Connecting to a CNC machine, pros and cons of USB vs PP vs Ethernet etc.

[Jess]

I think theres a little bit of a thing against laptops in the CNC world which comes from the problems you used to get with parallel ports on them

I think the parallel-port-voltage-thing was often solvable using a decent breakout board.

However, there's another reason: system management/maintenance interrupt 'stuff'. It's the utter enemy of anything that needs to be 'real time'. It's the usual reason for, say, a very slight pause every 30 minutes, or a little hitch whenever the processor gets slightly warmer and the fan speed needs to be increased or if there's a battery fault or...

Whilst all new machines seem to have more 'going on' in terms of SMI tasks, laptops tend to lead this trend, so it's just that poor performance or unreliability is more likely with a laptop.

It's less of a problem with a motion control board as that can hide smaller latency based sins but you really don't want to be a few hours into a milling job only for an intermittent computer problem to mess it up.

I did try a laptop, and the latency was fine...except for a very occasional (like once a day) SMI interrupt which put it it 'useless, even with a motion control board'. (You could also trigger it if you used any of the special laptop keys (like the screen brightness ones etc.,).

[JAZZCNC]

It's less of a problem with a motion control board as that can hide smaller latency based sins but you really don't want to be a few hours into a milling job only for an intermittent computer problem to mess it up.

I did try a laptop, and the latency was fine...except for a very occasional (like once a day) SMI interrupt which put it it 'useless, even with a motion control board'. (You could also trigger it if you used any of the special laptop keys (like the screen brightness ones etc.,).

It's not a problem using a Laptop if you have a good motion control card using Ethernet. I've been running machines from Laptops for years running jobs that last 30+ hrs without stopping and they never fail or drop out. The Same can't be said for USB it's irratic and unpredictable. There's so many things that can affect it IMO it's not a good or suitable connection for reliable Motion control.

[Jess]

Ethernet really isn't a natural partner for realtime stuff - being irratic and unpredictable was designed in from the beginning (because it's the easiest way to recover from certain network problems). Add newer things like spanning tree rebuilds and it's a big mess.

USB on the other hand, is - in theory at least - much more predictable; the host (the controlling computer) remains in control all the time.

However, despite the intrinsically random behaviour, Ethernet is much more attractive for industrial control: it works over greater distances, can handle multiple control stations and has features like isolation of multiple grounds and differential signalling built in...and it's actually an IEEE standard.

So, it seems likely that the combination of better motion control cards using Ethernet and capabilities for dealing with irratic and unpredictable behaviour having to be better on Ethernet will make Ethernet appear better, simply because Ethernet motion controllers are able to hide things that the USB motion control cards aren't able to.

(Even in the budget area this looks like it's true; unlike the USB smoothstepper, which seems to rely on a microcontroller for everything, the Ethernet smoothstepper adds an FPGA - presumably for offloading the generation of step/direction signals.)

(Also, don't forget that the only way to get an Ethernet port on some laptops is via USB, nor that some devices actually have a USB Ethernet controller on-board)

[JAZZCNC]

So, it seems likely that the combination of better motion control cards using Ethernet and capabilities for dealing with irratic and unpredictable behaviour having to be better on Ethernet will make Ethernet appear better, simply because Ethernet motion controllers are able to hide things that the USB motion control cards aren't able to.

There's No Seems about it. Ethernet is much better than USB for Motion control and I'm going thru experience using Both for CNC from several companys products not Theory or what some Spec sheet says should be better.!! . . . . BUT . . . . We are moving away from OP problem so lets get it back on focus.

[Jess]

There's No Seems about it. Ethernet is much better than USB for Motion control and I'm going thru experience using Both for CNC from several companys products not Theory or what some Spec sheet says should be better.!!

The danger of relying on experience in this way that find yourself reaching conclusions that your experience does not - and cannot - support. To put it another way; lifting 6 tonnes on rope specified for a safe working load of 1 tonne is a bad idea, even if you've 'done it before lots of times'. In fact, it's increasingly poor idea even as you gain more 'experience' that the rope you're using is 'much stronger than the spec sheet said'.

Why I'm objecting is that, whilst you have a lot of experience, 'Ethernet is much better than USB for Motion control' isn't a statement of your experience - you've extrapolated it into conjecture. You can tell it's not experience, because the existence of even one solid USB motion controller is sufficient to falsify your claim.

...plus, I've had enough 'memorable experiences' with Ethernet to know that Ethernet gear is plenty capable of being 'erratic and unpredictable'!

. . . . BUT . . . . We are moving away from OP problem so lets get it back on focus.

A little off but experience and theory and the spec sheet all seem to indicate that if you've got a USB motion controller and a laptop then that combination might be a source of problems. If the planned for desktop PC happens to be lying in a box somewhere, then it'll surely be worth trying with that; given it'll eliminate using a laptop as a factor.

A useful thing to do is to try to split the problem up; perhaps, it's possible to run the machine with the motor drivers disabled, so we're only exercising up to the motion control board? If that's reliable, then the laptop and inputs to the motion controller are less suspect, if it's unreliable, then we can look harder at the laptop and motion controller.

Other problems could be as simple as the connection to the board being poor - so it might be worth trying other USB ports. This is especially true for older computers where the ports may have got worn - especially with laptops which tend to get more connections and disconnections. Sometimes, a different cable might fix things; worth trying as it's easy; although IME cabling problems are fairly uncommon with USB -I've only had it on bus powered gear and extremely cheap cables). If you do have a cabling problem (or even a wonky device) then you might be able to find messages about usb connections/disconnections/resets/errors in the Windows Event Log.

