Are these servo/steppers appropriate?(Fastech EZI)

More
01 Oct 2010 00:01 #4459 by 1:1
the Mesa 7i39 - 3 phase, you mean 3 phase power ?

I'm single phase here. Unless it refers to something else ?

Interesting videos - the second one, that is what I do for a living - I am an automation programmer/operator/technician for theatre/event and I have flown people just like this - when I say programmer I mean I program the motion, not the code running the gear behind it ... More skilled towards setting up and running events (safely, creatively and timely) than building the systems - although there is a cross over, hence my interest here.

It is along the lines of what I want to do, although on more rigidly connected items where missed steps will mean some interesting inter-axes math so get back to a happy place - that 3D flying of people turns out to be usually pretty blunt in terms of resolution. Simple stuff could be programmed with g-code - but I think a system where things can be visualised in 3D over time and then the kinematics (if need be) and the rotational (motor) information is reverse engineered from those movements into velocity (step rate) is a better way to go, many examples online which I reckon you're already aware of... This is where a drive with its own internal loop would come in handy...

So, Brushless DC ... sounds great - are there suppliers that have a drive and motor - single phase 240v - high torque - hi incremental encoder count with div/mult - that can run its own internal PID *and* run 'open' with EMC2 ?

shucks ;)

Nick

Please Log in or Create an account to join the conversation.

More
01 Oct 2010 00:22 #4460 by andypugh
1:1 wrote:

the Mesa 7i39 - 3 phase, you mean 3 phase power ?

No, 3-phase motor. The mains power is irrelevant as it needs to be converted to DC for the drives anyway.

Interesting videos - the second one, that is what I do for a living - I am an automation programmer/operator/technician for theatre/event and I have flown people just like this

Actually, I think the people are just hanging. what EMC2 is doing is moving the ball. The G-code defines a ball position in 3D XYZ space, and then EMC2 converts that into 4 wire lengths on 4 separate stepper-motor winches (using a specially written kinematics module) and controls the motors to achieve that.

It is along the lines of what I want to do, although on more rigidly connected items where missed steps will mean some interesting inter-axes math

Then EMC2 might be exactly the right package.
linuxcnc.org/docs/html/motion_kinematics.html

You do have to be willing to delve into C-code, but you can compile the kinematics modules completely separately from the main source code.

So, Brushless DC ... sounds great - are there suppliers that have a drive and motor - single phase 240v - high torque - hi incremental encoder count with div/mult - that can run its own internal PID *and* run 'open' with EMC2 ?

Probably :) But you might have to find it yourself.
Actually, the encoder resolution downsampling is tricky, I would say it is worth getting a Mesa 7i43, it works under windows too, you can use it as just a 48 pin GPIO card, and it can be a USB device or a p-port device (P-port only with emc2 though, because of hard-timing requirements). I can see it finding a whole lot of uses for you.

But given your requirements, the Fastech drives look like they might be a good bet. If you can afford $500 for each motor then not getting a $79 7i43 to go with them is probably a false economy (I guess that half of EMC2 users go that way, or to a PCI equivalent when they run out of P-port pins or step rate or encoder count rate)
They do have pretty impressive torque and the speed is not dreadful. They definitely will work with EMC2, and a lot better than a standard stepper if not quite as well as a conventional servo. (And bear in mind that an awful lot of people, me included, are working perfectly happily with open-loop stepper machines)

Please Log in or Create an account to join the conversation.

More
01 Oct 2010 00:59 - 01 Oct 2010 01:00 #4461 by 1:1
Thats how we do 3D flying - but we use Maya or 3DS max to output the motor positions over a time period - then derive the required velocities from that ... same same but different ;)

What I meant by rigid bodies and inter-axes math was hmmm, ok imagine this scenario - you've got an object that rolls along the ground, it has 2 automated drive wheels (it holds itself up on casters or whatever) - it can rotate by driving each wheel relative to the other, then go forward by driving both etc... (very simple description) - imagine you can program it it to go somewhere on a cartesian plane, using bezier splines for the spatial path and a velocity profile over time (or whatever) - a very simple case in point>> you want to go forward x distance then somewhere along the time it takes to do this one wheel looses x counts, it's now rotated a little... PID ahoy! its now back on its correct speed - but its not going to hit its spatial target ... You can 'make up' the lost counts by over shooting or mirroring the error in the other wheel, again it wont make its target - what i want to make is something that can figure out *new* motion paths to adjust for errors as they happen to eventually it'll get to it target position - these new paths will probably look a lot more funky than the original path, and they'll get very interesting near the target, maybe asking for the motors to do something out of spec... Lots to think about I guess. Or simply enrol in a ME in mechatronics - ha ha

The 7i43 is for what exactly ? to get around all the I/O issues of the parallel port running 4 axes ?

