[dfbittner]

I am a newbie at this CNC thing and am seeking advice on controller options to retrofit a 20 year old CNC router I inherited. I'll the machine for hobby woodworking. I am not a controls person and will get some terminology incorrect. I feel confident I can fumble through some basic control system upgrades, though. The router is a K2 CNC that seems fairly well built with sound structure and acme screw actuators. The control box contains 3 Geckodrive G201 stepper drives, a decent power supply, and an old Parallel port interface between PC control and the drives. I'd like to upgrade the control system to something more modern and enable future addition of a 4th axis. I'm interested in either 1) An all in 1 CNC controller solution that does not require the PC sending G-code during machining or 2) A PC control based solution with ethernet communication to a control board. Any suggestions on preferably US based control board or onboard controller upgrade solutions with good support? A colleague suggested considering a SmoothStepper brand motion controller. This looks interesting, but has ribbon cable interface to the drives and the drives don't have ribbon cable connection. See image of one of the drives attached. Any and all advice is appreciated and please dumb down the controls terminology for me. Thank You

[peteeng]

Hi DF - I'd consider a Buildbotics but they are out of stock due to microprocessor shortage. But have a look they maybe in stock soon. This would replace your geckos its all in one box solution.... Peter

[jckstrthmghty]

Those 201's are nice stuff. Easy to add another drive. There are economical BOB's if you don't have space on the current one. You could consider going with a Ethernet Smooth Stepper. Not cheap but works well for mach3/4 setups.

[joeavaerage]

Hi,
I use Mach4 and an Ethernet SmoothStepper and have done so for eight years and Mach3 and a parallel port before that.

You are wise to ditch the parallel port, they are so last century.

An Ethernet SmoothStepper is US made, excellent support forum moderated by Andy, the company tech/developer. They cost about $190USD. They really need or benefit
from a breakout board downstream, and for that you may be able to reuse what you have.

The basic set up is that Mach3 or Mach4, the ESS works with either, runs on a PC. The ESS is connected to the PC via Ethernet and the output of the ESS is arranged as up to three
parallel port equivalents. Thus you could with an IDC-to-DB25 cable ($9) plug your existing parallel port board into it. Note that the ESS has three output ports so you could plug in another
breakout board or boards if you want or need more inputs and/or outputs.

Another alternative is to use an MB3 breakout board ($180 by CNCRoom in Malasyia) as this board is good quality, and has all three ports fully developed, including all the relays, PWM outputs,
differential motor outputs, that you will ever need to run a machine.

Just to give you a ballpark approximation of what this solution will cost:

Mach4Hobby License $200
Ethernet SmoothStepper $190
IDC-DB25 cable $9

Total $399


If you want an MB3 breakout board, a worthy choice for future expansion:

Mach4Hobby License $200
Ethernet SmoothStepper $190
MB3 breakout board $180

Total $570

So the cost Mach4 and ESS is aimed at the hobby market, but it is not cheap, but very effective. Mach4 is highly adaptable to different machines with a versatile scripting
and screen editing environment. Given that you have a machine already, and most probably of fairly conventional design, an adaptable motion control solution
may be rather moot.

There are other solutions that may appeal. For example there is LinuxCNC, it is free and open source CNC software that runs on a PC with a specialised Linux OS.
At its simplest it's output would be a parallel port, and despite my deprecations of the parallel port, the Linux parallel port is very good indeed. So for the cost
of a suitable PC to run Linux you could basically plug your existing parallel port board straight in.

Another alternative is GBRL. I don't know enough about it, but I can tell you its open source software that runs on an Arduino style board. Its widely used on 3D printers
and so on, and with some judicious modification run a simple CNC machine.

Yet another possibility is Centroid Acorn (US made). It is a software solution and combination motion controller for up to a four axis Step/Dir CNC machine. Its a sort of one
stop shop, around $350 and you supply the PC.

Craig

[jckstrthmghty]

Now if I was going PC less I'd buy Masso but there is a premium to it.

[dfbittner]

Thanks all for the excellent information thus far. Let me digest for little while, then may have follow up specific questions. Wanted to clarify too that I want to keep the Gecko's.

joeavaerage/Craig - From what I can quickly gather (again I'm not a controls type and BTW you did a pretty darn good job of dumbing it down for me) you assumed I do want to keep the Gecko's. I will read through and thoroughly understand all the options you laid out. Much appreciated content and detail, including costs!

[joeavaerage]

Hi,
yes, the options Mach4/ESS/MB3, LinuxCNC and Centroid Acorn would all retain the Geckos. I'm less sure about GRBL. Frankly I consider GRBL a bit of a toy, although I could
reasonably, and possibly accurately, be called a 'CNC snob' for it. You'll have to do your own research on that and make your own mind up as to it performance.

Craig

[dfbittner]

Wow Masso looks like great functionality, but rather pricy as you implied. I wanted to at least understand an option and price of ditching the PC to dole out the G code a line at time vs bulk offload/upload. Thank You for offering this option.

