New 7i95t Open Loop Stepper Driver Issues

I’m trying to test run my new Mesa 7i95t. I am completely new to LinuxCNC but have lots of CNC (and programming/linux experience FWIW). Before I start touching anything in my DIY machine (currently running on FluidNC) I just want to get a single stepper driver to run a single stepper motor so that I can complete the flow end to end. As a sanity check I’ve hooked up a DM556 stepper driver to the card’s stepgen 03 connector (tried both a differential connection and also just connecting PUL- and DIR- on the driver to stepgen’s ground). The stepper driver is powered by a 24v PSU and works perfectly if I just connect its PUL to a pulse generator. The ENA is left unconnected, enabling the driver. I’ve setup the Mesa config with pncconf using 7i76 as the card and then later switched to hm2_7i95.0 in the INI as per the instructions. Assigned stepgen 03 to the X axis. LinuxCNC starts fine, I can unlock the machine and jog, and see the DRO changes values - but the stepper doesn’t move. Looking into the generated HAL file, I see control type set to 1 (velocity and not position? does this make sense for a stepper driver like this)? Also, I see a bunch of pid initialization, and again it’s confusing to me why PID is needed for an open loop stepper driver that provides no encoder feedback to LinuxCNC. What am I missing here? Is there a sample HAL file that works with the 7i95t and a simple open loop stepper driver that I can test? Thanks!

forum.linuxcnc.org/show-your-stuff/47357...hines?start=0#256463
Attached Messer_big is for 7i95 (not T) and uses encoder feedback and is for plasma, so just take it as a reference to compare to.

The most common issue with lack of motion ins step/dir systems is too short
pulse width settings. I would always start with at least 5 usec (5000 ns)
for the step pulse length.

Another common issue is insufficient drive, which is why its suggested to wire 
opto-isolated step/dir drives in single ended format as this reduces the drivers
source resistance by 1/2.

Single ended:

7I95        DRIVE

STEP+    PUL+
DIR+       DIR+
GND       PUL-,DIR-

The PID setup is intentional.

Note that the latest pncconf ( in LinuxCNC master ) has direct 7I95/7I95T support

 

thanks, will test that.

btw - I have installed the latest release of the linuxcnc debian - 2.9.4 - and there is no 7i95 option, is the version on master newer but not released?

Master is 2.10

So you would need to build from source or get a snapshot from the buildbot
to update to master.

BUT... you should be able to just replace the current pncconf pieces with these:

forum.linuxcnc.org/39-pncconf/56464-mesa...thcad-pncconf-screen

( Thanks kworm )


Another option is MesaCT which has had 7I95/7I95T support for a long time.

 

pulse width was already 100,000 ns so that wasn’t the issue.

i downloaded mesact, chose 7i95t and when i switched to the firmware tab got a message that no firmware was found, prompting me to download it which i did.

then built the config, and it worked perfectly right off the bat so firstly thanks so much for the help here!

i’m curious what this missing firmware was - since no flashing happened im to understand this isn’t the board’s firmware, so i’m assuming then its firmware on my desktop for communication with the 7i95t?

before, i was using hm2_7i95 which was available to the HAL (not 7i95t which the HAL complained about being missing once I ran LinuxCNC with it initially). so this is just hm2 “driver” firmware for 7i95t which was missing on my debian?

There is no specific driver for the 7I95T, its supported by LinuxCNCs hm2_eth driver.

Normally you don't need to do anything with the on card firmware unless you have some
specific different I/O requirements. ( Like you want one step/dir output to be PWM/DIR or
you have parallel daughtercard  )

Not sure what explains the initial lack of functioning, possible an issue with the 7I76
config file conversion.

giaviv wrote:

pulse width was already 100,000 ns so that wasn’t the issue.

Was that "base-period" ?
Pulse width is 5000 by default, base period is 100000 by default, if not set otherwise, but base period is only present when you select using a parallel port, and this might have been the issue with stepgen numbering etc...