stepper to servo

[Stevie]

Hi guys

I'm just about ready to move to servos after a year with steppers
But you know i thought i knew what to do; turns out I'm kinda lost

I know steppers are step/dir but how do servos really work; is it the halls; or the usually attached encoders that tell the amp whats going on; the 10v pluses; can they come from the LPT or would I be better with some interface from the serial (AKA Rutex)
I have 3 beautiful brushless servos; all basically brand new; and 3 matched copley brushless amps; the headers are already setup for these servos
I have no connectors yet; I'll get some soon; but I want to start installing them into the panel; and rough in the PS etc

I think over the past year my brain lost all what I had read

edit ////////// should have added; turbocnc or Mach2 as controler

[Swede]

Stevie, I'll tell you what I know and not speculate on what I don't. My own servo system is a "modern" digital system in that the amplifiers, like a stepper, accept step and direction signals from the control. The encoder is the primary feedback device. Within the EEPROM of the amplifiers resides a number of parameters which the system uses to control the motion.

Let's say the control asks for a 5" rapid move. For simplicity sake, the servo may be viewed as a stepper with the number of "steps" equal to the encoder resolution. The control creates a ramping profile, and begins to move the motor. It has already determined the number of encoder "steps" that must be moved to make the 5". While the system is in motion, the control continually tracks the position and importantly the velocity of the servo. Deviations in position and velocity are continually corrected based upon a mathematical algorithm known as a PID (Proportional-Integral-derivative) loop. These values are known as the servo's "tuning" and contribute to the speed and accuracy of the system.

This whole "stepper-like" behavior of a digital servo system confused me at first. It seemed that with this method, why not use steppers anyhow? Well, you still have all of the torque and speed advantages of a servo over a stepper. The servo's torque does not rapidly fall off with speed. There is no cogging; the PID and the control create the smooth motion expected with servos.

I've also found that there is no advantage to a super-high res encoder. If I have one servo with a 2000 line encoder (8000 "steps") vs a 500 line encoder (2000 "steps"), to get any decent speed out of the 2000 line encoder, I have to tell the system to treat the ticks as blocks of 4 or 8, defeating the purpose of having all those lines on the encoder disk. With a 500 line / 2000 tick encoder, each tick consumes 0.0001" of axis travel on a 0.200"/turn screw direct drive, plenty accurate and usually beyond the mechanical capability of our home shop conversions.

The hall lines are for commutation of the brushless servo... they're part of the basic hardware that allows the servo to be rotated in a specified direction. Brushed servos don't use commutation, all of the info of the state of the servo (position and velocity)comes from the encoder.

That's all this CNC noob knows at this point. Having messed with steppers long ago, the speed and accuracy of a good servo system still amazes me.

Someone with more experience with older systems (resolvers/tachs) will need to pipe up with their theory.

[Stevie]

Great thanks Swede

So if these amps have no encoder input; i have to pipe the signals to something else; and it will have to tell the amp how much to move??

I have 1 of the new dig amps with step and dir input
I'd need 2 to convert my lathe

[ViperTX]

Stevie,
All the Gecko, Rutex and hobby equivalents have the Step/Direction....this was done for compatibility reasons....the way the "big boys" drive servos is a little different as is their software.

So, if you're going from steppers to servos should be a replacement of the drivers and power supply...the software should be the same....some parameters in the software will have require some change.

[ESjaavik]

Try this site http://www.worldservo.com/
It explains better than any of us how it works.

@Swede: Your writeup is very good! But I will go a bit deeper: There is one more thing to having a high number of lines on the encoder. A servo is driven by it's error signal. So it is not possible to make it stay within one count and keep it's position there. If you have a high number of lines, a small displacement will be known to the controller and it can take corrective action without becoming "nervous". But a good step/direction driven servo drive can usually be set up to your liking of how it interprets the step signal. There need not be any particular relationship between encoder lines and "steps". You can choose your step size (movement) to be just as big as you need to get the resolution your machine need. Your drive will use that to drive its Desired Position or target position if you like. Then it will use it's native resolution (encoder signal) to attempt to get there. The step size should however be larger than the encoder resolution. This is what you describe, except that in many drives the relationship need not be an integer number.
Let's say you like imperial and I work in metric. You tell me to move 2 1/4" and I can do that with as much accuracy as if you had told me to move 57,15mm because I will convert the target before trying to get there. But if I have a leadscrew with 5mm/turn I will have a hard time pinpointing my target using the handwheel graduation. If it is 1mm/turn I can do much better. So it is for a servo, the finer "graduations" it can "see", the closer it will get to the intended target. But what you give me as target position does not really influence how close I can get to it.

