Large input cap for Duet 3 6HC

[Nxt-1]

For a while now, I have been refitting my delta with a Duet 3 6HC and just when I was about done with all the new wiring an interesting thread poped up on the forum asking about the input voltage of the duet.

In short: the duet's components have a 38V V_in max recommended rating but generally 32V is used as max to build a buffer for the back-emf from decelerating motors. It was suggested by @dc42 that a large input capacitance could act as an energy buffer for the energy comming from the motors, thus limiting the V_in voltage rise.

I got myself a 10mF 100V cap to install on my machine but would like to verify some things first:

1. Inrush current: 10mF is a considerable capacitance and could, depending on the input resistance of the Duet, cause a very high inrush current. Is there any ball-park figure about the Duet's input resistance?

2. Charge bleeding: Large capacitances always get my safety senses all tingly. Idially I would have a power resistor in parallel with the cap to bleed its charge down when not in use. On the other side, the large capacitance does offer a buffer of energy for the Duet power loss detection system in the event of a power failure. A bleederw would ofcourse counteract this.

3. Mounting: The cap measures 40x60mm (d*h) so I can't easily fit it very close to the Duet's terminals. Does wirelength play a role here?

4. Just to make sure, I have 4 separate PSU's in my machine for all the voltages I need. All their grounds are tied together (and to PE as well). At this moment only the ~32V psu is connected to the Duet at this moment (maybe the 5V will be connected later as well). I assume them sharing ground does not cause an issue here?

[dc42]

@Nxt-1

1. I suggest you connect the capacitor across a spare pair of output terminals of the 32V PSU, then the Duet will not see the inrush current.
2. The current drawn by the Duet to feed the MCU and other components will probably be sufficient to bleed the capacitor, especially if you also have a PanelDue connected to it. You can use the blue LED on the Duet to judge the state of charge of the capacitor.
3. Wire length does not play a big part here because the higher-frequency components of the current from the motors when they are turned off will be absorbed by the on-board capacitors.
4. Shared ground is the correct connection.

[Nxt-1]

@dc42 I still am kind of concerned about the inrush current so I figured a NTC in series should be effective as a current limiter. Admittedly I have never specced an NTC before so I'd love your feedback.
The PSU I am using has a maximum output current of 4A. At 32V that would give me a R_25C of 8Ohms. And the 10mF cap at 32V would store 5.2J of energy which seems fine for every NTC I've found. In reallity the steady state current will be quite a bit lower but with future profing in mind, I took the max psu current as current requirement.

This led me to B57364S0100M000, a 10 Ohm, 7.5A NTC. Would this be a suitable part to place in series with the posive leg of the cap?

[rjenkinsgb]

@nxt-1 said in Large input cap for Duet 3 6HC:

I still am kind of concerned about the inrush current so I figured a NTC in series should be effective as a current limiter.

It could just possibly limit charge current, if that lasted long enough to heat it up - but once the voltage stabilises, the current to / from that cap will be minimal and it will revert to high resistance.

You could just use a low value resistor to limit the current to under the PSU maximum; it should be reasonably effecting overall.

Or run a 24V relay via a series resistor from the big cap, and get it set up so it only pulls in at near 30V, then connects a far lower value resistor across a mid value soft start resistor.

And don't forget that if the PSU is permanently connected to the Duet, its output smoothing caps are added to the Duet onboard caps, so need allowing for in any motor current dumping calculations.

An old Meanwell 10A 24V I have kicking about has 4000uF on its output.

Also, the PSU is only 4A, the motor current must be lower still, so the energy returned on shutdown should only be a fraction of that in DC42's calculations, as those were for 24A motor load; 6 x 4A motors.