Machsupport Forum
Third party software and hardware support forums. => CS-Lab => Topic started by: slovenec on April 09, 2016, 09:03:52 AM
-
Greetings,
i would need some help regarding the CSMIO/IP-A. The problem is in the PID regulation. When i try to regulate the axis i do not recive any signal from the linear encoder on the graph . Is this normal? Linear encoders are Heidenhaim ls 903. I have checked the connection and every thing is connected right. Attached you will find data sheet of the encoders and my screen of the problem. The encoders worked with the old controller without problems.
Thank you in advance.
https://www.dropbox.com/sh/5zyu75x5e756ku1/AADZO8KOkRyX_a6qm7OCSv-pa?dl=0 (https://www.dropbox.com/sh/5zyu75x5e756ku1/AADZO8KOkRyX_a6qm7OCSv-pa?dl=0)
-
Are your scales 5v square wave?
Often Heidenhain Encoders and Scales are sine wave 1v peak to peak, if so then they are no use unless you also have an additional interface box that I have seen on some systems. That box takes the sine wave signal and converts to 5v square wave.
Hood
-
Hood thanks for the answer. How can i check this ? I have loked at the data sheet that i atached but i coudl not find anything. The suppy is 5 V Do you have any idea who sells these boxes? If i understand correctly than the signal has to be square ?
-
Ok just looked, they are sine wave scales.
Do they have an additional box that came with them that they plug into or did they just plug directly to the old CNC control?
I have seen them on eBay but even second hand they are expensive.
You would probably be cheaper just getting compatible scales or using the motors encoders if it has them, if it is resolvers then might be better just to get encoders and fit to the ballscrews.
Hood
Hood
-
No they were direct plunged in the old controller. I do not have encoders on the motors. What i have to lok when i buy buy new linear scales?
-
If you look here and scroll down you will see the specifications you need.
http://www.ebay.co.uk/itm/JCXE-linear-encoder-glass-scale-for-digital-readout-650-mm-to-950-mm-travel-/221901376880?hash=item33aa5a4d70:g:mw4AAOxyOMdS9inu
As I said the other option would be to fit encoders, it may be the cheapest depending on the axis lengths.
Hood
-
Thanks Hood. Can you check if these linear encoders would be ok. Because we use them at work and i know they are good.
http://tela-ms.si/doc/NC%20linear%20scales%20TGM133_173.pdf (http://tela-ms.si/doc/NC%20linear%20scales%20TGM133_173.pdf)
-
From what I see, there are 3 versions of each of these scales. two seem to be sine wave, one mA (SI) and the other Volts.(SV)
They will not be any use.
The third kind however (DS) seems to be square wave and although it doesn't give the voltage it says RS422 which I think would suggest 5v but you would need to clarify that with the manufacturers.
Hood
-
Will borov tomorrow a linear encoder. See what happens. Thanks hood i will update as soon as i connect the encoder.
Regards.
-
Okay i borrowed the linear encoder at work and see attachment first response when i move it with hand. Tomorrow i will order new ones. Does anybody know how to remove the x axis linear encoder ( see atachment)
Regards
Rok
attachments :https://www.dropbox.com/sh/5zyu75x5e756ku1/AADZO8KOkRyX_a6qm7OCSv-pa?dl=0 (https://www.dropbox.com/sh/5zyu75x5e756ku1/AADZO8KOkRyX_a6qm7OCSv-pa?dl=0)
-
On glass scales there are normally slotted holes at either end that fixes them to the machine.
Hood
-
Yes that is true, but there is only one problem that the x scale is mounted between the x rails and table, its really hard to aces the scale.
Anyhow now i have anther question as i was ordering the linear scales the seller asked me where would i like to have a reference point for homing.
The reference signal in the middle of the measuring scale or a DCR (distance coded reference) that should mean that the machine trawls any ware for 20 mm
to get the reference signal.
Hood can you tell me which is the best for the CSMIO- Controller?
Many tanks and best regards.
-
Probably best with one that has the Index every 20mm or so as it means you can do Index homing and you just need to have the Home Switch positioned close to an Index. If you get one in the centre then you will have to have the home switch close to the centre.
Hood
-
Hello all together. I need some help (again). This week i decided to take some brake of work and work a little on my retrofit. I mounted the new glass scale and started to tune the axis. The first thing that bothered me was
that the axis was moving when the csmio-ipa was in emergency stop. Then i figured out that if i set the DAC offset to 0,18 V the motor/axis stays stil. Is this wright ? Or am doing something wrong ?
The next step was i had to figure out the steps per revolution so i gave in the pid tuning some small values and with help of the function axis calibration i got 1234.567901 per unit.
