NordicCNC's build log

[jarjar]

I'm surprised how much these have dropped in price, last time I looked, which must admit is quite some time ago, Yaskawa kits were like $900-$1100 range for 400W. I've never used them for this reason so I may look at these when a suitable job comes up.! . . . How easy or hard are they to setup/tune.? Do Yaskawa supply the Sigma tuning software for free.?

I'm surprised the absolute encoders are cheaper as well, not as many controllers about that can accept Absolute positioning so I would have expected them to be more. Thou absolute systems and Bus/Ethercat drives/controllers are coming down in price and more popular so maybe this is the reason.?

Yes of course software is free. I'm not yet to the point of tuning but having read the manual multiple times I don't expect many issues. You start with the auto-tuning and fine-tune from there.

IMO rotary absolute encoders are not worth the hassle. Their only use in a router/mill context is to spare you the homing after power-up. And obviously the controller must be able to interpret the special Yaskawa protocol and I think no hobby controller can at this point. It could be done with some knowledge with LinuxCNC+Mesa.

[JAZZCNC]

Yes of course software is free.

No, it's not obvious really, it's not uncommon for high-end Servo manufacturers to charge for specialist tuning software if not using their own controllers just like it's not uncommon for PLC providers to charge for programming software.

I'm not yet to the point of tuning but having read the manual multiple times I don't expect many issues. You start with the auto-tuning and fine-tune from there.

I'd be interested in hearing how you get on with tuning. What controller and which method of control +/-10v or Step/Dir .?

IMO rotary absolute encoders are not worth the hassle.

What hassle.? Far less wiring and setting up far as I can tell, esp if using Bus type.! . . . I've not used any yet but soon will be doing so I'll let you know.

[JAZZCNC]

To me it seems like you favor steppers at for your own builds, but I may have got this wrong. If I am right, I guess you favor steppers for the simplicity and the above reasons?

Yes exactly for that reason! The machines I build mostly go to MEN in SHEDS type users or small business where the speed or high resolutions etc servos offer isn't needed. Reliability and simplicity are more desirable than speed for these users. Also from my point of view it also means less work and happy customers.

A good properly sized and tuned closed loop system is more than good enough for most builds and far simpler.

By the way what would be a good supplier for belts and pulleys? What supplier are you using?

I use a local supplier.

[jarjar]

I'd be interested in hearing how you get on with tuning. What controller and which method of control +/-10v or Step/Dir .?

Step/Dir is the most reliable and flexible. With analog +10V you're forced to control the servo in velocity mode.
With Step/Dir you can control it in position or velocity mode.
I'm also convinced position mode (letting the drive close all the loops internally) is the best way since the drive can do it much faster than the controller.

What hassle.? Far less wiring and setting up far as I can tell, esp if using Bus type.! . . . I've not used any yet but soon will be doing so I'll let you know.

The interface of the servo drive is a whole other topic and has nothing to do with absolute encoders. For absolute encoders you need to connect a battery and 2 extra wires to request the position data when using the standard analog/pulse-train drives. And as I said above, the controller needs to be able to read the position data for the specific servo manufacturer.

Now if you use EtherCAT drives, indeed you just need a RJ-45 cable between the drive and controller (but still need the battery). But you have 2 more problems:
- you need to actually find one of those rare EtherCat drives. Yaskawa for example doesn't sell them in the Chinese market. So you would need to go through an official reseller and pay $$$, or buy another brand: Omron / Panasonic / Mitsubishi / Delta / Estun...
- be prepared to spend weeks and write some C to make the controller talk to the drive (LinuxCNC in this case, the only one supporting EtherCAT).

[NordicCnc]

Yes exactly for that reason! The machines I build mostly go to MEN in SHEDS type users or small business where the speed or high resolutions etc servos offer isn't needed. Reliability and simplicity are more desirable than speed for these users. Also from my point of view it also means less work and happy customers.

A good properly sized and tuned closed loop system is more than good enough for most builds and far simpler.

Glad to hear this actually. Of course it is nice with speed and the non-existent noise! Who would not want 20m/min if they could have it. Anyway 10-15m/min will probably be enough for most small businesses and hobbyist (like me) as you said. I will let this thought about steppers sink for a while before I make any decision. The design is still not finalized anyway so no need to rush things.

You said you are using Lichuan 8Nm 230Vac for the build with the ATC spindle and that you use Lichuan closed loop system nowadays for your build. I checked aliexpress and they are all either 2 phase or 3 phase. So they work with a 3 phase output but uses only 230V? I did not find any 8Nm steppers, but I found this one: https://www.aliexpress.com/item/3279...52606833P16XMw

I use a local supplier.

Okay, I will research area more. I read somewhere that GT2 is supposed to be really good quality compared to the commonly used HTD 5.

