[Willyb]

Hi Guys

When mounting the Encoder, is there anything to be gained by installing them on the Ball Screws verses having them on the Motors? My Encoders are presently mounted on the end of the Servo Motors shafts (the way they were when I purchased them) and I was wondering if it would be worth moving them to the Ball Screws? I am using a Belt drive setup with 2:1 Ratio and a 3/8” wide toothed Belt. Didn’t know if these belts stretched at all? Interested to hear your opinions.

Best regards
Bill

[balsaman]

Yes you can but why?

Eric

[sol]

This was ripped unashamedly and directly from Industrial Hobbies site. I think he won't mind the free advertising.

Why is the encoder mounted on the ballscrew?

The long and short of it is accuracy. Mounted on the end of the ballscrew the encoder gets the most accurate reading possible. Simple things such as belt slack (which changes over time) or crud getting between the belt and drive pulleys (remember this is a harsh environment) don't effect accuracy.

[balsaman]

This is true but if there is any backlash in the system (like belt slack), a leadscrew mounted encoder will make the motor hunt, sometimes violently, as it tries to position the screw properly.

When there is no backlash between the motor and screw it works great, and I have done it.

Eric

[ESjaavik]

The encoder should follow the motor shaft as precisely as possible. If mounted on the motor shaft it will be close to perfect. If on the screw, there may be differences because of slack and other factors. The maths in the servo drive cannot take this into account, and you get hunting, noises and other disturbances. If you calm down the motor with the servo drive settings, it will not respond very well. But you may save money by not being restricted to servo motors. The cheaper motors again having less suitable characteristics. So this may be just another example of making a cost driven choice look good in the sales blurb.

Since it seems you have a working setup just keep it.
The belt stretch is insignificant. The motor mount (rigidity) and belt / wheel wear are more important factors.

[Torsten]

On a 2:1 ratio timming belt setup mounting the encoders on the leadscrew would
mean your resolution is only half as seen by the motor.
To maintain your current configuration you would need encoders with twice the
resolution you have now.
The higher the ratios get the more problems you are gone encounter with this because
the uncontrolled movements between readings become larger.
Timming belt setups do have backlash they are just hiddden by the tension of the
belt unlike a gear where any play can be easyly felt and messured.
Other draw backs of this setup are the fact that you will not be able to
controll this motor when it is removed from the drive arrangement.
Don't you love the kinds of problems where your best approch is doing nothing.:-)
Good Luck

[Al_The_Man]

Here is some recent interesting quotes from some of the leading Encoder & Machine tool Manuf.

"HEIDENHAIN Linear Encoders
As the present trend of machine tools evolves toward increasingly higher accuracy and resolution, increased reliability and speeds, and more efficient working ranges, so too must feedback systems. Currently, linear feedback systems are available that will achieve resolutions in the submicron range.
Submicron resolutions, for example, are required in the semiconductor industry and in ultra-precision machining. Achieving these resolutions is possible with the use of linear scales which transmit displacement information directly to a digital readout, NC controller, or other peripheral device for display or evaluation.
As in rotary, linear scales operate on the same photoelectric scanning principle, but the linear scales are comprised in an overall straight construction, and their output signals are interpolated or digitized differently in a direct manner. One of these signals is always used by the accompanying digital readout or numerical control to determine and establish home position on the linear machine axis in case of a power interruption or for workpiece referencing. "
.....Rick Korte President Heidenhain USA

ANILAM CONTROLS : "Glass precision linear encoders are the most accurate way to cost-effectively measure distance/position in a straight line", comments Managing Director Phil Goulding.
"That's why around 90% of machine shops globally - and most OEMs - use this type of optical encoder rather than inductive types".
Anilam was the first company to produce a programmable DRO (a product which took the competition four years to emulate) and one of the first to use contaminant-protected glass scales over 30 years ago.
"There's no argument that glass encoders are more accurate", continues Goulding, "because they have 1270 points of measurement graduations per inch compared with the four points per inch that characterise many other alternative (transducer) types".
Accuracy is, of course, the key denominator when comparing encoders, he says.
"Anilam ensures accuracy by adding the hard chrome graduations - which occur at every 20 microns step - to the glass in a vacuum, then the scale is mounted in a one-piece extrusion using elastic adhesive.
Coupled with the fact that the roller bearing-guided reader heads consistently maintain their position relative to the grating, and the result is predictable and repeatable measuring to +/-5um/m".
This compares, he says, with encoders that are constructed of other materials that, conservatively, even if they can be produced to within 1um of each other means that their inaccuracy over a 1m length could be as much as 80um.
The longevity of glass scales and their suitability for working in the harshest metal-cutting environments are other myths that Anilam will be dispelling at MACH, where it will be showing examples from a comprehensive range of precision glass scale linear encoders having resolutions from 5 to 0.5um, to suit the application. ANILAM Engineering"

