[mechhamo]

hi members

i am converting a knee mill to a 4 axis cnc and i have been looking at different controller software for my conversion and have chosen to use linux cnc ,not so much thats its free but from what i have read it also looks to be very capable, my question is how hard is it to learn linux cnc, i have some programing knowledge but know nothing about g code
and further am i missing other software out there that is as capable and easier to learn as linux cnc that members know about,i have also read that mach 3 is limited when it comes to servo drives ,is this correct

Also if i was to use linux cnc what motion control board would be best for this application considering i want to control spindle speed using a vfd and linux cnc so the motion board would have to have enough inputs and outputs for limits and homing switch as well as other outputs like lubrication and vfd , also be able to drive a 5th axis, any help would be very appreciated


kind regards wayne

[joeavaerage]

Hi,
I'll leave the LinuxCNC questions to those who know more than I.

Mach3, or better still Mach4 can run pulse input servos. Just about all modern servos have Step/Direction pulse inputs and just about any control software
that produces Step/Dir pulses will run them fine. In many respects modern servos work identically to stepper drives, if you can drive a stepper you can drive a servo.

There are still a few analogue servos around. They require (from the controller) an analog voltage, and the controller has to close the position loop. Mach3 and
Mach4 CAN be made to do that, but only at the expense of hardware and simplicity, and overall I would say that Mach3/4 are not well suited to analog servos.
Analog servos, if you can still find them are more expensive than modern AC servos anyway. The only reason you'd shag around with analog servos is if you already
had them and you wanted to re-use them....poor sick little puppy!

i have some programing knowledge but know nothing about g code

You probably don't need that much programming skill to get LinuxCNC to work, this is after-all pretty basic stuff. What you will have to learn is the rudiments of Gcode,
all motion control software including LinuxCNC, Mach3, Mach4, UCCNC etc ALL use Gcode. Get used to it.

Craig

[Al_The_Man]

I assume we are talking about analogue drives?
The advantage is when used with the likes of Galil Motion card etc, where the trajectory loop is closed back to the controller itself, they can be used with simple analogue (transconductance amplifier) drives.
So different makes of motors and drives can be mixed and matched easily.

[CNCMAN172]

When it comes to Linux the most supported hardware are Mesa boards. Check out the various Linuxcnc forums for the best cards. When it comes to process you really do not need to learn much gcode, but it is helpful to understand some of the basics. The process is to design something in Cad, many programs such as Autocad and many others. The you must design the machining strategy using a CAM program, there are many like Mastercam and a host of others. The CAM program uses a post processor to output gcode for your specific machine control.

Probably one of the most popular software packages today is Autodesk Fusion 360, that combines all the function in one software package, so I would start there

Russ

[mactec54]

hi members

i am converting a knee mill to a 4 axis cnc and i have been looking at different controller software for my conversion and have chosen to use linux cnc ,not so much thats its free but from what i have read it also looks to be very capable, my question is how hard is it to learn linux cnc, i have some programing knowledge but know nothing about g code
and further am i missing other software out there that is as capable and easier to learn as linux cnc that members know about,i have also read that mach 3 is limited when it comes to servo drives ,is this correct

Also if i was to use linux cnc what motion control board would be best for this application considering i want to control spindle speed using a vfd and linux cnc so the motion board would have to have enough inputs and outputs for limits and homing switch as well as other outputs like lubrication and vfd , also be able to drive a 5th axis, any help would be very appreciated


kind regards wayne

No, Mach3 works very well with Servo Drives and if you want more from it you would either use a low cost UC100 or Smooth Stepper hardware, all require a simple Breakout Board, Linux can be run directly with just the use of a Breakout Board but does not perform very well in this configuration, works really well with the Mesa Hardware, Linux much more to learn than Mach3 or other controls like Eding Acorn UCCNC all can do what you want, not all are 5 Axis so if that is what you really need then you have to go with what can do true 5 Axis simultaneous motion

[mechhamo]

thank you everyone for the replies , i am going to use a/c servo motors so i think what has been stated is most software will work with them ,but what i also need to know is what controller software is able to install a 5th rotating axis i have read that uccnc can't do this ,is this correct and if so what software can do this,also it is hard to get mesa boards in australia
but i have read that uccnc motion boards can be used with mach 3 is this possible

regards wayne

[joeavaerage]

Hi,
I use Mach4, and Mach4 has up to six coordinated axes. I use an Ethernet SmoothStepper and made my own breakout board which has six axes, plus a spindle (ON/OFF & PWM)
An ethernet SmoothStepper is available in Australia from Homan Designs in Melbourne.

