[sn4k3]

Again great content!

Yes you can pull it high (or low) and thereby cause the servo drive to be permanently enabled. I intend (as I am still building my new mill) to have one Enable signal line (from my controller) that enables
all three drives. I also intend on having one Reset signal (from my controller) which connects to all three drives. Note that Delta drives tolerate an already active drive receiving a Reset signal. Some brands
a Reset to an already active drive would cause a fault.

You direct short them or have a diode per driver signal?
Also there's any advantage of having the controller enabling the servo opposed to permanently enabled? If yes what are the use cases?
About reset do you have a macro to trigger it or you tie it to any button? I'm thinking to activate reset once clicked "Override limits" button...

On manual it's not clear if Outputs+ are shorted to COM+ or VDD? For inputs+ they short with COM+ but outputs are not clear, do you know about it? I want to spare cables and nest

As far as various fault signalling schemes Delta drives are very flexible.

Lets start with limit signals. With most Mach systems, and I imagine similar to UCCNC, limit switches are connected directly to the controller. A limit event will cause the machine to stop.
Delta drives (and incidentally most other modern AC drives) offer another alternative. You can hook a ++limit switch and a --limit switch direct to two inputs on the drive. Should a limit
event occur the servo will stop, either a crash stop or controlled deceleration at your program choice, and also propagate a LimitEvent signal on one of the drive digital outputs
to signal to the controller that a limit event has occurred. If you are required to jog the servo back within limits the drive will know not to allow you to jog++ when the axis is already
exceeded the ++limit, but will allow you to jog--.

You might ask 'why would Delta bother to add complexity to their drive in the form of limit signals and logic when my controller can do it?'
The main reason is that Delta drives are capable of being used in a distributed motion control solution. With such a system you do not have a motion controller at all, your PC is the trajectory planner
and via EtherCat or CanOpen that data is signalled to each drive, and each drive is responsible for controlling its own motion. Thus each drive must be able to handle and control its own
limit and home events. EtherCat is a recent feature addition in Mach4 but is not available with UCCNC.

Technically we do not need the limit handling built in to the drive but I point it out that you may see some of the power and flexibility of the drive.
If you had limited input pins on your controller then having drive handle limits might be an advantage. Instead of two inputs (++ and --)being required
for each axis, you would require now only one, namely the LimitEvent signal. I know that most people combine ++ and -- limits to just one input but then you lose
the capability of determining which way to jog the axis. As the servos are so fast and powerful having a mechanism that absolutely prevents you from jogging in the
wrong direction seems worthwhile to me. I have not decided which method (drive handles limits and jogs verses the controller handling the limits and use a software interlock to prevent
erroneous jogs) to employ on my machine. Given that the drive handled signals are hardwired in the drive they would be less subject to programming error.

While i was exploring manual i notice that thing about limits, i will not use that feature but is always good to have, since we can ignore it it doesn't hurt having extra features and more the better
I'm in the group who short every limit sensor, in my case i have X-/+ & Y-/+ and will have Z+ soon, they are all common but in my machine is easy to know which limit was trigger with just a look, still that should never happen...
But in fact is not hard to tie them to driver, with a 3 pin F/M connector will run the sensor just in some minutes

Now with respect to other fault conditions, I refer you to 7-98 to 7-100 of the manual. There are a number of fault conditions which are combined to one output ALRM.
It is also possible to have fault events signalled separately on different pins. Your controller or control software could handle faults in different ways.
For instance you might decide that Overcurrent, Overvoltage, OverHeat should cause a machine wide EStop. You might decide that if you get a FollowingError fault
your machine software would execute a FeedHold event and thereby prevent loss of HomeReference that occurs with an EStop event.

This multilevel fault signalling is much more sophisticated than you or I as hobbyists are used to but are common in industrial practice.
I have yet to decide what level of sophistication of fault signalling I will adopt but I suspect I will opt for the solution above ie ALRM will cause an EStop
whereas a FollowingError fault will cause a FeedHold.

Programming these features without the use of Delta software is possible but problematic.

Yes that a big plus and in fact should be used even for us.
Regarding EStop, Stop Cycle & FeedHold i can't get a valid test... I did a test which show me that feedhold was acting like stop or estop, maybe only noticeble on high speeds/accels? When i press feed hold on UCCNC it immediately holds just like estop, it don't complete the line, for instance lets say a straight line from 0 to 100mm, if i feedhold on 50mm, it will hold as soon i click it, but as i understand feedhold as a safe thing that will hold on a safe position ie: start of next gcode line, stoping at middle of a line doesn't seen so safe, but well UCCNC does a really good job because i already try them all and both was able to continue without offseting the work. I also read somewhere people use cycle stop on uccnc instead of feedhold because of spindle and therebefore cycle stop = feedhold but turn off spindle.
Twice i also tried the extreme, after a long job i had to turn off at night, so i press RESET (estop), turn off cnc power and left motion controller and software on, on next day i resumed without home, i just cleared the estop and click cycle start, everything went ok, clean and without offsets (I'm am with luck?)