"CINCINATTI MACHINE TOOL: The change from rotary to linear digital scale encoder positioning technology on the Cincinnati range of VMC vertical machining centres and Hawk TC CNC lathes has improved positioning accuracies by almost 60 per cent.
Not only has this development led to highly consistent machining, but it has also enabled Cincinnati Machine UK to lay claim that its highly competitive price/performance ratio machines are now in a class of their own in world markets.
The linear scales form part of a control and drives package supplied by Heidenhain (GB) of Burgess Hill, that creates a closed-loop positioning system.
The Heidenhain package is also connected into electronic thermal compensation that monitors changes in the machine structure and provides instant feed back of relative data to the machines' Heidenhain TNC 426 and TNC 410 contouring control systems.
According to Pete Robbins, Sales Product Engineer at Cincinnati Machine's Birmingham build operations: "The fitting of linear scales to Hawk CNC lathes was a milestone.
It made us the first major machine tool builder to provide such high levels of cylindrical turning accuracy on a volume built production lathe.
This gave us an immediate competitive edge throughout the world, from the cold of Alaska to the heat of Arizona."
The advantage of linear scale technology is that it accommodates most mechanical transmission errors in the drive system of the machine tool, the effects of temperature changes, friction and any variation in slide positioning.
Rotary encoders are less expensive but have limitations because they are mounted with the motor and are detached from the cutting zone.
While rotary encoders are perfectly acceptable for most turning operations, they will only provide indirect control of the linear position of the slide, relative to the position of the rotor in the servo motor.
However, as axis speeds rise, problems arise from this arrangement which cannot account for any increased heat generation in the ballscrew and effects from thermal distortion that will occur" http://www.auto-met.com/heidenhain/n...ncoders_in.htm
Al

[HuFlungDung]

No doubt that the linear scale is better if there is negligible motor to screw backlash. Some of the fancier systems out there have dual feedback, one device on the motor, coupled with another one on a scale, I suppose. I don't know what logic they'd use to interpret the two feedback signals, but it should be pretty awesome performance.

[ViperTX]

Hu,
I believe the opposite is true....the linear scales will always indicate the true position (based on the accuracy of the scale) regardless of the amount of backlash.....the problem generally encountered is when you approach a position and need to back up a bit based on your error signal and you have a large amount of backlash. Anyway someone on one of the forums actually mounted their encoders on the ballscrew....I wonder what their experiences are....

[HuFlungDung]

Viper, I believe that true position is always the goal. Who cares if the motor performance is superb but the position is no good? That's the point of the dual feedback system: any system with backlash never moves far enough, so the encoder on the motor provides the PID feedback components, but the linear scale provides the error signal.

[trubleshtr]

We have a new machine at work that uses two feedback systems as you are describing, the machine just crashed yesterday, instead of milling a pocket it punched a hole straight into the fixture, After all the head scratching, the linear encoder was unplugged and we let the system run on only the motor encoder, for some reason the linear position felt it came up short and allowed the axis to compensate.....right into the fixture. I'm still not convinced 2 is better than one, just my 2 cents.........

[HuFlungDung]

Yikes! That's unfortunate. Too bad they didn't have internal logic to compare the two position signals and halt the machine if a certain limit was exceeded....unless one just up and died, but that should be caught, too.

[Bubba]

ViperTX,
When I first attempted to convert my mill/drill to cnc, I tried to use my existing DRO system (Rotary encoders driven by a cable system) that read the table directly. BAD NEWS (I had NOT converted to ballscrews at that time) and the backlash made for an unstable system.
I moved the encoders from the "table movement" to mounting them on the end of the lead screws and even with the terrible backlash that I had, I was very happy with he results. Therefore, when I converted to ball screws, I continued the same setup.
I use XL037 belts and pulleys for the reduction from the servos and feel the system is very stable.
The biggest thing you have to be careful of is that the leadscrew does NOT have any axial movement or you can mess up the encoder and of course your accuracy/resolution will go on vacation. (Don't ask how I know:{

Bubba

[ESjaavik]

@Al TM: What you're implying here is valid, but maybe not in amateur use. Hu just gave us the reason why. The point he lost is maybe that to us this is mostly theory because we don't have the software available to exploit this technology.

What we use is previous generation servo drives, cheap or open source control systems and bits from the junk bins of the world. What I'm trying to say here is that you're right, but as I see it, it does not apply to the thread-starter (Willyb) and most of us on this site. I drive an Alfa Romeo, but I would really like a Ferrari, I just can't afford it!

By all means let's keep on pointing to the current state of the art, but let's make it clear that today, it's beyond our budgets. That may change, as if there are any whiz-kids here that are ready to put as much effort into the (CNC) open software community as some do into virus development, we may get the opportunity to use these technologies soon. And we old farts will be following as best we can.

On a more serious note, I think that if applied as a correction outside the Current/Speed/Position loops of a "previous generation" drive, it would work fine. But there is no control software that I'm aware of for the amateur that have this possibility.

[trubleshtr]

Hung, We are waiting for the machine builders to arrive (from Itally) I think the original programmer made a "type-o" in the set-up, remember 2 negatives make a positive in math when multiplied? I got a funny feeling that instead of the machine reversing, it got the math wrong (due to programmer set-up) and went positive in travel.......The machine is still under warrenty, so for me this is a front row seat to "how not to program" a dual postioning system............now if I can only find my popcorn.

[ViperTX]

Hung, We are waiting for the machine builders to arrive (from Itally) I think the original programmer made a "type-o" in the set-up, remember 2 negatives make a positive in math when multiplied? I got a funny feeling that instead of the machine reversing, it got the math wrong (due to programmer set-up) and went positive in travel.......The machine is still under warrenty, so for me this is a front row seat to "how not to program" a dual postioning system............now if I can only find my popcorn.

Seems that you should have some mechanism to view the tool path of a part before it's machined and the software should know that the boundaries of the part have been crossed and indicate in bright red where there is a problem.