Questions about a Mesa 7i76E build with CL86T steppers

I am gradually building out a highly experimental linuxcnc router, mostly for my own education and research, but with the goal of eventually making it quite functional.

I previously have a bit of GRBL CNC experience, but am otherwise diving into the deep end. I'm a software guy (with 20+ years of Debian experience), so things are good on that front, but the electronics side is all fairly new to me.

Right now, I have one 7i76E and this closed-loop Nema 34 kit , featuring a CL86T .

I have been successful at following wiring things up using the handy connection sheet and running pncconf has generated a config that is able to jog the motor and move in response to the test program.

My main objective in getting this closed loop driver and motor will be achieved just through lower power drain that they will enable, which is important, as I will be sometimes running this setup off of a large 48V battery bank.

I am, though, also interested in various ways of closing the loop back to linuxcnc, as well, mostly to enable the "research" I mentioned above. I'm interested in adaptive algorithms, and what is possible through extreme monitoring, including vibration and sound sensing, current and torque sensing, etc.

So, to that end, my first goal is to just know how hard the motors (both the steppers as well as the future spindle I'll be adding to this) are working. I don't need it to be fast enough real time control, though I would like to be able to do soft-real time display of these logged parameters.

So...what's the best way to achieve my goal of monitoring/logging the motor current with a reasonably high(kHz) frequency?

• Hall effect current monitor using either a 4-20mA or +-4V output?
• Tap into the encoder information between the CL86T and the motor? (no idea if this is possible)
• Use a completely different driver/motor that are designed for this kind of thing?
• Something completely different?

Related question, would a 7i47 be a good addition to this setup in order to enable more/varied analog inputs? 

Bump.

Or is there a better place to ask this?

tupshin wrote:

I am gradually building out a highly experimental linuxcnc router, mostly for my own education and research, but with the goal of eventually making it quite functional.

I previously have a bit of GRBL CNC experience, but am otherwise diving into the deep end. I'm a software guy (with 20+ years of Debian experience), so things are good on that front, but the electronics side is all fairly new to me.Right now, I have one 7i76E and this closed-loop Nema 34 kit, featuring a CL86T.

I have been successful at following wiring things up using the handy connection sheet and running pncconf has generated a config that is able to jog the motor and move in response to the test program.

My main objective in getting this closed loop driver and motor will be achieved just through lower power drain that they will enable, which is important, as I will be sometimes running this setup off of a large 48V battery bank.

I am, though, also interested in various ways of closing the loop back to linuxcnc, as well, mostly to enable the "research" I mentioned above. I'm interested in adaptive algorithms, and what is possible through extreme monitoring, including vibration and sound sensing, current and torque sensing, etc.

So, to that end, my first goal is to just know how hard the motors (both the steppers as well as the future spindle I'll be adding to this) are working. I don't need it to be fast enough real time control, though I would like to be able to do soft-real time display of these logged parameters.

So...what's the best way to achieve my goal of monitoring/logging the motor current with a reasonably high(kHz) frequency?

• Hall effect current monitor using either a 4-20mA or +-4V output?
  • Tap into the encoder information between the CL86T and the motor? (no idea if this is possible)
    • Use a completely different driver/motor that are designed for this kind of thing?
      • Something completely different?
        
        Related question, would a 7i47 be a good addition to this setup in order to enable more/varied analog inputs? 

Sounds interesting, and a rather ambitious for a first LCNC project.

First observation: with some (rare) exceptions, stepper motors/drives will not accomplish what you want regarding data from the motors for analysis.

Stepper drives, even closed loop, generally operate at full current to the motors regardless of the motor load during a commanded move.  Nor do most drives have an output signal (analog, serial, or other) which could be used by a motion controller for a feedback scheme.  In most cases stepper drives have a function called (more or less) "idle half-current" where the drive drops the current to some value when the motor is stationary.

Stepper drives deliver full current to the motor, and the motor either achieves the commanded position or it doesn't - i.e. misses steps.  Closed loop drives can see if the motor has rotated the desired position and throw an alarm if the number of missed steps exceeds some user-programmable value.  They don't have a PID controller continuously measuring the current required to stay within the permitted following error - much less the ability to communicate that information to the motion controller or user.

As noted above there are exceptions.  Galil motion has a stepper control scheme where they manage a stepper motor like a 2-phase brushless servo and the coil current is continuously modified to stay within the following error.  Copley Controls, and perhaps some others, also have similar non-traditional stepper control schemes.  But those drives are not cheap (except on ebay) and they aren't simple.  And on the Copley (Stepnet), only the most recent models have an analog output available for use by the motion controller.  Without an analog output you'd have to write software to query the drives continuously over rs232 or some other data protocol, and then figure out how to get that data in to LCNC.

So yes, it's possible.  No, I'd not recommend someone just diving in to this motion control rabbit-hole try to tackle that.

If you really want/need load feedback from the axis motors without a ton of fuss, you need to switch to brushless servos with drives capable of monitoring and reporting the output current - either through analog, serial, or ethercat schemes.  This rules out the Clearpath servos - they don't have outputs like that even though they are 'real' servos.

The other thing to ask yourself is: what's the point?  Why do you need to know the axis motor load?  It's easy, even using steppers, to 'close the loop' to LCNC by adding linear scales to the axes.  This closes the position loop and you let the drives (dumb or smart) sort out the current and speed loops.

If you're interested in adaptive machining, then the VFD or other spindle control drive is very likely to have a current output function.  This signal goes to LCNC, and LCNC just slows down (or speeds up) the axis motors based on spindle load.  

Thanks for your response. I am aware that stepper motors normally operate at full current. However, these CL86T driven ones close the loop in the stepper driver, which dynamically adjusts current/torque. From the introduction section of the CL86T manual :

The system includes a 2-phase stepper motor combined with a fully digital, high performance drive and an internal encoder which is used to close the position, velocity and current loops in real time, just like servo systems.

And then in section 4.4:

The motor current will be adjusted automatically regarding to the load or the stator-rotor relationship. However, the
user can also configure the current in the tuning software. The configurable parameters include close-loop current,
holding current[...]

So this is why I think I can treat these motors as a bit of an exception to your main point.

I'm interested in areas of research adjacent to this work:
www.hindawi.com/journals/sv/2015/406294/
which was done using servos. Part of what I want to find out, is whether similar research can be done using this newer breed of closed-loop steppers as well.

In general, the effort required to get the spindle to a given point is the important data point, as this stepper should be able to throw an alarm if it is actually missing steps, so scales won't be that much use for this particular analysis.

So, given all of that, I'm optimistic that a Hall effect monitor could provide a meaningful stream of information for (mostly offline) processing. The one linked to  does 4-20mA output, but there are others that do +-4V. There's a rather insane selection here ,

I hope that clear things up a bit, and explains what I'm trying to do, even if it is quite odd.

Thanks again.
 

Sorry, the battery part of this scenario was probably too much detail and overly confusing. I live nearly off-grid, and am building this CNC in a 7x14 cargo trailer. The cargo trailer is equipped ith a 48V 300 amp hour battery bank and has an inverter able to put out nearly 14KW at 220V.

It will dock to shore power most of the time, so the amp hours aren't always critical, but when using it as a mobile shop and unhook it from shore power, I want it to be as energy efficient as possible. See my other response for a more detailed technical look at what I'm trying to do wrt current monitoring.

Thanks