Anyone seen this?

PCW wrote:

The LAN9254 can do 32 GPIO by itself (basically the same price as LAN9252/LAN9253)
 

Good call, it's literally about $1 more than the 52. I didn't really do the shopping As I don't need the baked in GPIO for the current thing.

 

I might add that building a custom Cia402 homing module would allow you to home some drives to a hard stop so no I/O would be required for limits. But most drives have some inputs for homing/limits and an output for an alarm. So its really just the encoder needed until you add a tool changer or something.... Plan out what an average machine needs. You might be surprised how little I/O is required.

rodw wrote:

I might add that building a custom Cia402 homing module would allow you to home some drives to a hard stop so no I/O would be required for limits. But most drives have some inputs for homing/limits and an output for an alarm. So its really just the encoder needed until you add a tool changer or something.... Plan out what an average machine needs. You might be surprised how little I/O is required.
 

Eh....sensorless homing has always bothered me...

I've noticed the same about EtherCAT drives having IO ports. You are right about a full etherCAT machine not really needing much separate IO except for the fact that EtherCAT drives are still really expensive. I think that's where your encoder/pulse/io modules come in, so you can still do everything over EtherCAT from the host. Your basically doing the Mesa card's job with some EtherCAT stuff at that point so I'm not 100% certain that there's a big reason to do it over just using a Mesa card. I'll have to figure out what kind of counting/pulse frequencies you can expect from a high end microcontroller vs what the FPGA cards can do to see if it's even worth talking about.

I think you use ethercat or you don't. Even the Rtelligent ECT60 drive can be configured with min limit, max limit and home using internal Inputs if you don't want to use sensorless. It also has two outputs, usually for alarm (we should use) and brake (rarely used) but can be used as a generic output. I suspect sensorless is more repeatable.

Whether you use Mesa or Ethercat is a matter for personal preference but if you are building a machine for resale customers will expect to see ethercat. The added cost may well be offset by reduced labour in the wiring.

rodw wrote:

I think you use ethercat or you don't.
 

That's a fair assessment.

Whether you use Mesa or Ethercat is a matter for personal preference but if you are building a machine for resale customers will expect to see ethercat. The added cost may well be offset by reduced labour in the wiring. 

I do agree with what you're saying here, but the machine OEM/professional crowd can keep buying what they're buying. I don't build CNC machines for sale...not in the traditional sense. Then again I'm not so sure it looks all that professional when you open a cabinet and see a bunch of Chinesium....EtherCAT or not. I'm sure it's fine but I don't understand why they always use that font lol....there's something about that font that says "stay away"

In drive ethercat is definitely the way to go.

Servo drives: costs is $40-60 per drive for ethercat (vs pulse input variants) across multiple brands.
You then don't need a stepgen card. Decent stepgen card is $120-150.
So cost difference is minimal. Ethercat in drive gives you a lot more and is a lot easier than having a stepgen card.
For a new system, there would be no role for ethercat -> stepgen card -> servo/stepper.

But there is a case where converting an existing step/dir setup would be nice. I have a small lathe with a 750W step/dir servo for the spindle. It's good quality, not a cheap servo. The axis are cheap stepper driven. I'd like to change to ethercat servos for the axis. But then I am stuck with what to do for the spindle which is step/dir.