[dfbittner]

Thank you again for all the great advice from this forum. Well I've decided to go with the SmoothStepper option as a starter kit. I assume this hardware choice alone won't classify me as a 'CNC Snob', primarily because I have much to learn about configuration, debug, and use of the software suite from solid model to spindle carving out what I want. However, if I arguably become a snob in the future, it'll be hard for me to pass on that Masso unit as an upgrade.

What I like about the SmoothStepper:
-Can run Mach3 freeware as starting setup, with option to pay and upgrade to Mach4
-Retains the Gecko drives I already have
-Option to add a 4th axis. I have thoughts of adding a 4th axis for mill-turn as future setup.
-Ethernet PC interface with onboard motion control (external to PC)
-Allows plug and play hardware setup, retaining my parallel port interface board to convey stepper outputs, but without a visible parallel port (great suggestion Craig)

The Acorn looks like a very nice controller and is a close 2nd place finisher for my upgrade.

If in the future I'm in the market for a control board with onboard drives, both the Openbuilds BlackBox X32 and Spark Concepts-xPRO V5 are great options, with the latter uniquely offering 4th axis capability.

Now on to acquiring the Smo0thStepper and associated hardware, starting my journey up the incline from Newbie to a becoming competent CNC user

[jckstrthmghty]

and ofcourse the best reason for ESS is the offloading the pulse generation from the PC to the ESS for smoother stepper operation and higher reliability as the code can be queued up on the dedicated device.

[joeavaerage]

Hi,
by and large I agree with your assessments.

-Can run Mach3 freeware as starting setup, with option to pay and upgrade to Mach4

This is not correct, however. Mah3 is not free. There are so many pirate copies out there that many people consider it free, but it's not free or ever has been.

There is no upgrade path to Mach4. Mach4 is a new product with virtually no code in common to Mach3 which preceded it. Even if you have a Mach3 license
you get no discount if you want to buy Mach4.

Craig

[peteeng]

Hello DF - I have been looking at the Duet3D companies parts. They are principally a 3D printer company but support CNC and automation applications very well. They have controller motherboards that allow external drives to be attached or boards with inbuilt drivers. The boards also have modules for extra axis as many as you want 10 is common in the printer world. Its a bit of a new world for CNC (and me) but I think starting there allows things to happen into the future that your current selections don't have. The 3D printer "ecology" seems to be richer then the CNC hobby world at the moment. The interfacing will be slightly different to the CNC world but I think its worth a look. The machine control software side is free (open source) look at Marlin or Klipper. There are 100s of videos with people unpacking and setting up these things and the Duet people have happily answered my questions promptly.

https://marlinfw.org/docs/basics/introduction.html

Duet 3D

In this way the machine control, motion control and external bits are all in one integrated environment. All upgradable via software or firmware modules into the future. Vs changing many bits that are connected together. They also have wifi, ethernet and can-bus connectivity options. I think they are more forward looking. Been researching this for a few weeks,

Another board company to look at is Bigtreetech. This is aimed at the hobby 3d printer market. But their CNC side is less developed but possible. https://biqu.equipment/collections/c...for-3d-printer

this board has 10 axis capability and you only buy the drivers to suit. start with 4 go to 5 in the future no problem. Pick a drive with the amps needed. The SKR 3 is good too only 5 axis... BTT is cheaper then Duet but not as well documented on the CNC side...

Just a thought for you. If your into electronics its ideal as it will take a little time to explore this area. if you want plug and play maybe not. My local 3D printer shop has been helpful with a few insights as well.

https://www.duet3d.com/duet3mainboard6xd up to 6 external axis for 150 pound... Cheers Peter

[peteeng]

Hello DF - Just to add to the prior post. The future is in 3D printing and in the future you may say lets make my CNC into a printer. Starting with CNC software and hardware and making it into a printer is not easy. But if you start with duet then making a rotary or a 3D printer is easy all the bones are there and they have better motion control...Worth a look I feel. Peter

[joeavaerage]

Hi peteeng

3D printer is easy all the bones are there and they have better motion control..

What is it that you believe that gives this solution 'better motion control' than conventional CNC software solutions like Mach and UCCNC?

Craig

[peteeng]

Hi Craig - Mach and UCCNC use trapezoidal velocity profiles with no jerk control. Which means they are prone to vibration as they get faster ie when using HSM techniques.

Duet and BTT have input shape functions which control accelerations like a wheel damper does or noise compensation in electronics. Firstly you run a vibration test on the machine using accelerometers. This produces a vibration profile of the axis. Then using a shaping function this vibration can be minimised. This can only be done if the machine controller techs and the code techs get to together and co-operate. Thats what's happened with Duet and Marlin and BTT and Klipper. You wouldn't have a car that does not use shock absorbers so why are we using motion controllers without either jerk control or input shaping? Attached is a paper from 1995 on the technology. It's been around for so long yet it has not been implemented in CNC hobby world not sure why. But 3D printers being light and fast need this sort of thing to be smooth and go faster. It would also benefit machining and I want to build a small 3D printer to get experience with this then migrate it to a CNC. BTT and Duet also use trapezoidal profiles but the shaper is in effect controlling jerk via controlling accelerations. Using s-curve controllers is similar but does not know anything about the machine. With the shapers the control is tailored to the machine via the accelerometer at the spindle which is a better way in my view.

