Testing Power Failure Feature

A Former User

That brings me back to exactly what kind of hardware is doing the power loss detection, and what kind of power supply is in use? It may well be the supply just doesn't have enough power after the fact.

Phaedrux

12v or 24v?

FairladyZ

@Phaedrux 24v
@bearer That is most likely the case. I have decided that since it saves the restart data and I am able to restart the print as long as the print head being close to the print does not ruin anything, I will leave this project until later. I think I should look into a different power supply but at this point I am fairly happy with how it works currently.

A Former User

Adding capacitors to the existing supply could help, but how effective it wold be would entirely depend on how the power loss is detected in the first place.

dc42

@bearer, the Duet detects power loss on-board. The power loss threshold voltage is set in the M911 command. using 24V power, you would typically set it to about 22V.

The heaters are automatically turned off when the voltage dips below the threshold set in M911. Typically, the power fail script should use M913 to reduce power to motors other than Z and E, to help save enough power to raise Z and retract filament. See https://duet3d.dozuki.com/Wiki/Setting_up_to_resume_a_print_after_a_power_failure.

nophead

@dc42 Is there a way to configure another input? I normally use an opto coupler to detect mains zero crossings to get an early warning of mains failure.

dc42

@nophead said in Testing Power Failure Feature:

@dc42 Is there a way to configure another input? I normally use an opto coupler to detect mains zero crossings to get an early warning of mains failure.

Not yet, although I plan to add one.

A Former User

But an arbitrary input assigned to a trigger could get the job done surely, and detecting mains loss instead of voltage sag would make adding capacitance more effective, if needed at all.

dc42

@bearer said in Testing Power Failure Feature:

But an arbitrary input assigned to a trigger could get the job done surely...

It wouldn't provide the same near-instant execution that the built-in power failure code does.

A Former User

Interesting, hopefully configurable inputs will be good to go by the time i get around to adding power loss stuff.

dc42

If we assume that the power draw of the machine after the firmware turns off the heaters and XY motors is 10W (basically the power needed to drive the Z and extruder motors, plus a little for the Duet itself), then adding 10000uF capacitance on the 24V power rail should provide 190ms additional time. On a delta where you can't turn any motors off, it should still provide 100ms extra time; and on a delta the Z move will be fast anyway.

How much extra time would be provided by an external power fail input will depend on the power supply and the nature of the power fail detection.

nophead

@dc42 It is far more effective to use the energy stored in the main reservoir capacitor in the PSU that is charged to 340-400V because E = CV^2, and it can usually drop a long way before the PSU output falls at all. That is why I like to look for two missing mains half cycles when doing power fail detection. On a decent PSU it can give you seconds. I haven't done it on a 3D printer yet though.

A Former User

I was thinking the same
(but thinking usually never leads to anything good )

dc42

@nophead said in Testing Power Failure Feature:

@dc42 It is far more effective to use the energy stored in the main reservoir capacitor in the PSU that is charged to 340-400V because E = CV^2, and it can usually drop a long way before the PSU output falls at all.

That depends on the PSU. If it has a single input accepting 110V to 240V without an input voltage selector switch, that will be true when you run the PSU on 220-240V mains, but not on 110V.

The half CV^2 applies to adding an output capacitor too. The stepper drivers can work down to less than 10V. If the power failure is detected at 22V, then by the time the voltage has dropped to 10V, 79% of the energy has been extracted from the capacitor.

nophead

@dc42 Free range PSUs normally use a PFC boost converter, so the reservoir is charged to the same voltage regardless of the input voltage. Otherwise the main down converter would need run at half the voltage and twice the current necessitating bigger heatsinks, etc.

A standard compliant PSU needs active PFC at these powers anyway, so free range input comes for free as it already has a boost converter to allow it to take sinusoidal current over the full cycle rather than just at the mains peaks.

Cheap PSUs just have a switch and voltage doubler. I haven't come across wide input range PSUs of any size that don't use one of these two techniques, have you?

A Former User

I think the CE regulative says that today any supply over some 40-50w needs to have active power factor correction, so pretty much every recent supply is universal input with the boost converter (at least in my neck of the woods). Even most supplies below the threshold has it. There, done beating dead horses for today:)

dc42

@nophead said in Testing Power Failure Feature:

@dc42 Free range PSUs normally use a PFC boost converter, so the reservoir is charged to the same voltage regardless of the input voltage. Otherwise the main down converter would need run at half the voltage and twice the current necessitating bigger heatsinks, etc.

In that case, surely the second converter would be designed to work over a limited range of output voltages from the first converter? In which case, the fraction of energy you can extract from the reservoir capacitor at the output of the first converter would be lower.

I'm not saying that detecting mains failure earlier than the unaided Duet can do it won't buy you additional time. It will obviously do that - especially if the PSU is a large one because it normally powers the bed heater. But for most users - especially those who don't want to get involved in mains wiring - adding an output capacitor is simpler and should give predictable performance.

nophead

@dc42 Yes the second converter sees a constant input voltage around 400V, so can have its heatsinks dimensioned accordingly. During powerdown heatsinking isn't an issue though, so it can potentially run down to quite a low value. It might not even have an under voltage lockout, just a current limit, so if the load is light it could run down to a very low voltage.

Yes a large output capacitor is simple enough but I don't use 24V on most of my machines.

dc42

@nophead said in Testing Power Failure Feature:

Yes a large output capacitor is simple enough but I don't use 24V on most of my machines.

I see, that's why you want to make use of the energy stored in the PSU reservoir capacitors. IMO no printer should be designed to use 12V power these days. Modern Trinamic drivers generally perform better on 24V, which also allows higher motor speeds and thinner wires to the heaters. Even the budget Ender 3 printer uses 24V power. The only reasonable justification for using 12V IMO is if you want to use a low-cost ATX PSU in a home-build printer design because you want to avoid doing any mains wiring.

nophead

@dc42 I power my bed from the mains and I have lots of 12V PSUs that I got for free. I do have several 24V PSUs and 36V and 48V but they are much bigger. I am trying to make a compact machine.

I do use 24V on my machines that have ball screws but with belt drive my motors go as fast as I would want them to on 12V. I easily get 200mm/s out of a Melzi on 12V with Marlin.

Are you saying the Trinamic drivers won't work as well at 12V or just that they will go even faster on 24V?