The real trouble with 4 and 5 axis machines is not really the controller or the CNC software, but its generating the 4/5 axis Gcode. The cost of CAM that can do 4/5 axis is enough
to make you weep. I use Fusion 360 and the basic subscription costs $495/year, and I think fair value. The machining extensions which do 4/5 axis, collision avoidance, toolpath
editing etc costs another $1600/year. I was of the opinion that Autodesk (Fusion manufacturer) was just gouging, but no it turns out all the CAD/CAM softwares that do 4/5 axis are
right up there.

I've got my fourth axis up and running but I can really only afford to write Gcode by hand, and thats a PITA.

Craig

[Al_The_Man]

There is also Dynomotion/Kanalog https://www.dynomotion.com/
There is a Forum here for them.
With Kanalog you can use the simple transconductance drives I mentioned and can mix and match motors.
Also the advantage of a system where the motion loop is closed back to the controller.

[dazp1976]

thank you everyone for the replies , i am going to use a/c servo motors so i think what has been stated is most software will work with them ,but what i also need to know is what controller software is able to install a 5th rotating axis i have read that uccnc can't do this ,is this correct and if so what software can do this,also it is hard to get mesa boards in australia
but i have read that uccnc motion boards can be used with mach 3 is this possible

regards wayne

You can use UC controllers with mach3 and 4.
I've set up my UC300eth to work with mach3 and/or uccnc.
Works perfectly with both. Only tested it to 3 axis so far.

[joeavaerage]

Sorry, double post, can't quite work out how I even did it!

Hi,

Also the advantage of a system where the motion loop is closed back to the controller.

I am intrigued why you consider the feedback loop being closed by the controller is an advantage. With AC servos the loop is
closed by the drive. Are you suggesting that a generic controller can close a servo loop better than a drive designed and built
by the same manufacturer as the servo motor?

I was surprised to learn for instance that my Allen Bradley servo drive has a piecewise liner approximation of the magnetic saturation
properties of the servo and applies that as feedforward when determining the servo currents. Do ANY analogue capable controllers
have magnetic saturation models built-in?

Craig

[joeavaerage]

Hi,
second question:

With Kanalog you can use the simple transconductance drives I mentioned and can mix and match motors.

Can you still even buy a transconductance amplifiers anymore? Certainly they were all the go twenty, thirty plus years ago but since the 90's AC servos
have ever increasingly dominated the market. The last time I tried to find a DC servo with tacho-generator to replace a faulty Z axis motor on a plasma table
the manufacturer wanted $2000USD for a 100W servo!!! In the event I replaced the entire servo and amplifier with an AC servo and drive as I was able to extract
the Step/Dir pulse from the controller. Cost $400USD for servo/drive and cables.

If you want to re-purpose old servos and amplifiers then it can be done, but its dubious to me that you are truly saving any money. I buy 750W Delta B2 series
(160,000 count/rev encoder) servo/drive/cables for $435USD plus shipping. Is there an economic case for old school servos and transconductance amplifiers?

Craig

[Al_The_Man]

I am intrigued why you consider the feedback loop being closed by the controller is an advantage. With AC servos the loop is
closed by the drive. Are you suggesting that a generic controller can close a servo loop better than a drive designed and built
by the same manufacturer as the servo motor?

When the controller not only develops the trajectory but resolves the PID feedback from the system encoder etc, This allows for features such as electronic gearing and Electronic cam etc.
This also absolves the drives from any required intelligence and simple torque mode (transconductance) amplifiers can be used..
The systems I have put together over the years was based on such controllers made by Galil Motion and Acroloop etc.
One other all ready mentioned is the advantage of simple mix and matching different makes of drives and motors.
Either DC or BLDC motors.
The Galil controllers were capable of encoder feedback resolution frequency of 12Mhz, I believe now it is up to 18Mhz with current models.
Gail are by no means out of date controllers.

[joeavaerage]

Hi,

This allows for features such as electronic gearing and Electronic cam etc.

AC servo drives have electronic gearing and CAM, nothing special there.

but resolves the PID feedback from the system encoder etc,

An AC servo drive resolves encoder feedback as well or better than any generic controller, after all the servo and drive are made as a matched pair.

This also absolves the drives from any required intelligence and simple torque mode (transconductance) amplifiers can be used..

And absolving the drives of the need for any intelligence is important why? It seems that even the humble refrigerator has a brain these days, so why
would anyone be concerned about the processor smarts built into a servo drive?. Aside from anything else can you even buy transconductance amps
any more? And are they cost effective even if you do?

The Galil controllers were capable of encoder feedback resolution frequency of 12Mhz, I believe now it is up to 18Mhz with current models.

The current model DMC-40x0 is good to 22MHz. A five axis Gallil DMC4050 costs $2845USD.

https://www.galil.com/motion-control...mc-40x0#info=0

I can buy Mach4 Hobby ($200), an ESS ($190), an MB3 ($180) and five Delta B2 series 750W servos/drives/cables ($435 per kit) for $2745USD
So I can get the whole shooting match including servos/drives for less than a Galil five axis controller alone.