I don't want the fastech stuff if it wont run "quite as well as a conventional servo" - the painful research continues for now ... Googling: "brushless servos for dummies"
Last edit: 01 Oct 2010 01:00 by 1:1.

Please Log in or Create an account to join the conversation.

More
01 Oct 2010 08:13 #4465 by andypugh
1:1 wrote:

Thats how we do 3D flying - but we use Maya or 3DS max to output the motor positions over a time period - then derive the required velocities from that ... same same but different ;)

Slightly different, in that doing it with EMC2 kinematics means that the calculations are done in real-time, so you can fly the ball with a joystick if you want.

you want to go forward x distance then somewhere along the time it takes to do this one wheel looses x counts, it's now rotated a little... PID ahoy! its now back on its correct speed - but its not going to hit its spatial target ... You can 'make up' the lost counts by over shooting or mirroring the error in the other wheel, again it wont make its target - what i want to make is something that can figure out *new* motion paths to adjust for errors as they happen to eventually it'll get to it target position - these new paths will probably look a lot more funky than the original path, and they'll get very interesting near the target, maybe asking for the motors to do something out of spec...


This is probably not an application for EMC2. You would typically set up an EMC2 machine with conservative parameters such that there was no real risk of missing a step, and if your feedback does report a missed step then you would normally simply stop the machine and report the error.
What you are discussing is much more in the realm of autonomous robotics, and I would be looking at stepper drivers and Arduinos if that was my game.

The 7i43 is for what exactly ? to get around all the I/O issues of the parallel port running 4 axes

Yes, and also to handle the high encoder count rates, it can track encoders and output step motor patterns up to MHz frequencies. a couple of orders of magnitude faster than software generation/counting.

I don't want the fastech stuff if it wont run "quite as well as a conventional servo" - the painful research continues for now ... Googling: "brushless servos for dummies"

They look like they should run a lot better than standard steppers, and are a lot easier to drive than servos. See them as super-steppers rather than sub-servos and they make a lot of sense. They can be driven directly by a microcontroller (such as the above-mentioned Arduino) whereas a Servo system needs a drive system, something to perform the PID calcs etc etc.

Please Log in or Create an account to join the conversation.

More
01 Oct 2010 13:43 #4469 by dab77
Thanks Andy for linking!
Hi 1:1, i'm the one who did the emc2-flying volleyball. is it real you're doing the same directly with 3dstudiomax out? interesting.
just right now i'm using splines in 3dsmax to draw the path, then i convert them with a script into g-code, and i feed emc2 with this code (slightly modified manually...). Then i still can jog the single motors, or the position of the object directly with keys, or with a joystick (..still haven't done..).
Interesting that we do the same job, where are you from?
I think you know the Fisher Navigator System..do you?

Andy, that system, used and sold by Los Angeles Fisher people, is the very professional system to do the 3d fly of objects (like cameras) and people!! they use this in making movies, for example...
in fact now my objective is improving the sistem, making it safe and powerful to do something like that..
ciao!

Please Log in or Create an account to join the conversation.

More
01 Oct 2010 16:45 #4473 by jmelson
1:1 wrote:

Hi,


Does that prove fastech is just an expensive stepper ? or is it quite good ?

Yes, pretty much. Maybe these are just a little better than a typical microstepping drive, but certainly not much. The smaller motor is down to half torque at 900 RPM and the big one at 1200. So, if you expect to get more than 500 RPM at the motor, you have to plan to have a serious torque reduction. Servos generally give full rated torque up to rated speed, which usually is several thousand RPM. You need to evaluate the torque requirements of your machine and figure the highest cutting and rapid feeds, and compute how fast the motor needs to spin at those feeds.