And the max speed was 3000 rpm. The biggest problem is the function auto-tuning in the PID section. I can not get it to work. How long does it take to finis this ? i waited like 10 min and nothing happens ( see my screens on the link at the end of this thread)
Im out of ideas what can i do ? Also how do i know how many Pulses/Rev does my encoder have ? i looked at the manual of my linear scales and i can find the information.
The motors are DC.
https://www.dropbox.com/sh/5zyu75x5e756ku1/AADZO8KOkRyX_a6qm7OCSv-pa?dl=0 (https://www.dropbox.com/sh/5zyu75x5e756ku1/AADZO8KOkRyX_a6qm7OCSv-pa?dl=0)
-
The Axes should never move when in E-Stop, if they are then it suggests you do not have an Enable signal controlled from Mach/CSMIO.
DAC offset is likely needed when CSMIO is in control and Mach out of E-Stop, to get the correct value you just press the DAC Offset button and it should work it out for you instantly.
Normally for steps per unit you would use the encoder counts per rev and divide by the distance travelled but as you have linear scales you will need to know the resolution of the scales.
If for example each graduation of the scale is 0.005mm then you would have 200 per mm, this however would likely need to be multiplied by 4 for quadrature, so 800 would be the value.
To know the pulses per revolution (encoder count) you would need to know the distance your axes travel for one revolution of the screw and then calculate with the graduations of the scale. if for example it was 10mm then your pulses per revolution would be 8000 (10 x 800)
AutoTuning should start instantly and not take too long, likely your steps per unit being wrong are the problem there.
Hood
-
Okay the resolution is 0.5µm. I was searching over the internet and i do not understand how to calculate this.
How did you calculate this ?
-
0.5um is 0.005mm so it would likely be as I mentioned above.
Hood
-
okay feel so stupid now :-[. So if i mark at the face of the spindle and than jog so long that the mark makes a 360 degrees turn and measure the length of traveling i will get the distance that i have to multiple with steps per unit. is this right ?
Thanks Hood.
-
Okay i must be doing something wrong. I can not tune this. Here are my steps.
At first i set up the steps per : 800
Than i marked on the end of the spindle a line to se when does it turn 360 degrees. a the same time i was measuring how far does the machine trawl. (see pictures)
And i measured the distance 5.08 mm therefore the encoder count 5,08 x 800 is 4064. I stil can not complete autotunning :(
measurement setup : https://www.dropbox.com/sh/5zyu75x5e756ku1/AADZO8KOkRyX_a6qm7OCSv-pa?dl=0 (https://www.dropbox.com/sh/5zyu75x5e756ku1/AADZO8KOkRyX_a6qm7OCSv-pa?dl=0)
Best regards
-
Try Steps per unit at 200 and Encoder at 1016.
Really though it should have worked with the numbers above, just moved 4x further.
In the Auto Tuning page, how many counts do you see for one rev of the ballscrew if you turn it manually?
Hood
-
Okay i fount the error, i was not patient enough. The auto tune function needed a lot of time. Now it has finished but there is stil a problem. Its like the motor is always correcting the position. I uploaded the video on my drop box if you would take a look. Is there a setting for this? When the auto tuning is finished , does it give any messages that it has finished ? I also attached the seetings.
The actual eror goes from 0 to -1 and 0 to 1 when the motor is correcting the position
details:https://www.dropbox.com/sh/5zyu75x5e756ku1/AADZO8KOkRyX_a6qm7OCSv-pa?dl=0 (https://www.dropbox.com/sh/5zyu75x5e756ku1/AADZO8KOkRyX_a6qm7OCSv-pa?dl=0)
-
That looks like you have a lot of backlash. For that amount of rotation of the screw I would be expecting a much larger count in the plugin, it is only moving 1 or 2 counts., so it would seem the screw is rotating but the axis is not moving and thus the scale if not registering movement.
I presume you have the correct value in Motor Tuning in Mach for STEPS PER UNIT?
If it is indeed backlash you will have to eliminate it as you will not get things tuned very well, that is the downside of linear scales rather than encoders, any backlash and it is extremely hard to get a decent tune.
Hood
-
Yes the steps per unit is 800 like we said before in the previous steps. If im understanding, you would say that i have a bad ball screw ? Yes the ball screw is constantly moving and the axis is still.
Oh man what a week everything goes wrong :/
Thanks Hood i would not know what to do without your help!
You are a HERO!
Best regards
-
That amount of rotation of the screw without movement would be much worse than I could imagine from a bad ballscrew.
I would think it much more likely that the fixing of the ball nut or the bearings on the screw are loose or damaged, check to see where the looseness is before contemplating replacing the screw. Bearings will be a lot cheaper than a screw and loose adjusting nuts will be free to fix :)
Hood
-
okay i have connected the Z axis and it has also movement (you can see it n the link of drop box). I was looking at the movements and i can not say if one is moving mor then the other.