It works by establishing the vibration profile of the axis then once this is known an inverse signal is added to the accel signal that cancels the vibration by modifying the accel profile around those bad freqencies. . Peter

[joeavaerage]

Hi peteeng,
I will try to find the scholarly articles that concluded that 'third order planner' motion control offers the same as a reduction in peak acceleration (jerk), ie you trade cycle time for jerk control,
in short not very useful. That there is tens if not hundreds of billions of dollars worth of installed equipment in the world today, the vast majority running second order planners,
suggests that the hype around third order planners is just that, hype.

Craig

[peteeng]

Hi Craig - There is mountains of info out there on this. Most people will agree that s-curve is better then trap. That jerk control is better then s-curve. These are passive solutions to use the car industry term. Then there are active solutions ie those that use feedback. Top end servo machine use velocity, accel and jerk feedback to make very smooth motion. Cars have passive motion dampers commonly called shock absorbers and some cars and race cars have active motion control just like top end cnc machines do.

Active suspension - Wikipedia

Shock absorber - Wikipedia the input shaper is equivalent to a mechanical shock absorber

The trick is to find an economical way to control machine vibration to improve motion. I think input shapers are the go at present. Some 3D printers now have accelerometers full time in their printer heads (like mine will and my cnc) and its not much of a leap to go from front end input shaping to full active control in the future. Printers are getting faster and faster and faster and faster. Hobby machining is not (I appreciate the physics is a bit different but all other areas of technology have leapt forward) and many of the control people are a bit behind I feel. This solution does not make compliant machines better but it will improve machining. Peter

by the way I don't like the term input shaper but thats it for now

Oh and just because there are billions of $$$ of a particular type of machine being used that does not mean that something will not disrupt it. Take film cameras, glass valves, IC engines, black phones on walls, drones vs battleships etc etc.... things can change very fast when needed to. We have to look outside of the box we are in occasionally to see whats coming or whats gone past....

[peteeng]

Hi Craig - If your happy with what you have then thats fine. DF is talking about adding options and stuff into the future that he can have now (or options of into the future) instead of paying for it all over again and going thru new learning curves. Just start at Duet3D and everything he talks about is available. Probably for similar cost to up front cost of M3, smoothstepper, etc etc... Peter

https://www.cnczone.com/forums/mecha...30780-cam.html

to save DF's confusion continue the discussion in the jerk thread. Peter https://www.cnczone.com/forums/mecha...30780-cam.html & DF don't be shy asking Q's

[joeavaerage]

Hi,
I have miles and miles of expansion and improvements to make before I'd waste my time on that. I have not long finished a fourth axis. I have just taken delivery of the gearbox that
will become my fifth axis. The two axes (4 & 5) and the Machining Extensions CAM software to exploit them have cost about $6500NZD. Thereafter I'll be focusing on a third spindle,
a high torque spindle (14Nm cont, 48Nm overload at 4000rpm rated) with ATC. After that I'll be saving for a 42000 rpm HSK25 ATC spindle.

The existing Mach software with its lowly second order motion planner will still exceed my ability to exploit it.

Any suggestion that a newcomer like OP 'requires' a third order planner is ludicrous. If I'm not mistaken you use UCCNC yourself do you not? Is there something lacking in UCCNC
that a third order planner will fix?

Craig

[peteeng]

Hi Craig - I'm not suggesting he does require jerk control but with mach its not an option. Your prior entry states that M4+smoothstepper + cable is $399USD. If you buy a Duet 3 mainboard it is $180USD, can control 6 external axes and has the options of as many more axes you want, plus input shaping, plus other inputs/outputs that M4 can't do. So the starting point is scalable and cheaper. You don't even need a computer to run Marlin/Duet as its a computer within itself. Marlin is open source well supported and development is relentless. It will always be free. Blackbox and Spark are deadends aimed at very hobby level users, they have their place and I have considered them but I need something that scalable to may machines.

In regard to better machine motion my enquiries are always asking for more speed. Once we determine what their current limits are (usually leadscrew or ballscrew whip or a soggy machine) then we can move forward. The 1000-2000mm/min group want to get to 4-6m/min and the faster machines want to go even faster. So I have to solve this issue as it is a market demand. If your happy with your lot I'm happy for you

In regard to my use of UCCNC I'm happy with it for what it is but I'm looking for the next level system, that's my nature always looking for greener grass. faster, smoother, tastier whatever. In the yacht racing and auto racing industries I have worked in every year we are expected to do better. If we don't then we lose and I'm out of a job... Sorry DF we may have derailed your thread... Peter