How to setup Chinese AASD Servo electronic gearing

[JAZZCNC]

Ok thought I'd make a short video because I've had lots of people ask me about these Servo's and when I've asked what controller they use most come back with one that is in 100Khz region which just won't cut the mustard for a 3000rpm servo with 10K encoder count unless we do a bit of trickery in the drive.
This video shows the trickery to get 100khz doing the job of 500khz and just how fu#king useless the Chinglish manual can be, which to be fair this chinglish manual is quite good.

Anyway hope it helps some of you. (LEE: We need a component section or if we have one and I've missed it please move.:thumsup:)


https://youtu.be/bIdzlIKDJeM

[routercnc]

Nice work. Good to know if I ever go that route :encouragement:

[Neale]

Quick question (although it might be a long answer!)...

The reason for using electronic gearing here is because you want to include the controller in the feedback loop but it can't handle the encoder pulse rate (as I understand it). In this case, is there an argument for using step/dir signals to the servo controller (which most of the ones I've looked at seem to support)? OK, controller is out of the loop but the servo controller can handle the encoder feedback itself, possibly with some kind of built-in auto-tuning (guessing here) and maybe give better overall performance? Or maybe I've misunderstood completely and this is about using analogue input controllers?

[JAZZCNC]

The reason for using electronic gearing here is because you want to include the controller in the feedback loop but it can't handle the encoder pulse rate (as I understand it). . . . . . . .. . . . . . Or maybe I've misunderstood completely and this is about using analogue input controllers?

Yep your miles off Neale.!! . . . .:hysterical: . . . Did you watch the video because I'm struggling to see how you came to that conclusion because think I made it pretty clear what the problem was.?

This as nothing to do with Closing the loop or if using Step/Dir or Analog it applies to both.

But I'll explain it here in text for you and anyone else who's not sure what was meant in the video.

To get the full 3000Rpm from the these servo's which have encoders with 10,000 pulses per rev the motion controller needs to output 500Khz frequency. 10,000 * 3000 =30,000,000 pulses per minute / 60sec = 500,000 pulses per sec or 500Khz.

Most of the low to middle range motion controllers only provide 100 or 125Khz and the parallel port only allows at best 50khz and still be stable.
This means with the only 100khz you won't get the full RPM out of the motors. So to get the full RPM the drives provide electronic gearing.
It does this by taking the incoming pulses and applies a ratio based on the gearing ratio that you program into the drive based on how short on pulses you are to the 500Khz needed for full speed.

The cost of doing this is that you lose some of the resolution the encoders provide. For instance, with a 5:1 ratio, our encoder count becomes 2000 not 10,000 and it's this value you use to calculate the steps per setting. So, for instance, say using a 5mm pitch screw the resolution drops from 0.0005 to 0.0025. No big deal because we had an excess resolution to start with.

So the point of the video was how to calculate the electronic gearing and program it into the drive because it's a calculation based on a formula that is about as clear as mud.!

Regards the comment about Step/Dir then yes they will take both Step/Dir and analog. The drives also provide an Encoder output that can be used by motion controller to close the loop if it provides this feature. Some controllers allow this for both Step/Dir and analog, however, most Step/dir controller don't. In which case the loop is closed between motor and drive.
All Analog controllers take Encoder feedback because they can not work without it.

[Neale]

Yes - miles off - story of my life! OK, I put my hand up - question asked before watching video :redface:. It's a personal thing but I tend not to watch tutorial-type videos too much as I prefer written text - my addled old brain can then skip the bits I know and I can go over the bits that are new or complicated as many times as I like more easily. In this case, reading the initial post made me think of something I'm sure I read a long while back, re the problem with slower motion controllers not being able to keep up with the encoder pulse count when the loop is closed via the motion controller. Probably applies more to analogue controllers, of course. A moment's reflection and I would have realised that the pulse output rate from the motion controller equally needs to match the encoder pulse rate and it's equally obvious that a 2500 line encoder with quadrature pulses is way more resolution than a typical stepper configured with its 1600 microstep/rev resolution. Thinking of my own machine, I would be looking at 3:1 gearing down from servo to ballscrew (matching max revs of servo to critical speed of ballscrew) giving 30K pulse per rev resolution. Obvious why electronic gearing is needed but with the big advantage over microstepping that there is no loss of torque involved in getting to the right resolution.

