Servo choices

[spoiledbrat]

Hello, all. I have arrived at the time of my build where it is time to purchase servos and drives. The machine is a 5x10 foot gantry router (duel driven gantry, rack and pinion), with 12 inches of Z (ground ballscrew). I have chosen 1kw, low inertia servos.

I have narrowed my selection to Automation Direct servos (SVL-210), and Yaskawa SGMSV-10A3A61. Both have nearly identical torque values for both rated and peak.

The inertia of the Yaskawa servo is 0.000174 kg-m (squared), and the inertia of the SVL-210 is 0.00026 kg-m (squared). The SVL is a closer inertia match, but not by much.

Purchasing the SVL-210 will save me over $3000 when compared to the yaskawa. My only holdup is that I have not heard of any industrial CNC using the SVL-210 servo. Yaskawa is everywhere, though. Will I see a difference if I choose not to purchase the mainstream, name brand servo?

My main concern will be rapid accel/decell values on the Y (rack/pinion/gearbox) and the Z (ground ballscrew) axis.

Both drives have the features I need. These will be controlled by an ethernet galil card and Mach3.

Please, someone help me!

Rob

[mactec54]

spoiledbrat

The difference between the 2 motors/drives is night & day if you can afford the Yaskawa I would get them, you could look at the Sigma II motors/drives as well they may be a little cheaper than the Sigma 5 & you should not need over a 850W motor for your machine, if you have 2 motors for the Gantry then you may get away with 2 500W motors

[YZF]

From a performance perspective two servo motors with the same (or close spec) should perform the same. Do the motors have a similar torque constant? Voltage constant? Coil resistance? Inductance?

Unfortunately it's not unheard of, even for big recognized names, to tinker with their specs a little bit and/or have variability in their process. Always ensure enough margin.

At the end of the day the brushless motor has magnets and coils. You drive current through the coils and you create a torque. A cheaper brand may use cheaper bearings, a cheaper encoder or somewhat inferior materials but that's not a given. I've worked with many servo motors and drives from many different vendors. (e.g. Parker, Baldor, Kollmorgen)

The inertia matching criteria is somewhat of a myth (google this if you don't believe me) and for many applications a higher motor inertia will provide better stability and allow higher gains. At the end of the day it is the stiffness of your system all the way from the motor to the load that will limit your performance.

Rated power is just one criteria for picking servo motors.

All your drive does is convert a voltage command from your controller (the galil) to a current in the motor's coils. It implements a current control loop to do that. Drives vary in features and may be less or more reliable but that's about it. More advanced drives use field oriented control which is a slightly more efficient way of driving than the basic sinusoidal drive.

So. The biggest impact on your system's closed loop performance is its stiffness or open loop frequency response. The second biggest impact is the control algorithm your controller implements and how well you tune it. The servo motor and the drive are much further down the line assuming they're properly sized/rated for your application...

[spoiledbrat]

In the spirit of backyard, over-engineering, I will be using 1KW servos on all axis. My needs require high acceleration for the Y and Z. I could care less about top speed. My gantry, with carriages and 4 spindles, will weigh over 1100 lbs. The Y-Z carriages will weigh about 150 lbs, and I will have two of them running slaved for mass (well, sort of) production parts.

I do appreciate your input, Mactec54. What exactly are the differences?

Rob

[mactec54]

spoiledbrat
The Yaskawa drives have autotuning,(with real-time adaptive load control) so once you have set the rigidity setting in the drive then you never need to touch the drive again I think there is like 3 parameters that you need to set & you are done, the other main thing on the cheaper motor is the encoder with only 2,500 count the Yaskawa that your were looking at has a 20 bit encoder has 1048576 counts per rev the larger the encoder count the smoother the machine will run, the encoders are mostly taken care of in the drives so you don't have to worry about the big number

That is why I said the sigma II motors would do the job for you they have a lower count encoder that would still work great for your machine, The sigma II motors have a 17 bit encoder (32768 cpr) but the drives for them only read up to a 16 bit encoder so that will give you (16384 cpr) with is still very smooth, all these numbers you x by 4 as well, ( quadrature ) There are many more things that you are paying for in the Yaskawa drives Motors, But when it come's down to it you need to make the choice what you want not anyone else

[YZF]

I think spoiledbrat was going to use a Galil controller in both situations. I.e. he wouldn't be using the Yaskawa as a controller.

[mactec54]

YZF
No one said he was using the Yaskawa servo dive as a controller (it is a servo drive not a controller), He is using Mach3 control software to the Galil card to the servodrive, what is in the drive is independent of the Galil & Mach3, If his computer is good he may not need the Galil card as the drives will take step & dir from the Mach3, but he will get more speed if he needs it by using the Galil card that is all the Galil is needed for

[spoiledbrat]

Mactec,

Thanks for the extended info. I have decided to go with the more renowned servo package on this build, as I cannot afford to buy wrong this time.
1 million counts? Wow. You said that the drive takes care of this. I will need this servo to be able to turn 50 revolutions per second, and the galil card can handle 3 Mhz (12Mhz quadrature). With a 2500 ct encoder (10,000 quad), 3000 rpm = 500,000 hz. Well inside the window. I will have to read up on the yaskawa drive a little more, to better understand how the drive handles this.

Thanks,

Rob

[YZF]

I'm feeling like I'm intruding on your private conversation here... ;-)

Many modern servo drives are also servo controllers. That is they implement velocity and or position control loops, not just a current loop. In fact if the drive can take step and direction pulses that means it has a position controller built in. So the controller features you are talking about would not be available if the Galil's servo controller is used. Then the servo tuning for the position and velocity loops would be done on the Galil, not on the Yaskawa, and the Galil would just be giving a current/torque command to the Yaskawa. If Yaskawa has a superior adaptive servo controller you would have to use its controller and not the Galils to take advantage of those features. Maybe it is indeed excellant - I haven't used that servo drive.

At any rate. If the original poster is happy by spending more money and the feel of a quality system he gets from that that's fine. If the Galil is closing the servo loop for both systems it's mostly going to be in the feeling he gets...

[mactec54]

YZF
No you should not feel that way, everyone's input is great,& if you had of used Yaskawa drives/motors you would be say the same thing, they are the best without any doubt, The quality & reliability is worth the extra money, & with more than 4 milion count encoder in quadrature, this gives you a supper smooth running machine

[Al_The_Man]

I have used Galil extensively, I like prefer the loop to be closed back to the card and the ability to use simple economical transconductance amplifiers such as AMC/Aerotech/Copley.
Al.

[spoiledbrat]

Wow. The deeper I dig, the deeper the rabbit hole gets! I am using the Galil card mainly to have something that Mach3 can communicate with, and attain higher rates of speed than the parallel port will allow. As this is a machine I plan to earn a living with, I do not want any non-industrial (read: hobby) parts on it.

As it turns out, the Galil card homing routine is homing to an index, as I was seeking. It would seem that the drive will do the same thing by itself.

Pointing out that the servo drive is really a single axis controller is really confusing me at this point. I thought that the Galil card was the controller, and that the Sigma 5 drive was the drive. I am not understanding how I am losing features here, just yet. I did learn that I can set the ratio for the pulse count to whatever I need to for the controller, though.

I will be implementing a DL-205 PLC for all of my I/O functions (This machine will have 2 ATC's, one pneumatically actuated dust boot with position sensor, two spindle temp sensors, coolant pump "on" sensor. I am not sure if I am even scratching the surface on the I/O that will be required. But, that PLC has 6 slots.

YZF, what features are you saying will be inaccessible to me since I am using the Galil card?

Thanks in advance,

Rob