Over the last two decades manufacturers like Yaskawa and Siemens have been producing intelligent drives, that not only close the loop but enact single axis
motion control autonomously as part of a distributed motion control system. Do you propose such world leading manufacturers are producing lesser
products than old school analog and transconductance amp types?

If you already have analog servos and transconductance amps FOR FREE, then perhaps an analog capable controller makes sense, but who in there
right mind would go out and buy new analog servos/transconductance amps when superior AC servos/drives are available for less?

Craig

[Al_The_Man]

Why? e.g. When I build a simple PID controller using a Picmicro, they have all the modules built in the controller, trajectory controller, QEI (Quadrature encoder interface) PWM output, so all that is required is a simple non-intelligent amplifier to drive the motor.,, it would make no sense to build processing logic into the amp also. The likes of Galil use a very sophisticated version of the Micro.
Using the likes of Mach3/4 etc, which does a good job, BTW, requires a means of closing the loop as this cannot be done satisfactorily using the basic PC Windows software.
Hence the need for intelligent drive to take over the positioning.
The electronic gearing I am referring to in Galil is the capability of up to seven motors, being slaved off of one motor encoder, with the ability to change the ratio on the fly, if needed.
They have a Flying Shear example on their website.
I realize that Mach is the choice of many DIY, simply because of the price.
For me, Galil offers a viable alternative, to Mach, I have been able retro-fit using Galil with a system that approaches the Mitsubishi CNC that I have also used, but a lot cheaper.

[joeavaerage]

Hi,

Using the likes of Mach3/4 etc, which does a good job, BTW, requires a means of closing the loop as this cannot be done satisfactorily using the basic PC Windows software.
Hence the need for intelligent drive to take over the positioning.

100% correct, Windows platforms make horrible feedback controllers, but as you say the drive can adequately close the loop, that shortcoming of Windows is accommodated.

The electronic gearing I am referring to in Galil is the capability of up to seven motors, being slaved off of one motor encoder, with the ability to change the ratio on the fly, if needed.
They have a Flying Shear example on their website.

OK, that is s distinctly different capability than I had imagined (and seen called) electronic gearing. I can envisage using AC drives to do this, but they would require a secondary
encoder input channel, and changing the electronic gear ratio on the fly is possible, but fairly slow. Depending on the dynamics of the change may make it too slow to be useful.
I would concede that the complexity of doing this would be disadvantageous by comparison to a Galil controller.

When I build a simple PID controller using a Picmicro, they have all the modules built in the controller, trajectory controller, QEI (Quadrature encoder interface) PWM output, so all that is required is a simple non-intelligent amplifier to drive the motor.

If I may ask, what is a Picmicro, or rather what sort of features does it have. I assume it part of Microchips micro controller range?

I have been using C2000 Launch Pad XL, a development board by Texas Instruments which has a TMS320F2869M micro controller on it. They are fantastic for rotating machinery
control, with more dedicated periphials than you can shake a stick at, single cycle 32bit floating point multiply and add and much more stil.... and the development board offers a galvanaiclly
isolated JTAG interface to the microcontroller, sort of a one stop shop and all for $35USD. I use them for motor control projects from basic brushed DC, BLDC through to multipole
AC servo motor drives. The only external circuitry required is the high current PWM modulators/inverters, power supplies and position/current feedback networks. Is this in any way
comparable to a Picmicro?

I would agree that it makes sense to use a Galil controller if you are retrofitting an existing machine, but do you think it makes economic sense to use analog servos and transconductance
amps as a new design?

Craig

[Jim Dawson]

Hi, .......,...,.

I would agree that it makes sense to use a Galil controller if you are retrofitting an existing machine, but do you think it makes economic sense to use analog servos and transconductance
amps as a new design?

Craig

Yes, the Galil controller is very flexible and easy to use. It can output analog, step & direction, and PWM command signals. My preference when using them with modern AC servos is to set the drive for analog velocity or torque mode depending on the application. Although you could use step & direction with the inner loop closed at the drive level and set up a dual loop system. I have only done this once and had good results. With the Galil doing the motion control heavy lifting, just about any computer is adequate for the user interface.

[samco]

I think maybe we are getting a little off topic from what the OP was wondering...

This is why I love linuxcnc. The motion controller is in the computer. You have such low level access to the motion/hardware abstraction layer you can pretty much do whatever you want. (You can do things like gear hobbing and such)

You can use simple step/dir drives or you can use drives that allow you to close the loop in the computer. (or a combination of both) or even have encoder/scale feedback with your step/dir drives..

It is not a motion control black box that you hope has the features you would like.

I was wondering if I could bore non-circular holes with linuxcnc.. Wrote a pretty short real-time component to test it out. This take the spindle position from the spindle encoder and calculates where the x/y axis has to be to create a polygon.