Or the short version: The step increment is the resolution of your commanded position. The encoder resolution is what the PID have to work with to get there without delay or overshoot.

The hall feedback is part of the electronics that does exactly the same job as the brushes + commutator on motors so equipped. Except that the electronic version can adjust the commutation timing on the fly. Just like the ECU in your can can fine tune the pre-ignition point while with your old style vacuum/flyweight distributor points it was fixed by the workshop/factory. You may know that some brushed motors are specified to run in a certain direction, and will do so more efficiently than in the other. This is because the brushes are skewed in one direction. They cannot be used as servos. A servo motor is adjusted to 0 degrees "pre-ignition" so it will run equally well in both directions. This makes it less efficient, but is necessary as it have to behave the same in both directions. With hall position sensors and electronic computation of the "ignition points", it can get high efficiency in both directions.

Why does it need both encoder and hall sensors? That is because the encoder does not tell it where it is as you apply power to it. But the hall sensors will tell it where the rotor is positioned relative to the frame even before any move is done. Admittedly with a very low resolution, but high enough to tell the processor which windings to energize to start a correct move.

I hope this made sense rather than confusion.

And Swede, I hope you don't take this as an attack on your explanation. I think it was good, I just hope to bring it further. It is useful to draw parallels to a stepper system in the beginning, and then further on realize the dynamic nature of a servo drive.

@Stevie: This guy use Copley drives http://www.lmwatts.com/cnc.html
Maybe you can glean some useful information from his site? He uses EMC as a controller (so do I). If you do your homework and ask qualified questions, I think he will give you an answer. He is resourceful and helpful. Keep up your work. Watching a well tuned servo drive at work is just amazing!

[ViperTX]

I think we have a disconnect:

What CAM software are you planning to use?
If TurboCNC or Mach2, I don't believe they will handle the Copley Amps.
EMC is the cheapest alternative for the Copley amps, others would be AhHa (low end), Galil....etc.

[Al_The_Man]

@Stevie: This guy use Copley drives http://www.lmwatts.com/cnc.html
Watching a well tuned servo drive at work is just amazing!

Those drives are DC brushed analogue input.
Just to put things in perspective as regards encoder feedback rate, the motion card I am using at present uses a Texas Instrument 32/64 floating point DSP processor, this reads in encoder counts up to 80MHz and has a 50µsec/axis servo update with a 250µsec trajectory update rate, Will multi task 8 motion programs and 8 plc programs.
BTW, if anyone wants to see a servo system being tuned with an explanation of PID, is to watch the video by Dr Jacob Tal on the Galil site, it refers to Galil products but the principle applies to all servo controllers.
Al

[Stevie]

great replies guys

I and I'm sure many others are all learing by this

[Stevie]

Those drives are DC brushed analogue input.
Just to put things in perspective as regards encoder feedback rate, the motion card I am using at present uses a Texas Instrument 32/64 floating point DSP processor, this reads in encoder counts up to 80MHz and has a 50µsec/axis servo update with a 250µsec trajectory update rate, Will multi task 8 motion programs and 8 plc programs.
BTW, if anyone wants to see a servo system being tuned with an explanation of PID, is to watch the video by Dr Jacob Tal on the Galil site, it refers to Galil products but the principle applies to all servo controllers.
Al

Should i be looking for a motion card for the 3 brushless servo amps i have?

I know the 2 3 channel amps i got are way different; they are for way down the road!!

The other copley i got is this and it seems to do just about everything under the sun

[glassgunner]

Where can I buy a cnc package that includes 3 servos, drivers, controller,
software, etc in one bundle.
Thanks,
Glassgunner

[Al_The_Man]

Packages that use these type of motion cards are Camsoft/Galil, Parker/Acroloop & Delta-Tau. All offer CNC software.
Al

[Swede]

This is great stuff. Einar, thanks for your better explanation. I am slowly coming to realize that servo systems relate to (and correct) errors in velocity and position. My understanding is that somewhere within the system, there's got to be an absolute encoder count being tracked, otherwise, the servo could not decelerate and brake to a halt at a specific location. That is why it occured to me that excessively fine encoder resolutions simply burden the system, in that they are forced to count every line to determine axial location.

I don't want to hijack this thread, the idea is to get Stevie's concerns answered. I'll lurk in the background.