Hello Guest it is March 28, 2024, 09:13:19 PM

Author Topic: 9x20 Chinese Lathe Conversion  (Read 88531 times)

0 Members and 1 Guest are viewing this topic.

Re: 9x20 Chinese Lathe Conversion
« Reply #70 on: June 26, 2012, 11:36:54 PM »
Nice Job Dan! 

Offline Dan13

*
  •  1,208 1,208
    • View Profile
    • DY Engineering
Re: 9x20 Chinese Lathe Conversion
« Reply #71 on: June 27, 2012, 10:51:46 AM »
Thanks JH. Working on the spindle servo now. Programming the drive to do all the stuff I want. A few little things still left then to get it fully running: fitting the home switches, the cable carrier for the carriage, fitting a guard on the chuck with a motor disable switch and fixing the tray and guards. Got them welded already, but the tray came out horrible. It's stainless and was TIG welded, but still got seriously (and I mean SERIOUSLY) distorted.

Dan

Offline Dan13

*
  •  1,208 1,208
    • View Profile
    • DY Engineering
Re: 9x20 Chinese Lathe Conversion
« Reply #72 on: June 29, 2012, 01:04:32 PM »
I am having a problem trying to tune the Ultra 5000 servo drive. Autotuning doesn't appear to work as the values it calculates don't give good performance and are the same values regardless of the mass connected to the motor.

I can't get it to hold position. It keeps producing this annoying buzzing noise. And it eats up all the current set in the current limit. If the current limit is small the buzzing sound stops and the current drawn in a standing state approaches zero as expected. But at this currents the system cannot operate as the motor doesn't have enough power.

Tried to play with all the parameters, but still can't get it stand still without producing the noise and drawing so much current.

I was able to tune the servo to fair results with the chuck removed, but have no luck with it mounted.

Does anybody have any thoughts? Must be something I am overlooking or doing wrong. JH? Hood? You set up far more servo systems than I did.

Dan
Re: 9x20 Chinese Lathe Conversion
« Reply #73 on: June 29, 2012, 01:10:12 PM »
Hi Dan,
  Till the pros get back ....
 I've only used the DDM and DSD drives but had this happen when I had the wrong motor set up for the drive.
Might double check ? ? ?
Russ

Offline Hood

*
  •  25,835 25,835
  • Carnoustie, Scotland
    • View Profile
Re: 9x20 Chinese Lathe Conversion
« Reply #74 on: June 29, 2012, 01:31:05 PM »
Dan, just read your PM and was about to reply but saw you had posted here so might as well also reply here.

Lathes spindles can be a real PITA to tune and I am really needing to redo mine since I added the chucking cyl;inder.
Anyway, AutoTune has never really worked that great for me, I have found that I need to set 50% or more for the current in auto tuning to get it to even work and setting nearer 100% gave best results.
Now to your buzzing, mess with your filter, it defaults to 150 but I have found I usually need to alter that, on some spindles I need more and some a lot less, I have only 1 on the lathes spindle and I think on the Beavers spindle I have 200 so mess about and see but I reckon you may have to go lower, try 20 for a start.

Hood

Offline Hood

*
  •  25,835 25,835
  • Carnoustie, Scotland
    • View Profile
Re: 9x20 Chinese Lathe Conversion
« Reply #75 on: June 29, 2012, 02:20:45 PM »
I just remembered its not an AB motor you have so what Russ says may be worth checking, if you do not have the custom motor set correctly you may get issues.
Also worth noting is having a low count encoder makes tuning a lot more difficult as an example here are a few screenshots of exactly the same PID settings and same manual tuning settings. The only difference is I have interpolated the encoder (smart encoder with 1024lines per rev) by x4, x256 and x1024. As you can see there is a huge difference between the x4 and the other two.
Hood

Offline Dan13

*
  •  1,208 1,208
    • View Profile
    • DY Engineering
Re: 9x20 Chinese Lathe Conversion
« Reply #76 on: June 29, 2012, 03:17:33 PM »
Thanks guys. I checked the custom motor parameters and found a small discrepancy: torque constant was set to 0.389 instead of 0.336. I fixed it, but this is a small error and I don't think it would be that much of a problem. But I will play with tuning tomorrow to check. Oh... and there is a parameter in the Ultraware motor editor called "Rated voltage" - the one that I filled in there was "Maximum bus voltage" from the motor specs. Assume this one is correct.

Hood, I am not sure which parameter you mean when you say you had to increase the current to 50% or more.

I tried the filter, but at low values the system is very loose and increasing the filter value brings back the buzzing.

Your point regarding the low current encoder might be the case here as the encoder on this one is only 1000 lines per revolution. The images you attached are very interesting. Didn't realise it would play such a significant role on a spindle.

Dan

Offline Dan13

*
  •  1,208 1,208
    • View Profile
    • DY Engineering
Re: 9x20 Chinese Lathe Conversion
« Reply #77 on: June 29, 2012, 03:23:17 PM »
Oh... and one another question. In the custom motor setting window there is a parameter called "Inertia", may be if I enter there a value accounting for the chuck inertia it will help? Or shouldn't I do this?

Dan

Offline Hood

*
  •  25,835 25,835
  • Carnoustie, Scotland
    • View Profile
Re: 9x20 Chinese Lathe Conversion
« Reply #78 on: June 29, 2012, 06:23:47 PM »
Its the step current in the auto tuning window, may be totally different in the 5000 drives though.
Regarding the inertia I dont know if I would do that or not as it states its the rotor inertia and not the system inertia.
I think the encoder actually affects the spindle more as it is a constantly changing mass and also you are varying the loads quite considerably all the time. The low count encoder doesnt allow the drive to respond nearly so quickly to changes. I can set the encoder to 4x and twist the motor shaft and I can feel it moving and fighting me but with exactly the same tuning  but the encoder set to 256x I cant even feel the shaft move as the drive corrects it much more quickly. On an axis the loads are not changing so dramatically so its not quite as important.

Going back to the step current for auto tuning, how I discovered that was like you mentioned auto tuning always seemed to throw the same numbers which when I came out of auto tune and enabled the drive the motor shook and jumped and then shortly after faulted. That is when I went and tried manual tuning. I set the scope to channel A as current command and set the scale to 50% of the drives peak which is about 47 amps (94 peak on  this drive) but when I started the manual tuning the current was way off screen. So I figured the drive was not able to use the set parameters as it was needing to use more current but that had been capped by the 50% setting.  I increased the scale to 80% of the drives Max current and it was on screen so I determined due to the mass I was needing a much higher current in auto tuning. I tried that and got much better results but they were still not great so I went back to manual tuning. As I have a back gear in my lathes headstock I have backlash between the spindle and driven shaft so I also only tune the spindle in one direction, this likely wont be a problem for you however as you are direct drive to the spindle.

The mills were much easier to tune as they had very little mass compared to the lathe and they didnt vary nearly so much.
Hood
« Last Edit: June 29, 2012, 06:31:24 PM by Hood »

Offline Dan13

*
  •  1,208 1,208
    • View Profile
    • DY Engineering
Re: 9x20 Chinese Lathe Conversion
« Reply #79 on: June 30, 2012, 01:50:08 AM »
Ah... no, it's the same with the 5000. Also have the step current there, but again no matter what (changing the step current as well) the autotuning gives same parameters.

I tried the inertia thing and it didn't make any difference at all.

I went back and removed the belt and so tried to tune the motor alone, but interestingly I was not able. Exact same values that worked for me before, didn't this time and I couldn't tune it any better than with the chuck together. Really have no clue what's happened.

Dan