If i fit the measurement gauge to the axis i donˇt get any movement. But when i give in to the axis calibration to move for 0.01 mm the axis moves for this amount.
Is it possible to tighten the ball nut ? I have no experience. Do you have any suggestions how can i check the bearings?
I have checked all the screws that hold the nut in place and they are all ok.
There is a strange thing when i click apply there is no movement for a couple of seconds, than it starts again to move.
My motivation is at the bottom wright now, because i have spend a lot of money on this machine and as far as i shows she does not like me! If it shows that i have to replace the screws and motor the machine will be new. :( Playing the lottery looks like my only hope.
Best regards
Dropbox z movement :https://www.dropbox.com/s/1lvwdd7wee6k7c1/20160428_114613.mp4?dl=0 (https://www.dropbox.com/s/1lvwdd7wee6k7c1/20160428_114613.mp4?dl=0)
-
Ok what to do to be sure is,
Move the axis one way, by maybe 1mm, with the gauge against it then move it the opposite way by the same amount and see if it returns to zero. If it does then there is no backlash.
You have relatively course pulses per rev due to the linear scales only being 0.005mm resolution and it could just be the way things will be.
A servo does not sit stationary like a stepper motor would, it moves back and forth a small amount to keep to the position. With good tuning and a highish resolution encoder you would see no movement as it would be so small but it will be moving.
I think your tuning is still not great, I think you will need to do the auto tuning at a much higher speed, I did mine at 100%. I aso had to fine tune the servo tuning after I did the AutoTune as it was fairly close but not as good as I liked. It was a long time ago I did the tuning so maybe the Auto Tuning has improved since the early days.
Hood
-
When i finish pid tuning and the try the axis calibration, then i get different steps per unit. I accept the steps per unit and i start again the pid tuning with the new value of steps per unit. Is this right ? Or im doing it wrong ?
Thanks for the previous answer will try to do this today.
Regards
-
I have never used the axis calibration as I prefer to work out the steps per unit from the mechanical data. The calibration will only ever be as good as you can measure, working iot out from known values will be as accurate as your hardware.
Hood
-
I had a loot to do these past couples of mots and yesterday i got back to retro lifting the machine. And here i have a new problem, i have connected a few buttons and now i have a button for go to zero. And when i joog the y axis to a point lets say 200+ and i push the button go to zero and before it reaches the zero i push the stop button. The y axis starts to move back and forth like in calibration and it does not stop. Also the y coordinates stay put . The don t move at all. If i don t push stop on the machine it moves normaly to zero.
-
How have you set up the button, is it OEM Trigger or Brain or Macropump ?
Sounds like you may have the input duplicated in other scripts or brains or i/o.
Hood
-
I have set up in hot keys OEM and then i assigned the values 1000 for cycle start and so on... I have looked at the inputs and there is nothing double.
i can attach a video if you would like.
-
Ok, video would be good but also if you could attach your xml and I will see if I can find an issue in it.
Hood
-
Hood in the link below you will find the movie ( sorry for the bad quality and shaking) and my settings.
Thanks for your help. YOU ARE AWESOME!!! :)
https://www.dropbox.com/sh/5zyu75x5e756ku1/AADZO8KOkRyX_a6qm7OCSv-pa?dl=0
-
Ok I looked through your xml and all seems fine.
I will try this out when I get a chance but when that will be I don't know as I am very busy at the moment.
One thing, if you have Ver 066 of Mach I would advise you overwrite it with an older version, maybe 062.
The reason I say that is Ver 066 is known to have problems, whether yours is one or not I have no idea but iot would still, in my opinion, be better to not use 066.
Hood
-
No i'm not using the 0.66. Which one are you using ?
-
I use Ver 062
-
Ok today I cloned my Chirons' profile and reassigned some buttons so I have one as Stop and the other as GoToZero. I moved away from XYZ Zero by 200 on X and 100 on Y and then pressed the GoToZero button, the Z raised to safe Z then X and Y started to move, I then presed the Stop button and it did exactly that. I did not get any Y vibration like in your video, so not sure what is going on in your setup.
Hood
-
Okay did not see the replay until today. I have still the same problem. And i talked to cs lab they said send us the complete folder. I did and now no answer for an month.
I checked all the connections but can not find any errors. Now im running out of ideas.
Hood i have another question. If i would change the dc motors for servo. Is it possible to tune the servo controller and motor so good that i get rid of the movement that i have now and than use the feedback from the linear glass scales ? Also the only thing that i have to look for is that the servo drivers have analog speed controller ?
