block processing time

Jay,

I just tried cutting an ellipse on a Byspeed machine and it cuts it without the hesitations that a typical Fanuc Control would. I will certainly try one of these days to program a circle using your instructions.

Of course, Bystronic makes their own CNC and PC Based Control and it seems to do well.

Mayron

Hi, Mayron
Interesting.
I'll have to try it on some of the newer machines on the floor.
(I'm back at our alma mater for a while)
cheers
Jay


Jay Maechtlen

hi
I think like Mayron, its depend of the cnc series!!
I also try ellipse on LC XI, the lastest amada and I reach 22-25m/min on 1mm aluminium (soory for m and mm Im french)
ans there was not problem !!

the pooh

The problem is not the controller but the servo positioning. 5-6 in/min with 0.002 in steps is equivalent to 50 steps per second. That's pushing against the settling time requirements any high-performance positioning servo.

TTFN

"Settling time"
That's something I am ignorant of.
What is the purpose of that?

Below is my first response, which still (I think) applies:
Most modern servos have a servo update of what? 5ms or less?
That would be 200 updates per second.
That's velocity update to drive toward an established coordinate.
On the ancient Prima CRG 306 robot controller I worked with, the best block response was 32 ms, which is...
32 new coordinates per second.
Of course, that was running on 8-bit CPUs....
With points .002 apart, 6 ipm = .1 ips = 50 points per second.

That's only 50% better for the Fanuc than the 10 year older Prima machine: I am unimpressed.
Remember, to test this, you need to program coordinate points very close together. Ignore material requirements- leave the laser off and just dry run it.
See how fast the controller will process the program points.
regards
Jay



Jay Maechtlen

Settling time is the overshoot/damping that brings the plant into position that it was commanded to. Physics is physics. You're assuming that the total command/response time is due to the controller, which it obviously cannot be.

Plus, your experiment is designed to aggravate the mechanical aspects of the system. Consider the following: It has a 32-bit processor that can do servo updates at 33 kHz.

If you were to command your servo for constant motion, you could move and fire the laser at a much higher rate, using the encoders to determine correct firing position. By forcing the servo to stop and start, you've forced the system into a very slow process that's more dependent on the acceleration/deceleration and servo stability than on the processor throughput.

TTFN

This is a cutting /contouring system we're speaking of.
The axis should not stop while contouring, it should maintain the commanded velocity through the commanded coordinate points (or "close enough" to them according to various control parameters)
Therefore, I'm not at all convinced that 'settling time' applies to this test.
If these were G00 instructions, we might expect a different result. Another factor is the "contouring mode" versus "exact stop check" mode. In "exact stop check", the control must drive to each end point exactly, in contouring mode the controller can drive "close enough" on the way past a point en route to the next coordinate.
Fanuc does have some hardware/software options which may address this- "high speed processing" I think?

Anyway- anyone bothered to make a test?
Regards
Jay

Jay Maechtlen

Jay,

Why don't you e-mail me the dwg or dxf you are trying to test.

I will do it next week sometime when I'm back in the office.

nexba@yahoo.com

Regards,

OK Jay

You can also give the dxf file !

remi-1978@wanadoo.fr

I'll test as soon as possible on lasers equiped with fanuc 18i and mitsubishi CNC

But I can say already that amada laser equiped mitsubishi
can do it "normally" because I tested multiG01 - ellipse without any problem !!

The pooh