[glbreil]

Hello, looking at these Leadshine Ethercat motors (EML2H-0750LA80F) they have a 23 bit optical encoder and I am wondering what that means in terms of repeatable positioning accuracy. I understand imperial measurements .001, 0001, 00001 etc. So without considering any sort of backlash in the system what sort of repeatable positioning accuracy does a 23 bit optical encoder on these motors give you?

Thanks in advance! Gary

[joeavaerage]

Hi,
theoretically a 23 bit encoder has an angular resolution of 83,388,608 counts per revolution, or 0.0155 seconds of arc per step.
This is incredibly fine resolution. In truth the linearity of the encoder means in practice the genuine or achievable resolution is very VERY much less
than this, down to maybe 100,000 counts per rev or 13 arc seconds per step. Even this resolution is very fine indeed.

Its probably fair to say that all modern AC servos are near enough to 'exact' with respect to their angular position.

The question of accuracy and repeatable accuracy is determined more by the linear motion components than the servo. You will no doubt have seen and read about the
different grades of balllscrews, with lowly C7's (50um/300mm), being very common amongst hobbyists on the basis of cost and availability. The better grades known as 'ground' ballscrews
are C5's (18um/300mm) and C3's (8um/300mm) are used in quality industrial CNC machines and cost ten and more times that of a C7.

There is one specification of ballscrew accuracy that receives very little attention but is very much tied to your question. It is called 'cyclic error'. Lets say you have a 10mm pitch ballscrew.
You rotate it precisely one turn and measure its movement and find that it moved 10.0000mm, so you might say its perfect....but it may not have moved linearly during that rotation.
At a half turn it had moved 5.05mm while in the last half it moved 4.95mm, for a total of 10.000mm.

While a C7 screw has a max error of 50um/300mm, and to be fair is actually very good and many fantastic machines have been made with them, the cyclic error is uncontrolled and not specified.
It has been measured on selected screws and be shown to be as high as 35um/360 degrees. This is not so good. Using the same idea above, lets imagine you had a 10mm pitch C7
which is near perfect, ie 10mm per revolution but it has 35um cyclic error. At one half tun it will be out by 0.035mm, even if it catches back up by the end of the turn to 10mm.

C5's and C3's the maximum cyclic error is specified and measured, with C5's it is 8um per 360 degrees and C3's its 6um per 360 degrees.

So even with a servo that is 'perfect' a machine with a C7 screw could exhibit something like 35um of error over 10mm while a C5 equipped machine will exhibit 8um, and a C3 machine 6um.
Clearly for industrial machining purpose the better C5's and even better C3's are highly desired....but the cost is shocking!!!!!

I use Delta B2 series servos. They have an encoder of 160,000count per rev. I cannot use anything like that resolution, I would have to signal the drive Steps at the rate of
many MHz, ie radio singals!!! to get full speed. Using the electronic gearing parameters I have programmed my drive to rotate one revolution with 5000 Steps. The servos are direct connected
to the 5mm C5 ballscrews, meaning that each Step is 1um. This is good and certainly adequte resolution. The cyclic error of the ballscrew, even at 8um/rev is likely larger than the resolution.
I have come to regard the practical and useful resolution of my machine is 5um per Step or 0.005mm. Realistically the servo must contribute some error but it is very small by comparison to
other sources of error, particularly cyclic error and backlash if you have any. Even these errors are small and so changing temperatures is likely to have an even bigger effect.

To answer your question, I would say that any well-made and programed AC servo should be considered as close to perfect as you are ever likely to get, so the actual resolution, accuracy
and repeat accuracy of a machine is going to be determined by the ballscrew and/or the thermal expansion/contraction of the machine itself.

Craig

[glbreil]

Thanks for the information, it was an interesting read. The machine is an older Hardinge CHNC II Super Precision or so it says, so I assume the screws are fairly high grade, but you are correct that I don’t have it in a precision controlled environment and I am not doing any particularly precision work. I just didn’t quite know how to relate the encoder bits to encoder lines and how that related to accuracy. Sounds like these are more than I would need.

Thanks much Gary

[joeavaerage]

Hi,

Sounds like these are more than I would need.

Yes, I think that any modern AC servo is way more accurate than any machine that you or I could afford to put them in.

I would suggest before you commit to buy, find out if there is setup and tuning software. Better still download and inspect the software. If you've not set up servos before
you'll want/need the software. There are hundreds of parameters in a modern servo, many may be left at default, but others cannot. Its a mine field without good software.

Craig

[automationtechinc]

Yes
We have EtherCAT Servo Motor Kit, 750W 220VAC in stock from Leadshine in Streamwood, IL

https://www.automationtechnologiesin...it-750w-110vac

or this one

https://www.automationtechnologiesin...t-1000w-220vac

[joeavaerage]

Hi,
do either model have a full suite set-up and tuning software package?