Best regards
Rok
-
i have a similar mikron WF3 DP with csmio-A, linear scales, heidenhain EXE602 converters and kollmorgen servos (AKM42 for xy and AKM44 for z) with kollmorgen 603 servo amps and i have similar problems with the pid tuning. my machine was an older version with 3 dc motors and one amp.
use your machine trapezoidal spindles and gears between spindle and servo or is it a newer version with ball screw spindles?
http://www.machsupport.com/forum/index.php/topic,33125.0.html
i think, i change the spindles to an ball screw type an belts.
-
woo cant believe there is one more unlucky soul with the same problems than me ;)! Just kidding!
Yes i have ball screws and gears between dc motor and spindle and 3 dc motors and one amp. And i can not get a good tuning mode. If i can see it correctly
you have also changed the amps ? Did you also change the motors ? I m looking to buy new amps and motors but this is a lot of money.
-
i changed the amps and the servos from callan type DC motors to modern servos, because i can´t find amps with +- 10v input, 120v DC and 90A peak for the callan type motor..
used servos about 200€ and amps about 250€ on ebay.
-
Will search on ebay. What is your problem ? Is the motor constantly correcting the position?
-
Is the motor constantly correcting the position?
no, that isn´t a problem. this is normal with linear scales on modern milling machines.
the problem is the precise and quick positioning with linear scales and spindle clearance.
i think, it would works better with encoders and spindle clearance.
-
But than you lose the precision ? Or I'm mistaken ? . h. Now i would like to know can you remove the back slash just between the servo driver and servo motor ? And than use the linear scales for feadback to the csmio?
-
But than you lose the precision ? yes
Or I'm mistaken ? . h.
Now i would like to know can you remove the back slash just between the servo driver and servo motor ? And than use the linear scales for feadback to the csmio?
i use the linear scales with csmio.
the servo amp use the servo encoder signal (internal), but i can connect the csmio input to the servo amp encoder simulation output.
i receive next week 2 ball screw spindles for x and y axis and than i remove the gears and change to belts.
-
Thanks for the answer.
Can you tell me which belts and pulley are you going to use?
-
hdt belts, 8 pitch, 25mm wide
22 tooth pulley on the servo side and 44 or 32 (y axis, no more space) tooth on the spindle side.
-
the x and y axis conversion kit is ready to mount.
i hope, it works...
-
i switched the x and y axis now to belt-drive.
the pid tuning is much better than with gears, but my spindle is not the best qualitiy.
i think, with zero gap spindles it works perfekt.
https://www.machsupport.com/forum/index.php/topic,33125.msg232442.html#msg232442
-
Greetings,
i have been also working on my machine these past couple of months. To sum up what i have done so far.
I have changed the energy chain and the linear scales. Also i have painted the whole machine. Now i have changed the operators panel
and added an cabinet for the new drivers for servomotors. For now i will stick with the gear transmission between the motor and spindle.
Will add a movie of my progress as son as i find some time.
Also the tuning is going to be a pain in the A**.
Here is also my thanks for HOOD he has helped a lot ! Thank you !
Best regards
Slovenec
-
https://www.machsupport.com/forum/index.php/topic,33125.msg236542.html#msg236542
the gear changing works.... not perfekt, but ist works. i must now optimize the spindlespeed macro.
-
Ok I have changed the servos and from Friday morning I am trying to tune the motors with no success. Now I am questioning my input data.
Please chek my calculations:
I am using Kinco servo motors and for feedback position linear scales. The resolution of the linear scale is 0.5µm if I convert it is 0.0005. To get the PPR I divide the Lead of the screw wich is 5mm with the resolution. And the PPR is 10000.
Now if I want to calculate the steps/per I just divide the 10000/5mm and then i get Steps per=2000.
If I set this parameter into the Mac3 and start auto tuning it finished but as soon as I tried to move the axis I get e-pid fault and in case I change the max error to a larger number i get sudden movements by jogging.
The transmission between ball screw and motor is with gears.
Can anybody please help me? what am I doing wrong.
Best regards
-
The "steps per" value that CSMIO wants to know is "how many pulses should I expect when the ball screw rotates 360 degrees". The steps are normally supplied by an encoder on servo or ball screw shaft. In you case, using a linear scale you know that the precision is .0005mm. (ie. one pulse every .0005mm). Your ball screw moves the nut 5mm per revolution. So the correct "steps per" for your machine is 5mm / .0005 = 10k.
Mike
-
Mike what is the encoder count then ?
And also thanks for your answer.
Slovenec
-
Sorry Slovenec for that ambiguous answer...
For your machine, the way I understand it, your "encoder count" is 10k. The "steps per mm" is 10k / 5mm = 2000.
So your calculation is correct... Are you sure there is no reduction in the gears that connect the servo to the ball screw? It is common to have a 2:1 reduction. If this is true of your machine then your "steps per mm" number will need to be 10k / 2.5mm = 4000
Mike