Having problem with blending arcs together

[Nandi]

If you need some more informations about the system setups, i will be here to answer.

Yours sincerely,
  Nandi

[Jochen]

Hi Nandi,

I have found a workaround for this issue but I don't know if this is feasible for you. Apparently the velocity glitches that you see when blending arcs with different radiusses are inherent to the blending algorithm on the board and can't be changed without major rework of the firmware. Still you can get rid of this problem by keeping the angular velocity constant.

The angular velocity scales directly with the radius, so before loading the move constraints for an arc, you should scale the vector velocity with the radius (and maybe a constant value). In my tests this has eliminated the glitches completely. The drawback of this method is obviously the fact, that the vector velocity doesn't stay constant for arcs with different radiusses.

I hope this helps,

Jochen

[Nandi]

Hi Jochen,

Thanks for the fast replies. This means the velocity glitches appear if we blend two circle with large angular velocity drop? Maybe i need to decrease the smaller arc's vector velocity to reach the larger arc's angular velocity. I should filter out this rare cases and transfer the velocity values only there.

[Jochen]

In my tests I have found that the glitches become more significant, the lower the velocity is and the higher the difference between the two radiusses is. That means, that as soon as you blend two arcs with different radiusses without adjusting the angular velocity, a glitch will occurr, but in most cases it will be very small and thus not noticeable.

Jochen

[Nandi]

Hi Jochen,

Replacing the motion control board, can be a solution? It's very inportant to keep the plasma head at constant velocity for good cutting quality.

[Jochen]

Hi Nandi,

constant speed is a typical requirement for laser or plasma cutting applications. Unfortunately you have come across an issue in our firmware that has been there for a long time, but it has not caused any troubles to other customers so far. As this part of the firmware is the same in all NI motion control boards, changing the board type won't solve the problem. For now contouring would be the only feasible option.

As I have already told you, I have escalated this issue to R&D. After talking to my colleagues in Hungary we have agreed to raise the priority of this issue significantly. After hearing back from R&D we should discuss our further proceeding.

Jochen

[Dre-motion]

I have the same problem with a Plasma cutter. We have a 7344 card with two servo's. Blending line to arc, and line to line works fine. Blending multiple arc moves gives motor jumps, and the blending does not occur. Our system also has M and G-code input. I see three possible solutions: (1) contouring mode: this requires quite some programming converting our G- codes into a contouring pattern including maintaining constant cutting speed, it looks to me that the biggest disadvantage is the deviation from our software architecture, and losing the possibility for "infinite" cutting length. (2) NI comes with the major bug fix, this is the option I hope for (3) we switch to another motion card manufacturer wich does not have this problem, this is the option I fear in case NI can't deliver a bugfix.

Does this bug also occur in older versions of the NI firmware? Maybe switching to NI-motion 6.xx could help?

[Jochen]

Dre-motion,

I'm sorry that this issue is causing trouble to you. As much as I know, R&D is working on this issue, but I currently can't provide exact information about the release date of a driver version that contains the fix. I'm in contact with R&D again and I should be able to provide better information soon.

The issue was also present in older firmware versions, so downgrading is not a resolution. Could you please provide some information about your deadlines?

Thanks and kind regards,

Jochen

Message Edited by Jochen on 12-08-2008 01:47 PM

[Dre-motion]

Hi Jochen,

The plasma cutter is about finished, except for the arc-blend bug (and some minor post processor bugs).

We planned to ship this machine at the end of Januari 2009.

André.

[Jochen]

Hi André,

I have received some information from R&D. The issue was analyzed thoroughly, but apparently the issue is deep-rooted in the architecture of the 73xx firmware and there is no  short- or mid-term solution in sight.

Besides the alternative options that you have mentioned, another idea comes to my mind: The 7350 provide a low-pass filter for the control output. For servos the velocity glitch that you see could  be damped significantly by adjusting the Cutoff Frequency appropriately. I know, that you are using a 7344 and not a 735x, but if your local NI branch could send you a 735x for testing you could give it a try.

I don't think, that this could fix the issue completely, but it could reduce it to an acceptable level without redesigning your code.

Regards,

Jochen