Craig

[automationtechinc]

Yes
We can email you the Motion Studio to test or tune for both 750W and 1000W

[glbreil]

Thanks, when I used the links I cannot tell whether the motors are ELM1 or ELM2. Centroid recommended ELM2H-0750MA0E with EL7 drivers is that what I am viewing on the 750watt link?

Thanks Gary

[joeavaerage]

Hi,
is there a particular reason that you are going for Ethercat?

I was not aware that Centroid even had an Ethercat solution. Ethercat servos attract a premium of somewhere between $50 an $150 per servo.

Craig

- - - Updated - - -

Hi,
is there a particular reason that you are going for Ethercat?

I was not aware that Centroid even had an Ethercat solution. Ethercat servos attract a premium of somewhere between $50 an $150 per servo.

Craig

[glbreil]

Centroid recommended their hickory controller with the ethercat motors and drives for my application. After looking over the whole system,it seems like a simple solution, and I don’t really see where the cost is any higher but I haven’t made a final decision yet. Thanks, Gary.

[joeavaerage]

Hi,
the Centroid Hickory is $1400 (plus software upgrade) whereas the Centroid Acorn is $330 (plus software upgrade).
Make no mistake the Hickory is $1000 more. The servos are likely $50 to $150 more each over their 'plainer' Step/Dir siblings.

Don't get me wrong I very much looked at a Kingstar/Interval Zero/Mach4 Ethercat solution for my new build mill. A cost comparison between the Ethercat solution
and the more regular Mach4/ESS/BoB solution was about $600. Added to that the servos (Delta B2-E) are $50-$75 more expensive than the non-Ethercat model.

I would presume you have a reason that you would pay a very distinct premium for Ethercat. That the vendor recommended it is hardly a surprise, they are in the business of selling after all.

Second issue that disinclines me to the Hickory, over and above the cost.

Ethrcat is a realtime communication protocol, but a Windows PC is not realtime. Thus the Hickory is still a buffered motion control, just with some fancy smancy way to communicate with
the servos, it still no more makes the PC realtime. When all said and done its just another fancy and expensive motion control board.

The Kingstar solution uses a PC but one of the cores of a multicore PC is loaded with an Interval Zero realtime scheduler. You might say the PC is now host to a realtime section,
very clever. Then the Kigstar software enacts an Ethercat master and communicates by realtime Ethercat via the realtime core. So with this solution YOU DO NOT HAVE A SEPARATE BOARD
AT ALL. That is no motion board, all the servos and IO nodes are daisy chained together. IO, things like limit switches, home switches, probes, and anything else are likewise just Ethercat modules,
which get plugged in to the daisy chain. You might have a Beckhoff Ethercat IO module with eight digital IOs, but lets say you need more and you want an analogue input as well, then you buy another Beckhoff
Ethercat IO module with say another four digital inputs and two analogue inputs. I don't mind telling you that Beckhoff modules are not cheap....so you're not gpoing to buy lots of them, but you buy what you need.
This means that not only are your servos Ethercat but your entire machine IO is Ethercat as well.

The Kingstar solution is $1600USD, it includes a re-furbished PC, with a top grade Ethernet card, has a Kingstar runtime license, an Interval Zero runtime license and Mach4Hobby license.
The Hickory is $1400 plus software, say $1000 and you still have to supply your own PC.....plus you IO is not Ethercat. No, I'm sorry to say I think the Hickory is a rip.

Craig

[joeavaerage]

Hi,
I just checked the current price of the Kingstar/Interval Zero/Mach4 solution, and the price has gone up to $1800.

https://www.automationtechnologiesin...ntrol-computer

Notice there is no hardware board. There are solutions offer on the same site that do have Ethercat boards, and they are rather like the Hickory. Quite frankly don't mess around with these boards, they
are like a 'Mister in-between'. You either get a traditional Step/Dir motion control board (cheapest and tried and trued) OR your get a genuine Ethercat solution, ie no motion board whatever.

This combination is the same core setup, namely Kingstar/Interval Zero/Mach4 but also has three Leadshine Ethercat servos for $3350:

https://www.automationtechnologiesin...h4-750w-3-axis

This includes a rebuilt Dell PC. Did you note in the video you linked to that Kingstar are not inclined to support such a rebuilt PC, but will support their own build PC, presumably a high reliability industrial PC.
That combination costs more ($4980):

https://www.automationtechnologiesin...h4-750w-3-axis

If you want Ethercat...you can certainly have it, but a genuine Ethercat solution is not cheap, even if it is vastly cheaper than it was even only five years ago. I

Craig

[joeavaerage]

Hi,
given that Automation Tech are listening to this thread, perhaps they would answer:

The solution that I looked at and liked was the rebuilt Dell PC, but the video suggested that Kingstar would not support a re-built PC , but would support a new industrial PC of their
spec?

