 (0x0FFE -- erster Nibbel soll stets 0 sein!)

man kann die Spindelgeschwindigkeit auch mit 64kHz (= 8MHz F_CPU / Divider 128) senden.

Wäre das eurer Meinung nach eher ein Fall für einen externen Arduino, oder kann das der Raspberry ebenfalls erzeugen und ausgeben?

I maintained series cabling, switch is just between rpi and first ECT60. Only purpose for switch was isolate Ethernet disconnect to side or other. After adding switch i cant see ethernet disconnect messages in rpi kernel logs any more. I am having RTT60V202 versions.

 

I haven’t run my ect60 drive for more than 30 minutes, so I haven’t seen this problem.

I appreciate all your documentation however. I’m also running a pi4b.

When you installed the Ethernet switch, did you maintain the series cabling for all the drives? Put another way, do you have just the pi and the first ect60 connected to the switch?

I wonder if a loop configuration using the Ethernet port on the pi, plus one more usb dongle Ethernet port would also solve the problem.

unknown wrote:

I'm not too sure if this may help
github.com/MichaIng/DietPi/issues/6577

The power saving feature shouldn't be kicking whilst there is traffic, but stranger things do happen. And I know the RPi is on a later kernel.
Actually that reminds me I'll have to talk to Rod about Ethercat repos for the new image.

Hmm, that puts the kybosh on the kernel.org docs saying EEE is disabled by default. Always disable it!

I'm not too sure if this may help
github.com/MichaIng/DietPi/issues/6577

The power saving feature shouldn't be kicking whilst there is traffic, but stranger things do happen. And I know the RPi is on a later kernel.
Actually that reminds me I'll have to talk to Rod about Ethercat repos for the new image.

That one i have already tried but it did not help

Well, I don't have a pi with linuxcnc and only googled a bit.
forums.raspberrypi.com/viewtopic.php?t=362866#p2229969 worth a try I guess if you haven't done so.
 

May be I have found least workaround and some glue what's wrong. I tested adding ethernet switch between my Raspberry Pi 5 and ETC60 Ethercat stepper controller. After adding this switch i have not got any synchronization errors or got Raspberry ethernet link up/down causing that error.

Because i have got same error between multiple tested Raspbelly Pi 4 and 5 boards and least three controller boxes having Rtlligent ETC60it cant be just a faulty unit but more common compatibility issue. Least adding switch is workaround fix.

My general experience from Linuxcnc / Ethercat / ECT60 is very positive. First I replaced always failing PC hardware with Raspberry Pi and PIO controlled steppers but I wanted better and Ethercat felt best choice. In beginning i needed to build everything from sources. This joint x motion error / synhcronization error was only unresolved issue and it took long time to find out that it was not caused by other factors like insufficient real time performance.Now when everything works and there is easy installation package i think that it will be my number one option. Next I will try Lichuan Ethercat servo and lets see is do I need switch between Lichuan ethervcat controller or not

here is the result of running sudo dmesg


[ 4.057274] Bluetooth: HCI UART protocol Broadcom registered
[ 4.083683] brcmfmac: F1 signature read @0x18000000=0x15264345
[ 4.087327] brcmfmac: brcmf_fw_alloc_request: using brcm/brcmfmac43455-sdio for chip BCM4345/6
[ 4.087734] usbcore: registered new interface driver brcmfmac
[ 4.291603] brcmfmac: brcmf_c_process_txcap_blob: no txcap_blob available (err=-2)
[ 4.291861] brcmfmac: brcmf_c_preinit_dcmds: Firmware: BCM4345/6 wl0: Apr 15 2021 03:03:20 version 7.45.234 (4ca95bb CY) FWID 01-996384e2
[ 4.393502] Bluetooth: hci0: BCM: chip id 107
[ 4.393719] Bluetooth: hci0: BCM: features 0x2f
[ 4.394870] Bluetooth: hci0: BCM4345C0
[ 4.394873] Bluetooth: hci0: BCM4345C0 (003.001.025) build 0000
[ 4.426812] Bluetooth: hci0: BCM4345C0 'brcm/BCM4345C0.hcd' Patch
[ 4.462459] macb 1f00100000.ethernet eth0: PHY [1f00100000.ethernet-ffffffff:01] driver [Broadcom BCM54213PE] (irq=POLL)
[ 4.462471] macb 1f00100000.ethernet eth0: configuring for phy/rgmii-id link mode
[ 4.469213] pps pps0: new PPS source ptp0
[ 4.469263] macb 1f00100000.ethernet: gem-ptp-timer ptp clock registered.
[ 4.470153] warning: `iwconfig' uses wireless extensions which will stop working for Wi-Fi 7 hardware; use nl80211
[ 4.470300] ec_master: loading out-of-tree module taints kernel.
[ 4.470306] ec_master: module verification failed: signature and/or required key missing - tainting kernel
[ 4.475936] EtherCAT: Master driver 1.6.8 1.6.8.g2543cc5-1+27.1
[ 4.476617] EtherCAT: 1 master waiting for devices.
[ 4.490862] zram: Added device: zram0
[ 4.628070] ec_generic: EtherCAT master generic Ethernet device module 1.6.8 1.6.8.g2543cc5-1+27.1
[ 4.628086] EtherCAT: Accepting 2C:CF:67:24:E6:CD as main device for master 0.
[ 4.628096] ec_generic: Binding socket to interface 2 (eth0).
[ 4.645137] EtherCAT 0: Starting EtherCAT-IDLE thread.
[ 4.747291] Bluetooth: BNEP (Ethernet Emulation) ver 1.3
[ 4.747296] Bluetooth: BNEP filters: protocol multicast
[ 4.747300] Bluetooth: BNEP socket layer initialized
[ 4.752789] brcmfmac: brcmf_cfg80211_set_power_mgmt: power save enabled
[ 5.129671] Bluetooth: hci0: BCM: features 0x2f
[ 5.131033] Bluetooth: hci0: BCM43455 37.4MHz Raspberry Pi 3+
[ 5.131037] Bluetooth: hci0: BCM4345C0 (003.001.025) build 0342
[ 5.131328] Bluetooth: hci0: BCM: Using default device address (43:45:c0:00:1f:ac)
[ 5.150144] Bluetooth: MGMT ver 1.23
[ 5.155700] NET: Registered PF_ALG protocol family
[ 6.530993] macb 1f00100000.ethernet eth0: Link is Up - 100Mbps/Full - flow control off
[ 6.537016] EtherCAT 0: Link state of ecm0 changed to UP.
[ 6.545022] EtherCAT 0: 4 slave(s) responding on main device. Re-scanning on next possibility.
[ 6.545024] EtherCAT 0: Slave states on main device: PREOP.
[ 6.545025] EtherCAT 0: Re-scanning now.
[ 6.545194] EtherCAT 0: Scanning bus.
[ 7.356283] EtherCAT 0: Bus scanning completed in 808 ms.
[ 7.356289] EtherCAT 0: Using slave 0 as DC reference clock.