[arturod]

Checkout the Ethernet Smooth Stepper Rigid tapping with a custom code:


You can also use a tapping head.

How much rigid tapping are you going to do? Do you have a fast spindle?

There are numerous options available; it ultimately depends on what works best.t for you.

Arturo Duncan
https://cnc4pc.com

[CallumRD]

So if I bailed on the synchronous rigid tapping requirement, would something like just a single an Ethernet smooth stepper board be able to support 3x Delta B3 servos with multi turn absolute encoders and a servo spindle motor? Would I need to be looking at any additional hardware to get up and running?

Edit: I have a 10k rpm spindle as most of my work is with smaller tools that can take advantage of the speed.

[joeavaerage]

Hi,
an Ethernet SmoothStepper really needs a matching breakout board. There are simpler and cheaper breakout boards like the C10, C11, C25 etc......but there is one, an MB3, which is
if you like the Rolls Royce. It has all three ports developed (51 IO's), with a good balance of inputs and outputs. The outputs intended for axis motors are differential allowing you to exploit
the best speed and resolution of your servos. It also has a good PWM to analogue output circuit for controlling a VFD or whatever. The current price is $196USD, so all those features are not free. Note that the ESS
plugs directly atop the board, no cables required.

https://cncroom.com/en/product/mb3/

My own breakout board is, if you like, my own take on an MB3. It too has a good PWM to analogue circuit and differential motor outputs. Note also how I have arranged the plugs such that each servo drive (up to six)
has its own dedicated socket and thereby considerably simplify the wiring. My own board has more inputs (31) and fewer outputs (20) than the MB3 (23 and 28 respectively). If I discount the time taken to
design and build it, it costs well less than an MB3, but if I counted my time, well......

As for the multiturn encoders on the B3...that is a question I am less sure on.

I use B2 servos which have a 160,000 count/rev incremental encoder. I do not feed the encoder back to the ESS and Mach4...I don't need it. I could do so but that would require a lot of inputs for all the encoder wires,
and inputs are precious so I don't waste them like that. The ESS reports to Mach4 the position of the servo(s), but does so open loop. That is to say if the ESS has commanded a servo to a given encoder count,
UNLESS the servo faults out 'following error' the the ESS assumes the servo got there. I've been using this for years without fault.

So strictly speaking you do not need to feedback the servo encoder to Mach4, it does not need it, it will display what the ESS tells it, or alternately the encoder if you insist on it.

Second thing to realise is that Mach4 is a buffered controller, like all Windows based CNC solutions. That is to say that Mach4 (the PC) issues trajectory data (Position, Velocity over Time, PVT) to the
ESS in 1ms time slices in a buffer. The ESS has a default buffer length of 180ms. Thus it could be that a movement commanded now, this instant, will not be executed for up to 180ms. Even if therefore the
encoder were hooked back to Mach4 there may be a small delay which renders that data out of date.

The last thing to understand is that Mach4 has two operating parts. One, the most important is the Trajectory Planner and Gcode Interpreter. When a program is running it digests the Gcode and issues the
required trajectory data (numerically) to the ESS motion controller. The second part of Mach4 is the Screen Script. It is what displays the DRO's, the tool path, all the buttons and everything else.
The two sections of Mach4 go back and forth between each other, usually seamlessly. At need however the Gcode Interpreter and Trajectory Planner asserts primacy, and the screen will not be updated.
The only time I see this is with really tiny circularly interpolated holes, where the screen does not update until the hole is done, and then it catches up. In this last circumstance it matters not whether the
encoders are hooked to Mach4 or not, Mach4 is that busy running the machine that it does not display them anyway!

Really what the multi-turn encoders do for you is so that you do not have to home the machine, not so much that you need to continuously feed back the encoders to Mach to keep it updated, the ESS does that already
without the need for feedback. What would be useful is that when you turn on the the machine that Mach4 can read the encoders and can then populate its machine coordinate registers. This can be done
using MODBUS. Mach4 has a good MODBUS module. It could very easiliy read the encoder registers in each servo drive and use that data to populate the machine coordinates. MODBUS is not hugely fast,
milliseconds, rather than micro seconds, so can only emulate Ethercat but it can do the same thing.

What I envisage is that when you turn the machine on, at the first run of the PLC script, Mach4 would read all three (or more) servo drive encoder registers and then use them to populate the machine coordinate
registers in Mach4. This would require some coding, I am not aware of any native support for that scenario in Mach4...but could be mistaken.

Given that my own servos do not have multi-turn encoders I have not explored it. I home the machine, but I never turn it off, unless I depower for some other purpose, maintenance for instance. Thus my machine remains
homed for weeks and sometimes months at a time without me ever having to refresh it. I did find about two days ago that my machine had some conflict with SoftLimits (which I have permanently engaged) which in turn
suggested a machine coordinate fault....so I rehomed the machine. Turns out the last time I did so was 56 days before!

The short answer to your question, is that yes an ESS and an appropriate breakout board will easily and simply handle three B3 servos. If you wish to exploit the multi-turn encoders such that you never
have to home the machine then using MODBUS that is entirely possible, but would add some coding complexity.

If you have an angular position capable spindle even rigid tapping is possible.

If you are trying to achieve a good result but simply, then Mach4/ESS/MB3 is a good approach. If you don't mind the extra complexity, and possibly cost, then LinuxCNC and Ethercat would be useful.

I myself have been using Mach4 and an ESS with various breakout boards, culminating in my own design board, for nine years....and its makes good parts. Really, making good parts is the only test of a
machine. Whether it has this feature or that is still secondary to 'Can it make this particular widget'.

Craig

[peteeng]

Hi Callum - Have you looked at the Kogna board from Dynamotion? Has a forum on cnczone here. Peter

https://www.cnczone.com/forums/dynom...kflop-kanalog/