Do you guys have any experience of this?

My concern is that were I to step up to Ethercat, and I'd like to, and I would not be buggering around with any motion board, it would be Kingstar/Interval Zero/Mach4, that were I to find
that Kingstar could not or would not help me should I encounter a problem....that is a risk I'm just not prepared to take. Thus if support is offered BUT ONLY for a new PC....then
the true cost of the purchase MUST include that new PC.

I have been using Mach4Hobby and an ESS with my own designed and built breakout board for 8.5 years. When I considered the Automation Tech solution it rather meant I had to buy another PC
plus the software licenses. Given that I already have a PC and a Mach4Hobby license it rather seemed like I was having to re-buy what I already had. Having said that I would not be at all confident
that my little PC would run Kingstar...so would perforce have to update it.....and then pay for shipping to New Zealand. As much as I wanted to go to Ethecat the costs were just too steep....at that time.
Its not really an upgrade from where I am at, but rather a new purchase. Shame, I would dearly like to do so.

Craig

[glbreil]

In its day the CHNC II+ was a very accurate and precision made machine. It had the best of equipment used and the machine is still in great shape, but the electronics are way outdated and I have never liked working with the GN10T control. On the other hand I have Mill that was factory fitted with a M400 Centroid control and I have always found it easy to use and reliable. The few times I have had issues help was readily available and I was able to get parts and fix it myself.

Staying with the Centroid Software is appealing because it is what I already know and have a little understanding of. Originally I intended to go with the Allin1DC and stay with the original DC servos because I have extras, but I found out they are to small to meet the minimum amperage requirements. Then I was recommended the Hickory and I can see the a real ease of installation, as well as it has 32 inputs and outputs without adding additional boards. The CHNC II has all kinds of stuff that will require I/O use. Spindle brake, Collet closer, Hydraulic cut off tool, parts catcher, way lube, coolant pump, and on and on, not counting the funky tool turret that will require several.

The idea of learning a new control other than Centroid after I get this done is not appealing at all to me, and the only other thing is that the most of the cheaper boards use step/direction and all you read is controversy with folks going back and forth arguing over whether they are suitable for anything other than a hobby machine.

Granted all of my machining is hobby, but it isn't really a hobby machine. I wish I had the knowledge and understanding and want too, to dive into something totally new, but really this is just a means to the end of me being able to make parts easier and simpler when I want to.

With that said, if I could do it and save some money, I am all about that. I really enjoy what you are writing and sending and I have learned a lot already. Please send more if you have additional thoughts or ideas, but I think I would like to stay on the Centroid page if possible.

Thanks Much! Gary

[joeavaerage]

Hi,

most of the cheaper boards use step/direction and all you read is controversy with folks going back and forth arguing over whether they are suitable for anything other than a hobby machine.

That is complete rubbish. A well sorted Step/Dir control system is performance limited by the machine, not the control system My machine running Mach4 is Step/Dir and is every bit as good as any other,
resolution to 1um and yet still do 5000rpm or 25m/min rapids. If there is a limit its the mechanics of my machine not the control.

A well sorted Ethercat controlled machine is performance limited by the machine, not the control system.

Where Ethercat does shine is that it can have 100 nodes, thus you could have several machines with conveyors and part manipulators between each and all be controlled by one PC, and without huge bundles of cables
throughout the installation. Ethercat is superb for industrial automation, its advantages evaporate when talking a single machine.

Staying with the Centroid Software is appealing because it is what I already know and have a little understanding of

That is understandable. The Centroid Acorn is a four axis Step/Dir controller and uses identical software to the Hickory. I imagine you'd want the Pro or Ultimate version of the software, but that cost is the same for
both the Acorn and the Hickory. The Acorn is a little over $1000 cheaper. Would a good four axis control run your machine? Remember also that Ethercat capable servos attract $50 to $150 premium over
the plainer Step/Dir models from the same manufacturer.

The savings of $1000 for the motion control (Acorn vs Hickory) and the $300(est) saving on servos is worth thinking about. If you do some research I'm sure you'll find many an Acorn user reporting
superb results.

If you are of the opinion that your machine is deserving of a 'better' control system then I do not believe the Hickory is the best choice. Look at Masso. Look at Kingstar/Interval Zero/Mach4. Look at
Linux CNC (Ethercat). Look at Sinumeric808. There are better choices (than Hickory), although all of them do use a sightly different software.....but that is a small price to pay verses being ripped by Centroid.

Craig

[joeavaerage]

Hi,
this is the sort of thing I mean about IO modules:

Ok, it is a cheap one, about $100USD. Just plug it in the daisy chain along with your servos and it gives you 16 digital inputs and 16 digital outputs. Does not get much simpler than that. If you need more IO just plug in
another block in the daisy chain.

Craig