Jon

Please Log in or Create an account to join the conversation.

More
01 Oct 2010 16:52 #4474 by jmelson
1:1 wrote:

hmmm,

couple of off topic questions:

-Does 1 parallel port have enough I/O for EMC2 to run 4 closed loop axes ?

No. You really can't use the software encoder counter for servos. If the encoder count rate is greater than about half the BASE_PERIOD, then you will lose counts, which will lead to disastrous servo runaways.
With a 50 us base period (20 KHz) you can't safely have more than 10,000 counts/second.

To reliably count servo encoders you need a hardware encoder counter that can keep up with the count rate.
There are the Mesa boards, as well as Pico Systems. The Pico Systems Universal Stepper Controller can be used with step/servos like this in a closed-loop mode. It can reliably count 500,000 count/second on all axes. It can also control up to 4 axes on a single parallel port. It can generate 300,000 steps/second on each axis, and generate much smoother pulse trains that the software step generator.

The parallel port becomes a communications link, so there are no discrete step pulses between computer and USC board, and no individual wires for each axis.

Jon

Please Log in or Create an account to join the conversation.

More
01 Oct 2010 16:57 #4475 by jmelson
1:1 wrote:

I need to get drives that can run PID internally for other projects where I'll be feeing step/dir from other dumber devices (ones I design my own, ha ha) ...

Looking at the Gecko 320X brushed servo driver - can it run 'dumb' so EMC2 can run the PID and it effectively becomes just a power supply ?

Maybe EMC2 isn't for me ?

No, but what's the problem? Anyway, with a G320, you need some way to copy the encoder signals to the computer for closed-loop (to the computer) operation. I make a "Gecko Interface" that does that, powers the
encoders from the Universal Stepper Controller and makes an opto-isolated copy for the Gecko drive. It also controls and monitors the fault/res pins of the Gecko, and powers them off when in E-stop. This allows you to switch between manual-with-DRO mode and CNC mode and keep the position alignment throughout. The G320 handles most of the gain and stability, and the USC and EMC just manage position.

Jon

Please Log in or Create an account to join the conversation.

More
01 Oct 2010 17:09 #4476 by jmelson
1:1 wrote:
[quote
So, Brushless DC ... sounds great - are there suppliers that have a drive and motor - single phase 240v - high torque - hi incremental encoder count with div/mult - that can run its own internal PID *and* run 'open' with EMC2 ?
Nick[/quote]
I make brushless drives that run off DC, the specific voltage can be selected to match the motor. Up to 122 V DC, up to 20 A. We have tried it with motors from Pittman (NEMA size 11) up to Fanuc Red Cap that weight 15 Kilos.
Keling sells some size 23 (tested) and 34 (haven't tried these yet) motors at a very good price. I can set up the size 23 motors from Keling with a CUI encoder, and have them on my web store for $120. These would be awesome on a desktop or smaller mill, but the size 34 might be better for a Bridgeport-size machine.
The Keling motors would run fine on 50 - 75 V DC, just get an appropriate transformer/rectifier/capacitor.
I can advise on some of those details.

These would be used with our Univeral PWM Controller. You can see all these items at :

pico-systems.com/oscrc4/catalog/index.php?cPath=3

Jon

Please Log in or Create an account to join the conversation.

More
01 Oct 2010 21:15 #4478 by 1:1
andypugh wrote:

This is probably not an application for EMC2. You would typically set up an EMC2 machine with conservative parameters such that there was no real risk of missing a step, and if your feedback does report a missed step then you would normally simply stop the machine and report the error. What you are discussing is much more in the realm of autonomous robotics, and I would be looking at stepper drivers and Arduinos if that was my game.
....
They can be driven directly by a microcontroller (such as the above-mentioned Arduino) whereas a Servo system needs a drive system, something to perform the PID calcs etc etc.


Yup, I wasn't expecting EMC2 to do this part for me - and yes, an Arduino Mega is sitting on the desk right next to me :lol: although I plan to learn how to use the faster more grown up chips around also ...

Please Log in or Create an account to join the conversation.

Time to create page: 0.515 seconds
Powered by Kunena Forum