Is it realistic to write a new firmware for the dongle? Is it to improve on the reset?

Sounds like you identified blocking wait states as at least one big part of the problem.
Realistic to rewrite those sections in the component?

fyi, I have the radio version, works well for me needs.

I just arrived but I just fixed a new Chinese 12 feet CNC press brake, and I have to say the control box HMI and the motor pump is very clean installation. The one that I fixed was 120T. I recommend you to build your own from a 5-10HP motor low speed, buy the oil pump, just the cylinder block. and connect 1/2" hoses to your reservoir. The already made will not have what it takes to be sufficiently powerful to be happy with repeatability.

Thank you Peter, thank you Tom.
I will try to increase the interval.
No pressure with these MX2. I have a running solution with the USB-Dongle.

Thank you !

Sorry, I was on mobile, not on PC so I wrongly uploaded platformio.ini. 
I am using SPI2 pins as per your previous suggestions and it was working ok in remora-spi with TMC.
So I am using same for ETH version.
I am uploading actual platformio.ini
 

Please look at ldscript, which is I have made with reference to already available ldscripts.

If I didn't look at the second picture I didn't have a clear idea on how to make the steppers interact with your valves, they are mechanically driven like a joystick right? Y could move them in portions or one way or other. But with the motors they are electronically regulated. Which is so, so nice! also the clean and direct way of doing it. Is SUPER! Even if you don't have a proper feedback to know if both hydraulics go down exactly as they should. You could even use a g2 timing belt as a rack and small pulley with a DIY encoder disc to count movement.
Thanks to your photos I have many ideas now.

Hi, 
So you are using the spi1 pins for the w5500? I think this will be a problem if your config uses tmc5160, as the module uses software spi and configure the pins as standard gpio. 

If you have time, please watch the video.
Even with a very small P value, the axis starts to oscillate.

PS. I understand that Linuxcnc is fighting with the internal PID controller of the servo driver.

This example config include a separate ohmic.hal and the hypersensing component is in the components folder.
github.com/rodw-au/showstopper

Like it could be a cooling issue? I guess I can pull the cooler off but the fans are barely spinning and fdoesnt feel warm at all.

I am trying to configure the spindle as a C-axis, but I am having a problem and cannot find a solution.
Could someone please take a look at my configuration and the video and give me some advice?
I was not able to post this in the original lathe thread.





#########################
# AXIS C JOINT 2
#########################

setp  pid.c.Pgain     [JOINT_2](P)
setp  pid.c.Igain     [JOINT_2](I)
setp  pid.c.Dgain     [JOINT_2](D)
setp  pid.c.bias      [JOINT_2](BIAS)
setp  pid.c.FF0       [JOINT_2](FF0)
setp  pid.c.FF1       [JOINT_2](FF1)
setp  pid.c.FF2       [JOINT_2](FF2)
setp  pid.c.deadband  [JOINT_2](DEADBAND)
setp  pid.c.maxoutput [JOINT_2](MAX_OUTPUT)
setp  pid.c.error-previous-target true

net c-index-enable    =>  pid.c.index-enable
net c-enable          =>  pid.c.enable
net c-pos-cmd         =>  pid.c.command
net c-spindle-pos-fb  =>  pid.c.feedback
net c-output          <=  pid.c.output 

# ---PWM Generator signals/setup---

268 net c-output            =>  hm2_7i98.0.pwmgen.00.value
269 net c-pos-cmd           <=  joint.2.motor-pos-cmd
270 net c-enable            <=  joint.2.amp-enable-out
271 # enable _all_ sserial pwmgens
272 #net c-enable            =>  hm2_7i92.0.7i77.0.1.analogen

276 net c-spindle-pos-fb    <=  hm2_7i98.0.encoder.02.position
277 net c-spindle-vel-fb    <=  hm2_7i98.0.encoder.02.velocity
278 net c-spindle-pos-fb    =>  joint.2.motor-pos-fb
279 #net c-index-enable     joint.0.index-enable <=> hm2_7i92.0.encoder.06.index-enable
280 net c-pos-rawcounts     <=  hm2_7i98.0.encoder.02.rawcounts

282 # ---setup home / limit switch signals---

284 net c-home-sw           =>  joint.2.home-sw-in
285 net c-neg-limit         =>  joint.2.neg-lim-sw-in
286 net c-pos-limit         =>  joint.2.pos-lim-sw-in


[AXIS_C]
MAX_VELOCITY     = 360
MAX_ACCELERATION = 500
MIN_LIMIT        = -900.0
MAX_LIMIT        = 900.0

[JOINT_2]
TYPE             = ANGULAR
HOME             = 0.0
FERROR           = 1000.0
MIN_FERROR       = 500.0
MAX_VELOCITY     = 360
MAX_ACCELERATION = 500

P        = 0.0000001
I        = 0.
D        = 0.
FF0      = 0.98
FF1      = 0
FF2      = 0
BIAS     = 0
DEADBAND = 0.1
MAX_OUTPUT = 10

ENCODER_SCALE    = -40
OUTPUT_SCALE     = 10
OUTPUT_MIN_LIMIT = -10
OUTPUT_MAX_LIMIT = 10

MIN_LIMIT = -1000
MAX_LIMIT = 1000

HOME_OFFSET   = 0.0
HOME_SEQUENCE = 1

Well Ive spent all day trying other installs or live versions and no dice. Cant get any to run live as they just hang at a blank cursor screen. Cant even get anything installed due to a "bad archive mirror" error.

Tried latest regular Debian and got that installed and then followed directions to install linuxcnc and got it up enough to latency test aaaaaand 4,550,000ns....

What causes this so I have some idea what to avoid in the future? What in the hardware would cause this?

We did some tests last year comparing IP network access vs raw packets
with Mesa HostMot2 and the statistical improvement of peak latency  was
minimal so decided it was not worth the effort.

There no issue with root access and LinuxCNC as direct hardware access
requires this and the blocking of normal network traffic on the real time
Ethernet channel also requires root priviledges.
 

thank you for the information Rod. mine has an hpr 130 torch on it and it uses technical gases while cutting, it is dual shielded. the probing happnens from the outer shield on wich the copper consumable touches the table and completes the loop with the ground cable. the wire for probing is connected to the outer shield itself, i can send a picture tomorrow if needed. i'll order two THCAD boards. how would i proceed to connect the one used for probing to the Mesa 7i97T? is it the same as voltage sensing? connected to an encoder input?