Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: Sweep on May 24, 2010, 05:05:13 PM
-
After all the other niggling problems that I have encountered I now have a machine that is coupled and working with Mach3 and has a MPG controller. Great.
So, this weekend I thought I would do some axis tests to ensure that my programmed movements equated to the physical movements. Not great!
First of all my native and operating units are metric.
Using my Imperial (inch) DTI I positioned it on the X axis and moved the table left and right using MDI mode. The distance I moved was 12.7mm (1/2") to both keep the movement within the range of the DTI and to allow the DTI needle to fall on the same point of the scale.
Unfortunatley every time I moved from 0 to 12.7mm I was showing around 0.004" (0.1mm) short.
I have tried moving the table by hand and there is no play as witnessed by no movement of the DTI needle.
I have adjusted the axis acceleration down to single figures (5mm/s^2), have the velocity around 500mm/min, step and dir pulse as 5us but still i'm losing.
Tonight I double checked all the mechnicals and found 5mm pitch ballscrew (RM1605-UF), 2.5:1 timing belt drive (12x30T) and 400step/rev motors so I would expect that my steps per mm box is correct at 200.
I can solve the problem by 'tuning' the steps per mm to some spurious figure like 201.838 but that can't be correct as it has obviously been built to a round number and it should be 200, shouldn't it??
??? Sweep
-
Do you not use any microstepping in your drives?
Hood
-
Sorry Hood, what do you mean?
-
What make of stepper drives do you have?
Hood
-
The drives are Parker Digiplan SD2 units set to 400 steps/rev (half step) mode. I am just reading through the manual and cannot, as yet, see anything that relates to microstepping.
Earlier I did try setting the steps per mm on Mach3 to 100 and leaving the drives set at 400 steps/rev and the distance moved was have that indicated, but with missng distance, so I put it back to 200.
-
Just a thought but I wonder how the direction signal is interpreted if it is received at exactly the same time as a step signal. Will the drive change direction first then move, or move then change direction therefore posibly losing 2 steps?
-
Ok thats fine, that ties in, your steppers are normal 1.8 degree steppers (200 steps per rev) and you are using half stepping so that gives you 400 steps per motor rev.
Can you set the stepping in the drive to 2000, that would be 10microsteps, ie 200 motor steps multiplied by 10 microsteps = 2000.
If you can do that then set your steps per unit to 1000 and see how it goes.
Hood
-
As to your last post, refer to the manual and see whether your step/dir signals should be active low or high, also see if it says what the pulse width and Dir Prechange should be.
Hood
-
Fancy having a quick browse as I am unsure what I am looking for...
http://www.denfordata.com/bb/files/sd2_179.pdf
-
Seems to be saying active High and a pulse width of 10uS min so in Ports and Pins try active High for your Step and Dir signals. In Motor Tuning set the Pulse Width to 10uS and Also the Dir Prechange to 10uS
You can only set the drive to full or half stepping according to the manual so you will not be able to set to 10 microsteps like I earlier suggested.
Another thing I noted is it says 20KHz is the max pulse rate, although with half stepping that will likely be plenty.
Hood
-
Appreciate your assistance Hood.
I will try this out tomorrow night, but according to my manual, and what I recall on the screen, the range of step pulse is only 1 to 5us ?
-
You can set it as high as 15 if I recall, if that doesnt help you may need to try Sherline Mode.
Hood
-
No luck.
I have dropped the acceration right down, increased the step and direction pulse width from 5us to 20us, besides trying other settings in-between, I've tried changing from active high to active low on the step, but still i'm coming up short everytime by exactly the same amount. I even increased the acceleration to see if I could lose more distance but it performed the same.
I stripped off the X motor and counted the teeth on the timing pulleys and they are as expected 2.5:1 (12x30).
I measured the ballscrews with a rule and they are 5mm pitch and identified the ballscrew nuts as Warner RM1605 UF.
I have tried in Sherline mode (restating Mach3 as instructed) but nothing changes
???
-
What happens if you double or treble the distance?
Hood
-
I have only tried the measurements over 12.7mm as my Imperial DTI is limited to around 1" range. I could experiment with larger distances but would have no way to accuratley measure them. For example at 127mm I wouldn't be able to get 0.030" accuracy. I suppose I could fasten my digital caliper to the table and have a go with that.
-
ok, the results of tonight's futile experiments!
