View Full Version : DRO equipped Mill retrofit
Paraprop 02-14-2005, 10:07 AM I have just retrofited my Mill with linear scales with 5 micron meter resolution on all 3 axes and a DRO unit which has a serial port able to output the X,Y, & Z position.
If I want tu retrofit it further to become a CNC mill by puting servo motors on each axes, do you think that I can use the DRO position data as feedback for the controller knowing that the screws of all 3 axes have quite a bit of backlash ?
What servo controller type would you use for such an application ?
As it is quite a big mill and the Z axis takes quite some power to move upwards, what is the best method to size the servo's ?
Which PC CNC software could handle such an application ?
Al_The_Man 02-14-2005, 11:29 AM I would say you would have to use the scales direct to the control due to any delay in relaying position through the serial port, this is usually meant to record position data, but without concern of doing it in real time.
If you use the scales direct to the control, and have alot of backlash, then you would probabally need a type of controller that will allow 2 encoder inputs/axis, one on the motor and the other the scale.
Al
ViperTX 02-14-2005, 01:28 PM I believe most of the step and direction controls are open loop, the only feedback is an error signal that the count was not reached...or did I miss something?
Personally with alot of backlash....I think you'll constantly be chasing yourself....I would eliminate as much of the backlash as you can before you CNC it.
Al_The_Man 02-14-2005, 03:16 PM I believe most of the step and direction controls are open loop, the only feedback is an error signal that the count was not reached...or did I miss something?
He was asking what type of controller he would need, I was just pointing out that using scales with backlash in the system, he would need the more sophisticated type of controller this would include the software that uses the high-end PC based motion cards or the commercial type controller where they accept 2 encoders per axis in order to be able to use a motor encoder and a linear scale on the final motion.
Right now he just has a DRO
Al
ViperTX 02-14-2005, 05:52 PM Paraprop,
1st things to do. Reduce the backlash (new ballnuts, ballscrews or ACME screws, or Moglice).
2nd - Find the interface schematic for your DRO's, are they differential quadature or absolute encoder types.
3rd - Determine what you need for servos, you could estimate what you need by measuring the torque required to turn each axis and use that to compare to any servo specifications or you could just select 800 oz-in servos.
4th - Now that you know your servo requirements, your DRO interface, and have reduced your backlash, now look at the controllers from Gecko, Rutex at the low end, and AhHa, Galil, etc. at the mid to high end.
Take the steps, don't overwhelm yourself.
Paraprop 02-15-2005, 07:32 AM Hi Al & ViperTx,
Thanks for your precious input,
It is going to take a while to get the screw's backlash sorted out.
The linear feedback scales are the Model MKT from Fagor, accuracy +-0.0004 inch, Resolution 0.0002 inch. The output signal is TTL the period of the output signal is 0.00002 m.
I am not sure if this data will allow you to identify a servo / stepper controller which can make use of this signal.
Evodyne 02-15-2005, 09:21 AM Paraprop,
Your scales have essentially three outputs: A, A', and Ref. The A and A' are square wave quadrature outputs that pulse once every 20 uM. Your Ref pulses once every 50 mm for your scales.
Being TTL (Transistor-Transistor Logic), the output logic levels are defined by as follows:
Logic 0 0.0 to 0.4 VDC
Logic 1 2.4 to 5.0 VDC
This can be important-a device with TTL inputs is guarenteed to work with these voltage levels, but other types (such as CMOS) work with the extremes, i.e. 0 VDC is logic 0, 5 VDC (or higher) is a logic 1. Thus a valid TTL logic 1 of 3.0 vdc might not be seen as such by a MOS type input. But it would be seen by TTL inputs.
So look for stuff with TTL inputs.
Also, most TTL devices have a fanout greater than one. What is fanout? Example: a thingy with a TTL output having a fanout of ten can have that single output wired in parallel to ten devices at once and still work. All ten devices will read the output simultaneously.
For you, this means you can drive your DRO and a servo driver both using the raw outputs from your encoders. Make sense?
