LinuxCNC vs Centroid

Another vs thread, bit boring I know but useful to get opinions from the more experienced.

Centroid (Step/Dir only)
Cost:Starter kit $299
Pro version + $139
Digitized + $399
Requires Windows 10 (£80-upwards)


Slick GUI, simple to setup, and reports suggest smooth running.
Looks like you need the pro version if you're serious but what does the digitizing version really offer (There's additional probing support). What do they mean by digitizing, is it still purely control, or does it encroach on some of what Aspire does?



LinuxCNC (Look ahead functionality, step/dir/velocity/torque mode)
cost: (7I83 + 7I76E) - $280, open source software
The above boards probably provide, and some, all the outputs I'd need. I see others purchase the small 7I73 for a pendent but I believe the two boards do include that connectivity.


There is a modded GUI (gmoccapy) but not with the polish of the Centroid. Users do start projects creating new ones but I guess being busy people, they stay in development, or are for personal use.
Compared to other combinations, there isn't the same video content showing setup and usage, although there is a strong forum community. (I've used Linux before, wouldn't go back to using it as my main OS but perfect for something like this)


The above Mesa combo provides analogue feedback. Can anyone summarise the advantages of having this feedback, and re using DMM servos, do they work as intended with said feedback circuit?



Out of the two, can anyone give a comparison, is one obviously stronger than the other. The Centroid obviously works out more expensive, I've also got a number of old PCs that would run Linux fine, so no extra costs there.


Thanks folks

The Centroid Acorn board and software are done by professionals with years in the business and yes you need the Pro software. LinuxCNC is done by...

One main difference is that LinuxCNC is open and extensible where Centroid is closed and you will get charged for every option.

LinuxCNC is probably more work to set up but you get much greater control/flexibility than you have with Centroid.
If you have a simple system and you are not doing anything out of the ordinary Centroid will be likely simpler to set up.

If you need more than the basics LinuxCNC is probably a better choice.

Hardware & I/O wise the 7I76e is more capable than the Acorn:

32 digital 12/24v inputs (4 can be borrowed to support 2x real time MPGs)
16 digital 12/24v outputs
5 differential step/dir outputs to 10 MHz
Isolated Analog spindle interface for VFD
1 spindle encoder input (single ended or differential A/B/Z)
2 parallel expansion ports (for additional high speed I/O like step/dir, encoders, PWM etc)
1 serial expansion port

In general you would not need the analog I/O

Quote:

---

Originally Posted by wmgeorge

The Centroid Acorn board and software are done by professionals with years in the business and yes you need the Pro software. LinuxCNC is done by...

---

Mesa electronics are professional products and from what I've seen they are bullet proof products . Linuxcnc , originally emc has been built by a very long standing community who fully support the software and are always working on it .

There is actually some interesting developments with GUI's with LinuxCNC. Like this one
https://qtpyvcp.kcjengr.com/showcase/mill_vcps.html

Also, I can't speak highly enough of Mesa's support. After I shorted out an encoder input, PCW from Mesa posted revised firmware for me overnight on the LInuxcnc forum while I was sleeping. This saved me the expense of a new 7i76e card. I don't think any other hardware manufacturer would go aboive and beyond like Mesa do to support their Linuxcnc users.

Thanks guys, appreciate the help.


Yeah the Centroid looks attractive for the reasons outlined. Costs are $540 vs $200-280 but I do believe in paying to save money later.

Saying that, the Mesa cards are very robust (with great support) and I'm trying to look ahead... i.e possibilities of laser (and pointer), plasma, probes, mpg etc. The mesa + LinuxCNC probably fit that bill better.


As to the analogue feedback, I was really just wondering if anyone knew what advantages that brings, including running speed/torque modes vs the step/dir, which is the only available mode using Centroid.

Ok I did some more digging, and saw some informative posts by @mactec54. I must confess what little I do know, is more likely to be dangerous (and posts often contradict each other) but I don't think the feedback loop via the controller will be an advantage to me.


re the 7I83, the output is 13 bit (@2.5Khz), which may include a sign, + factoring in small errors, which will limit the potential resolution. I'm guessing having the PID loop closed via the DMM drive/motor, the processing will actually be quicker, & at the motors inherent resolution, vs PID loop via controller. (i'm aware my terminology is a bit pants but I'm still getting my head around it all)

As to Step/Dir, Torque, Speed modes, again, I get the feeling I probably wont see much difference. (at the moment, I haven't worked out if feedforward requires a particular mode)


Conclusion, Mesa 7I76E + LinuxCNC is tipping the scales at the moment. Its low enough cost to make it not painful if the decision was made to switch to another control setup.

Quote:

---

Originally Posted by rvectors

Ok I did some more digging, and saw some informative posts by @mactec54. I must confess what little I do know, is more likely to be dangerous (and posts often contradict each other) but I don't think the feedback loop via the controller will be an advantage to me.


re the 7I83, the output is 13 bit (@2.5Khz), which may include a sign, + factoring in small errors, which will limit the potential resolution. I'm guessing having the PID loop closed via the DMM drive/motor, the processing will actually be quicker, & at the motors inherent resolution, vs PID loop via controller. (i'm aware my terminology is a bit pants but I'm still getting my head around it all)

As to Step/Dir, Torque, Speed modes, again, I get the feeling I probably wont see much difference. (at the moment, I haven't worked out if feedforward requires a particular mode)


Conclusion, Mesa 7I76E + LinuxCNC is tipping the scales at the moment. Its low enough cost to make it not painful if the decision was made to switch to another control setup.

---

Actually the analog output resolution does not have a very significant effect on position resolution of the feedback loop.
Consider that with LinuxCNC and a standard 1 KHz servo loop, how far can you deviate from the velocity_command*Time
in one ms? Say you only had 8 bit analog resolution (so ~127 velocity steps per direction and velocity quantization error of 1 part in 256)
so with a 600 IPM machine (10 ips) the maximum actual endpoint error due to the low (8 bit) resolution would be 10ips/256/1000
= ~40 uinch. Very low velocity setpoint resolution will cause other issues at low speeds and static positioning as will low resolution encoders
due to the poor velocity feedback they provide at low speeds.

For velocity mode drives with LinuxCNC closing the position feedback loop, the parts of the loop that require
high speed (current=torque and velocity) are done in the drive and LinuxCNC only does the (relatively low bandwidth)
position loop. With velocity mode drives, FF1 is needed and =1.0 if the output is scaled properly.

That said In general I would not suggest analog interfaces unless you are retrofitting an existing machine that
has analog only servo drives,or really need feedback all the way back to the controller.


There are some advantages of having feedback to the controller:

1. The controller can monitor/display the following error in real time
2. You can disable the drives without losing position
3. Dual feedback with linear scales is possible for better accuracy
4. If you use Torque mode, (which needs feedback) you can monitor applied torque