Trying other ports could also help with a problem that occurs with few older computers is that some equipment has an expectation of USB 2.0, but the USB ports are USB 1.1 (or only one of the USB ports is USB 2.0).

[JAZZCNC]

The danger of relying on experience in this way that find yourself reaching conclusions that your experience does not - and cannot - support. To put it another way; lifting 6 tonnes on rope specified for a safe working load of 1 tonne is a bad idea, even if you've 'done it before lots of times'. In fact, it's increasingly poor idea even as you gain more 'experience' that the rope you're using is 'much stronger than the spec sheet said'.


Why I'm objecting is that, whilst you have a lot of experience, 'Ethernet is much better than USB for Motion control' isn't a statement of your experience - you've extrapolated it into conjecture. You can tell it's not experience, because the existence of even one solid USB motion controller is sufficient to falsify your claim.

Stop talking Bollocks your twisting things to fit.!! . . . . . Object all you like but end of the day we are talking MOTION CONTROL not Networking etc. I can honestly tell you with Hand on Heart I've never had a Machine stop or Dropout thru Ethernet failing on Any Motion control device I've used and I've used most of those in the Serious Hobby user market and some of the more expensive ones as well.
Like wise I've had many many experiences thru helping others and costly experiences of my own thru lost material and time with USB driven Motion control cards or devices dropping out on Both Box PC's and Laptops. So I must object because My experience is very wide so therefore Valid and very much relavant to CNC MOTION CONTROL.!! . . . . Which is what we are talking about here.!!

No more will say on this because as Said it detracts from OP problem.!!

[Jess]

I must object because My experience is very wide so therefore Valid and very much relavant to CNC MOTION CONTROL.!! . . . . Which is what we are talking about here.!!

Experience has limitations. You almost seem recognise this when you say 'we are talking MOTION CONTROL not Networking etc.' but in falling over yourself to claim limits to the experience of others, you're forgetting applying it to your own.

It's much derided, but theory is necessary to understand what the limitations of one's experience are; ie., the situations it's useful in, and the claims that that experience allows one to make. It allows one to look at what is going on, rather than be confused by a name. For instance, you can fixate on it being called 'motion control', or you can recognise it being called Ethernet is a pretty big hint - ie., that it's a networking technology for, well, networking things and that it doesn't much care if the device you've plugged in is a desktop running a web browser, a VoIP phone or a motion controller.

Simply put - having used some motion controls doesn't qualify you to say that USB is the problem. For instance, given that USB controls tend to be in the lower end of the market whilst Ethernet motion controls tend to be at least 'serious hobby' - so an alternative conclusion would be that USB motion controls tend to be cheap with performance to match - nothing to do with the cable you used to connect it to your PC.

This difference is important, because if USB is the problem, an ethernet motion controller for side-hobby money is a great deal; if the issue is instead that it's simply unreasonable to expect a decent motion controller to be cheap, then one should be wary of any motion controller for less than serious hobby money.

No more will say on this because as Said it detracts from OP problem.!!

The only reason why I'm even talking about Ethernet is that you brought it up. The only reason I'm still talking about it, is because I'm replying to you.

The majority of my post was, in fact on OP's problem. None of yours was.

To try to get it back on track again, if the USB motion controller is likely to be a problem, could it even be the root of the present issue - as in, something goes wrong (eg., command buffer empties during a run) and the controller goes into e-stop mode? (And that guard switch is sorted?)

[JAZZCNC]

Simply put - having used some motion controls doesn't qualify you to say that USB is the problem. For instance, given that USB controls tend to be in the lower end of the market whilst Ethernet motion controls tend to be at least 'serious hobby' - so an alternative conclusion would be that USB motion controls tend to be cheap with performance to match - nothing to do with the cable you used to connect it to your PC.

Ok you pushed me that bit more and you Sir trying to put down experience over theory and untested conclusions shows me you are Full of Shit.!! . . . I'll take my Real world experiences using Real world MOTION CONTROL CARDS of all shapes and prices in Real CNC enviroment using most connection types with MACH3 control software (Big Clue here) in a Buffered not realtime windows enviroment over your Theory and untested Conclusions anyday.! . . . Now I'll Jog on.

[Jess]

over your Theory and untested Conclusions anyday.!

You are the one who is making claims about the performance of motion controls you've not tested simply on the basis that you've used other motion controls that happen to have the same connector on them.

[JAZZCNC]

You are the one who is making claims about the performance of motion controls you've not tested simply on the basis that you've used other motion controls that happen to have the same connector on them.

FYI with exception of a few Cheap Chinese rubbish USB controllers not worth the effort I've tested and used all Motion control cards that work with Mach3. Because after all we are talking, or I am, about expereices using Motion control cards working with Mach3 and not about the Protocol used to do this. I don't give a toss if I used a piece of string to Talk to Machine in Klingon so long as it's stable and reliable in use. My Experience with USB and Mach3 hasn't been good or stable regardless of Device Supplier or PC.

Regards the Sieg USB device if that's what your refering to as the " Untested" then Chances are it's using one I've already tested or some variant.
I could with a quick phone call findout exactly what or who's device it uses as one of the main developers of the Sieg machines I know quite well.! . . . . Chances are he is the one doing the support in UK.!