[JAZZCNC]

Step/Dir is the most reliable and flexible. With analog +10V you're forced to control the servo in velocity mode.
With Step/Dir you can control it in position or velocity mode.
I'm also convinced position mode (letting the drive close all the loops internally) is the best way since the drive can do it much faster than the controller.

I agree about step/dir being simpler when drives handle position but not so sure that it's easier when your closing the loop back to the controller. Analog gives a little more control is my experience. But that said, my experience with servo's isn't massive and mostly with full closed-loop Analog controllers.

The interface of the servo drive is a whole other topic and has nothing to do with absolute encoders. For absolute encoders you need to connect a battery and 2 extra wires to request the position data when using the standard analog/pulse-train drives. And as I said above, the controller needs to be able to read the position data for the specific servo manufacturer.

Now if you use EtherCAT drives, indeed you just need a RJ-45 cable between the drive and controller (but still need the battery). But you have 2 more problems:
- you need to actually find one of those rare EtherCat drives. Yaskawa for example doesn't sell them in the Chinese market. So you would need to go through an official reseller and pay $$$, or buy another brand: Omron / Panasonic / Mitsubishi / Delta / Estun...
- be prepared to spend weeks and write some C to make the controller talk to the drive (LinuxCNC in this case, the only one supporting EtherCAT).

To be honest I think we are getting crossed wires a little here.! I know exactly what's involved with interfacing drives and wiring Absolute encoders etc.
My point was there's no more hassle using Absolute than incremental and there's actually less wiring required because you're not fitting home switches etc. Also if your using EtherCat or Bus type then there's even less wiring involved and less tuning involved because the controller and drives work closely together.

Also, I wasn't talking about Yaskawa Ethercat or trying to make Ethercat work with Linux or any other DIY controller. I'll be using an industrial Bus type controller with matched drives so i won't be needing to write any code etc.! . . . How long it takes to get up and running however will remain to be seen... .Lol

[JAZZCNC]

Glad to hear this actually. Of course it is nice with speed and the non-existent noise! Who would not want 20m/min if they could have it. Anyway 10-15m/min will probably be enough for most small businesses and hobbyist (like me) as you said. I will let this thought about steppers sink for a while before I make any decision. The design is still not finalized anyway so no need to rush things.

You can have 20mtr/min speeds with steppers if you choose the right drives/motors along with linear components, the 10x5 vertical machine I've been showing will happily whizz the gantry around at 30mtr/min.

You said you are using Lichuan 8Nm 230Vac for the build with the ATC spindle and that you use Lichuan closed loop system nowadays for your build. I checked aliexpress and they are all either 2 phase or 3 phase. So they work with a 3 phase output but uses only 230V? I did not find any 8Nm steppers, but I found this one: https://www.aliexpress.com/item/3279...52606833P16XMw

Yes, those are exactly what I'm using for Z-axis with 12Nm on X&Y. I think the heading is an error, if you look further down at the spec you'll see it says 8Nm.
This size would be way overkill for your little machine thou. The secret to the speed with steppers, in general, is the relationship between inductance and voltage, but even more so with large motors which is why Mains powered drives work best.

Okay, I will research area more. I read somewhere that GT2 is supposed to be really good quality compared to the commonly used HTD 5.

It's Got nothing to do with quality, that's down to manufacturer and spec. HTD, GT2, etc is the tooth profile. GT2 is actually a modified HTD tooth profile which supposedly offers better handling of torque and speed than HTD.

[jarjar]

My point was there's no more hassle using Absolute than incremental and there's actually less wiring required because you're not fitting home switches etc.

You still need a homing switch to setup the absolute encoder reference. The only difference is that you don't need to home on subsequent power ups.

[JAZZCNC]

You still need a homing switch to setup the absolute encoder reference. The only difference is that you don't need to home on subsequent power ups.

Been told doesn't need a physical switch the system just needs to be calibrated with encoder offset for the home position which can be obtained through software..! . . . I will find out soon enough.

[NordicCnc]

Some design progress has been made.

X-axis design



- L-shape gantry with 120x80 (heavy) Item profiles 1200mm long, with approximaterly 950mm travel length . The profiles for the L-shape are bolted together from the bottom. I plan to use aluminium plates as guides in the T-slots. Should these guides be fastened somehow or is it enough to insert them into the slots and shim away the clearance?
- High side gantry profiles with 120x80 (heavy) Item profiles, 160mm long. These are bolted to the bottom profile of the L-section, from the top of the bottom profile.
- 20mm aluminium gantry side plates, bolted to both the L-section and the high side gantry profiles. It is also bolted to the bottom plates, that will be mounted to the Y-axis bearing plates. The purpose of the bottom plates being separated from the Y-axis bearing plates, is to allow some adjustment later when squaring the gantry.
- 20mm HIWIN linear rails.
- 2010 ballscrew, 1150mm long.
- Delta 400W servo motor with a 2:1 reduction, to limit ballscrew RPM.



Y-axis design will continue from here!