The quadrature outputs are just like those on the rotary encoders guys attach to their servo motors. So your slide is basically interchangeable with a rotary encoder from an electrical point of view.
I have to agree with the other contributors that slop is going to kill you. A well-tuned servo loop is going to hunt or oscillate. You can detune the loop, but then it will have a wider deadband, poor positional accuracy, and sluggishness. You won't be happy with it either way.
What kind of screws are you running? In almost any case you should be able to add a second nut and tension it against the first to eliminate the bulk of the backlash.
Good luck!
Paraprop 02-15-2005, 09:43 AM Evodyne,
looks like you are right.
Further specs. say:
Supply voltage 5V +- 5%, 100 mA
Output signals : 2 pulse trains A & B shifted 90 Deg.
Marker pulse Io Synchronized with A & B signals
Voh.>= 3.5V
Isource <= 4mA
Vol <= 0.4V
Isink <= 4 mA
IO pulse period T/4 (T= 20um)
However for the backlash, is there not a way to get around it ?
The reason is that backlash will only happen when we change the direction of rotation. Is there no software method to get around this ?
Paraprop 02-15-2005, 10:18 AM How would you proceed to measure the required torque to, for example, wind upwards the Z axis, knowing that on my mill the hand wheel is arranged horizontaly ?
Please refer to the 2 enclosed pictures of the mill taken before the DRO was put on it.
HuFlungDung 02-15-2005, 10:19 AM Paraprop,
There are plenty of circumstances when the cutter will actually pull the work across the empty "backlash space" when climb milling the work. If you are cnc machining, you will want to climb mill, guaranteed. This means that the control is going to have to try to adjust for this sudden error in the commanded position. The cutter may also get a giant mouthful that was not anticipated and break the tool, or throw the work out of position/the vise.
Even though you can adjust for lost motion in software, it is not truly eliminating the mechanical slop. Trust us, you'll hate it if you don't fix it.
Paraprop 02-15-2005, 10:39 AM HuFlungDung,
Your point makes sense to me.
The machine is close to 40 years old.
The screw used on my machine is not the ball type. The thread is straight square. How do you call this type ?
What should I do with it to cancel the backlash / slop ?
Sorry the picture where not attached with my earlier mail. Here they are.
ViperTX 02-15-2005, 10:52 AM HuFlungDung,
Your point makes sense to me.
The machine is close to 40 years old.
The screw used on my machine is not the ball type. The thread is straight square. How do you call this type ?
What should I do with it to cancel the backlash / slop ?
Sorry the picture where not attached with my earlier mail. Here they are.
Short of replacing the ACME screw which will probably be expensive, you should read about moglice. There is a good article in Home Shop Machinist about it, don't know if you have access to it. If not then go to the http://www.moglice.com website and read about. I've been skeptical about it's use, but many journeymen machinist and tool rebuilders use it. :cheers:
ViperTX 02-15-2005, 11:01 AM Paraprop,
You want to measure the required torque at the point where the servo will be connected. Use a torque wrench, pay attention to the torque required to get the axis moving and the torque required to keep the axis moving...there should be some variation. Based on your torque readings you can then determine if you can directly couple your servo to the axis or whether you'll need to add some mechanical advantage (using timing belt pulleys, or gears) depending on what you have access to.
Evodyne 02-15-2005, 11:02 AM Roton has a nice page that will help you identify all the properties of your screw. Look here. (http://www.roton.com/web/identifying.jsp)
I guess, in order of effort/cost, your choices are to:
1. Rebuild/replace the worn nut you have now with a "tighter" unit.
2. Add a second nut and tension it against the first.
3. Replace both your screw and nut with a better type-if I were going to go this far I would take the plunge and get machined screws and ball nuts (NOT a rolled screw/ball nut). A little patient searching on ebay can turn up some deals.
Evodyne 02-15-2005, 11:05 AM This stuff looks pretty interesting. I hadn't heard of moglice until just now. I took a look at their page: cool. Thanks ViperTX!
|
|