As mentioned earlier, I am pretty limited to the range of accurate measurement by the 1" capacity of my DTI.
Taking on board what Mr Hood said earlier and adapting it to my limited range I thought would smaller diatances result in smaller errors being recorded.
For anyone who hasn't picked it up, the reason for all the odd figures like 12.7 and 2.54 etc is because i'm operating in mm but have an imperial (inch) DTI.
So starting with the DTI zeroed I drove the table G0 X12.7 (1/2") and the result was 4thou short.
Then driving the table G0 X0 the DTI returned to zero, therefore the error is apparent in both directions.
Next I thought what about "electronic" backlash so I first drove the table from zero to G0 X-2.54 and the same 4 thou error appeared. Driving the table back to G0 X0 returned it to zero on the DTI.
Now the intresting bit. When I now drove G0 X12.7 the needle hit the 1/2" mark right in the middle. Brilliant.
When I drove the table back, i.e. introduced a direction change, to zero the 4thou error reappeared.
Next, because i had just sorted my nice big fat pro MPG I set it to single step mode with a step of 0.1mm (0.0039") and did the same experiment. This gave exactly the same results; take up the electronic backlash and the DTI would read true, change direction and i'd lose some distance, well if I moved 0.1mm (0.0039") whilst changing direction I would only move half that distance.
Now I know that i'm only losing, say, 2thou on the shortest of distances and twice that on long distances but I think we can see the problem.
Question is, is there a cure?
...Sweep
-
oh, by the way, pushing the acceleration right up to 500mm/s^2, dropping the step pulse down to 5us still disn't cause a problem if i'd already taken out the 'electronic' backlash.
I was just playing with these silly high accelerations just to see what would happen and have returned them to a sensible figure.
-
Ok, make sure you move in one direction, zero the DRO and also zero your clock (DTI) and then still going the same direction move your 12.7mm.
What do you get in both DRO and on the Clock?
Next do a move back to zero, what do both say now?
Hood
-
If I zero the DRO and DTI after a G0 X12.7 move, then do a G0 X-12.7 move the DRO will read 0.00 and the DTI will read -0.4996" (0.004" adrift by eye but probably will be 0.00039 due to my metric machine).
...Sweep
-
What you need to do is keep moving in the same direction for this test. Move in a positive direction, zero the DRO and zero the DTI then move x12.7 thus still going positive.
Hood
-
If I keep going positive I don't accumulate errors, the DTI will stay at 0.004" adrift. If I reset the DTI to "zero" after the first positive move, then any subsequent positive moves will not add to te error. The error is only introduced when I change direction.
-
It sounds like you have mechanical backlash. This could be in the ballscrew itself but can also be in the angular contact bearings or even loose gibs can give your table a twist thus giving you an error.
If you can put the DTI on the end of the table and have the poiter on the end of the ballscrew shaft then step slowly one way then reverse in small steps and see if the DTI needle moves, if it does then you have float in the screw which means either flex in the brackets or bearings need adjustment.
Hood
-
I am pretty sure that there is no mechanical backlash as have already pulled the table too and fro (well applied pressure in both directions) and the DTI needle never flinched. I think an experiment on the Z axis would prove or dispell your toughts as the weight of the whole spindle assembly will naturally remove any mechanical backlash. If there is 'electronic' backlash the error will still appear.
-
Sorry,to fully answer your last post, i'll try move the axis by my smallest resolvable increment (200steps/mm) of 0.05mm and see what movement of the axis is measured on the DTI
-
The error is only introduced when I change direction.
The only two reasons I can think of for this are lost step pulses on dir change or backlash. The former is usually a single step pulse lost but for your res and error you'd have to be losing around 20 pulses! That can't be right... Don't know about the rest of your system but do belts really have no backlash?
Cheers
Ian
-
ah, looks like I wasn't paying attention at school during maths.
1 / 5mm/rev = 0.2rev/mm of ballscrew shaft
0.2rev/mm x 2.5 belt ratio = 0.5rev/mm of motor spindle
0.5rev/mm x 400step/rev = 200steps/mm of motor spindle
therefore 1 step = 1 / 200 = 0.005mm/step or 0.00019"
I would, therefore, agree that 20 steps would need to get forgotton about to realise such an error. I'll have to have a closer look at mechanical backlash as I would agree that this now seems like the obvious problem.
thanks for the lesson Sterling
...Sweep