Anyway, although I'm happy with sums and sorting the numbers involved, I'm sure that your translation of the manual is going to be very useful for the increasing number of people using servos like these and who don't speak Chinglish as their native tongue!

[Urban]

Thank you for explanation, that solve my same problem, with a little plc 20khz output frequency.
Your suggestion are perfetc, now the motor (BLDC,1kw, max 2500 rpm) run perfectly.
Only one question, if there are particular setup to fix for correct rotation, CW and CCW, because now, with your setup and same pulse generator, the motor run only in CW direction.. From the manual, very bad, seem not be particular setup are requestd for the rotation, in both direction.

[Muzzer]

Thanks for posting that, Jazz. I've come across this a couple of times now but still learning. And the machine translated Chinglish can be a puzzling thing!

The reason for having separate numerator and denominator might make sense if you have odd pulley ratios. For instance I have a 34t and a 60t, so the ratio is 0.5666666666666667 etc, whereas it is a simple 34 and 60 or 17 and 30 with no need to use a calculator.

[eMTea]

Hi!
I know the thread is already a few years old, but hopefully you're still here :)

I wonder, my planned controller (UC300ETH-Max) is 400khz, and the servos are 1500rpms.

Is the 400khz divided for all simultaneously sent pulses to all moving axis, or a separate frequency for each axis output?

[JAZZCNC]

Hi!
I know the thread is already a few years old, but hopefully you're still here :)

I wonder, my planned controller (UC300ETH-Max) is 400khz, and the servos are 1500rpms.

Is the 400khz divided for all simultaneously sent pulses to all moving axis, or a separate frequency for each axis output?

The 400Khz is per axis output.

[Doddy]

Jazz answered your question perfectly. But I'll answer with another question. How are you going to interface the UC300eth with the servo? The thought process behind that question is, if you're targeting 400kHz signalling then you can face a couple of challenges for correct operation. Look at the respective devices - and cables - in the signal path between servo and the UC300eth - make sure you're not setting yourself up for a fall. If you believe you need 400kHz (and there's some advantage, just make sure you balance that against the disadvantages). Look at establishing a differential drive to the servo and either avoid, or be very aware of any component in that signal chain.

[dazp1976]

Yep.
You want your step/dir signals wired differential.
Using the open collector way like you would with steppers is no use.
I was told OC is only good for 150khz reliably. So a breakout board that can wire differential is desirable to achieve the wanted 400khz such as the UB1.
There are converter boards to go between the bob and servo that can convert OC wiring to differential but I've never tried them.

Adjust the electronic gearing to suit whichever kHz setting is being worked with.

[JAZZCNC]

Yep.
You want your step/dir signals wired differential.
Using the open collector way like you would with steppers is no use.
I was told OC is only good for 150khz reliably. So a breakout board that can wire differential is desirable to achieve the wanted 400khz such as the UB1.
There are converter boards to go between the bob and servo that can convert OC wiring to differential but I've never tried them.

Adjust the electronic gearing to suit whichever kHz setting is being worked with.

Dazp suggests a good board because the UB1 is an excellent choice if you are using servos due to the fact its differential line drivers are nice and fast, however when you get into the 400khz + frequencys then everything becomes important.?
Cables and grounding must be quality and done correctly otherwise any stray electrical noise will cause chaos. Noise is the main difference between single-ended and differential signals and why it's better to use differential signals when the frequencies start to go up above 200khz

That said I don't think you will be using 400khz for 1500rpm motors unless they have very high count encoders. If they are using typical 2500 line quadrature encoders then you'll only need 250khz to reach 1500rpm which shouldn't cause you too much trouble and you could get away with using single-ended signals provided you use good grounding practices and careful cable routing away from electrically noisy components.

However, I would still prefer going to the trouble of using differential signals with quality shielded cables and careful grounding than selecting a lower frequency and using electronic gearing at the sacrifice of resolution and smoothness.

[eMTea]

Thanks for the heads up guys!

I'll be using the UC300ETH-Max, (UB-1+UD-1).

I'm retrofitting an older machine, and want to keep the performance.
So I want to keep 12000mm/min on the 4010 ballscrews, so 1200rpm on the servos.

(1200rpm x 10'000ppr) / 60sec = 200khz.

So if the 400khz was max combined, it would slow my total down, but since it is per axis, it is no problem.