Now i tried 10000 mm acelarations but that shakes my machine and bench, that was a extreme to see how it behaves, i'm now with 800mm which i think its ok for my machine
I will order a second kit to compose the machine and get rid of steppers, these servos are a bless.

Here my adaptor and servo
https://i.imgur.com/qO6lTE6.jpg
https://i.imgur.com/Ggyzey6.jpg

[joeavaerage]

Hi,

You direct short them or have a diode per driver signal?

You can do either. Look very closely at the warning sticker that comes attached to the drive. It shows two ways that you can wire
to digital inputs, but one way is correct and the other will result in an overcurrent fault of the input photodiode and destroy it.
You need to understand that warning.

The attached pic is from 3-46 of the manual. It shows an open collector configuration to the pulse and direction inputs, but is in fact
representative of all opto-isolated digital inputs. Note that it indicates the use of the available auxillary 24VDC suply (500mA max) within
the drive for signalling purposes. The 24VDC is applied to pin 35 via a link from the 24VDC output pin 17. Note how I have highlighted
the built in 1kOhm current limit resistor in the circuit to each of the photodiodes. Thus if you shorted pins 37 or pin 41 to COM-, pin 45
the current through the photodiodes would cause both inputs to the drive to be logic high but the current through each photodiode is limited by
the 1kOhm and the two 51Ohm resistors. Clearly the 1kOhm resistor effectively sets the current and limits it to a safe value.

The second pic also from the same page as the manual shows that you can hook directly to the same input photodiodes through
pins 36 and 43 but these DO NOT INCLUDE the 1kOhm resitance, just the two 51 Ohm resistors. If you attempt to feed 24VDC through
the photodiodes WITHOUT the 1kOhm limiting resistor you will blow the photodiode. That is what the sticker is about.
Note that you could apply 5VDC to this wiring setup, the two 51Ohm resistors provide sufficient current limiting for that.

My inclination is to use the included 1kOhm resistor as that means I can't accidentally damage the photdiode input. The open collector
wiring option limits the signalling speed to about 200kHz whereas if you use a differential line driver of 5V you can get to full 5000kHz.
In practice I've found I can get to closer to 250kHz with open collector and in any event 200kHz is more than enough for the resolution
I require.

Also there's any advantage of having the controller enabling the servo opposed to permanently enabled?

I believe it is better to have a controlled enable signal, it would be available as a saftey for instance that would prevent unexpected movement
from your machine, say, when you had the doors open. Depending on your machine you might consider that extra safety precaution unwarranted
in which case by all means permanently short the Enable signal to earth.

I'm in the group who short every limit sensor, in my case i have X-/+ & Y-/+ and will have Z+ soon, they are all common but in my machine is easy to know which limit was trigger with just a look, still that should never happen...

That may be your choice but I refuse to countenance it. In industrial practice a machine would be determined to be unsafe if an operator could
jog a machine out-of-bounds beacuse the limits had been combined in such a manner that the control software could not prevent it.
Such a machine would contravene employment safety law here....so why would I consider it for home use.....after all I know who would get hurt!
My new build mill will weigh over 600kg and have axes powered by 1hp servos. It may have been acceptable for my mini-mill to have combined
limits but with a bigger and much more powerful machine such a shortcut is foolhardy. I have the required inputs and the control software is well suited
to the task....why would I not use it to make a safe machine?

In Mach, and I suspect UCCNC, although I cant be sure in UCCNC, an Estop or a StopCycle causes an imediate stop which aborts motion data
in the buffer, which results in loss of home reference. FeedHold on the otherhand allows the motion buffer to drain before the machine stops.
My motion controller has a default motion buffer length of 180ms. Thus if I hit <FeedHold> I could potentialy get another 180ms of movement
before the machine stopped. In most cases you have to look very hard and critcally to determine the difference between Estop and FeedHold.

Just as a matter of interest lets imagine your machine is idle at a given moment, that is not moving at all. If you hit <Estop> the machine stops
imediately but as there was no motion data in the buffer then there is no loss when the buffer is aborted. Then if you re-enabled the machine
it would maintain accurate home reference. Note some control software (Mach4 certainly) allows you to set what happens to Home Refernce status
after an Estop or Stop event. To be 100% you should in fact re-home atfer an Estop, although I have my machine just to resume the current machine
coordinates. It relies on me to determine whether the Estop could potenially have caused a loss of reference. Likewise in Mach4 you can determine
what happens to the spindle on an Estop, CycleStop and Feedhold.

Twice i also tried the extreme, after a long job i had to turn off at night, so i press RESET (estop), turn off cnc power and left motion controller and software on, on next day i resumed without home, i just cleared the estop and click cycle start, everything went ok, clean and without offsets (I'm am with luck?)

I suspect that at the time you hit <Reset> your machine was stationary and therefore no motion data was lost and then, yes, it is permissable to resume
without re-homing. If you were in an industrial environment would you want your employees to just start using a machine without confirming its home
reference?. If they get it wrong they can crash a machine. Indeed many industrial machines have absolute encoders such that the machine can be powered
up and it retains reference without a homing procedure, and machine owners pay a very considerable price tag for that added security.

Craig