Power outage - how do you deal with it?
-
Yet another power outage (5 hours into a 9 hour print) has me thinking about doing something to save a print from a power outage. The most obvious answer is a UPS but if you want to keep the print attached to the build plate then you need to keep that heater going which means a VERY substantial UPS (1200W build plate heater)
I understand that one can save the position of everything and power down during an outage but how realistic is that? If you loose power to the motors, you loose positions which would mean the printer needs to be re-homed when the power comes back on but how do you re-home if you have a big print on the build plate?
What is your strategy (other than to start the print from scratch)? -
@jens55 said in Power outage - how do you deal with it?:
VERY substantial UPS (1200W build plate heater)
Only if you need it during heatup. Steady state heating is quite minimal. Get a Kill-a-watt and see how much draw you actually have. You may be surprised.
-
@jens55 said in Power outage - how do you deal with it?:
how do you re-home if you have a big print on the build plate?
Manually. Identify the exact layer it failed on. Then you know the Z position. Touch the nozzle to the print by jogging and manually set the Z position. Z max endstop can work to, but must be accurately calibrated.
-
@Phaedrux, you are of course correct in that a heated bed only requires a fraction of the power if it is already at operating temperature. The problem with that approach is that the UPS must still supply the rated power for the heater. While the average power is much reduced, you can't use a 800VA UPS to power a 1200W heater because whenever the heater gets cycled, the UPS must supply the full 1200W which an 800W UPS will not be capable of.
The reduction in average power used will mean a longer UPS battery life but you still need a UPS that is spec'd to be able to supply full power to the heater. -
@Phaedrux, am I correct in assuming that this method of recovery is only viable if the print has not lost adhesion to the build plate?
-
If the print has detached, there's nothing to resume!
-
@jens55 said in Power outage - how do you deal with it?:
If you loose power to the motors, you loose positions which would mean the printer needs to be re-homed when the power comes back on but how do you re-home if you have a big print on the build plate?
On a Cartesian or CoreXY printer you can normally home X and Y even with a print on the bed. With Z there are a few options:
- Have a Z max homing switch to deal with this situation.
- Assume that the Z position won't have changed since the power failure. In practice it is likely to have changed by up to +/- 2 full steps or possibly dropped as much as 3 full steps, however if your Z steps/mm is high then that may not matter.
- Raise Z a few mm, clean the nozzle, then manually lower Z until it's at the same height as that part of the top layer that was printed before the power failure; and use G92 to set the current height to be the same as the resume height.
You need to set up your resurrect-prologue.g file to do whichever of these (or something else) you choose.
-
@jens55 one option would be if it is running off a ups, use an input to run a macro or something of the like to limit the max power for the heater to say 50%. If you needed to change the PID parameters you could also do that I guess (not sure how the PID algorithm would like changing those parameters on the fly)
-
Thanks for all the input!
I have come to the conclusion that for my situation it is not cost effective to implement a UPS solution. Trying to recover a partial print also seems too involved for me since it doesn't happen often enough for me to become comfortable / proficient with the whole procedure.
In addition to the above, when we have a power outage, it seems that probably half of the power failures are beyond recovery anyway as the print will have lost adhesion due to the cooling build plate. -
What is the print surface on the bed plate and what material are you printing?
I have found that PEI seems to hold on to prints (ABS, PETG, and TPU, not sure about PLA because I don't use it very often) even if the bed cools. I usually have to work a bit to get prints off the bed when they are finished, which usually happens hours after the bed has cooled to room temperature. Putting a few drops of IPA at the base of the print usually helps it release from the bed.
-
@mrehorstdmd, I use a glass surface and print mostly PETG and some PLA. I try to set the conditions (bed temperature, squish factor) so that the print releases when the bed cools down just a bit. It doesn't have to go to ambient - just 30 or so degrees drop and the print should (ideally) release. Yes, PETG can attach itself to glass in an almost permanent way causing chunks of glass to break out. I try to avoid that kind of adhesion.
I have not had much luck with a PEI build surface in the past although I have not tried that in a long time. My main problem was the failure of the glue sheet between the glass bed and the PEI sheet (PEI curling up at the edges when the bed temperature is fairly high)). I have never tried a metal surface that is coated with PEI. It likely would not have that issue. -
@jens55 I use a 24V DC UPS with two very small (and cheap) 7Ah batteries wired in series to give me 24V. My bed is 400mm X 400mm X 10mm thick with 12mm of insulation underneath it. So it takes an awfully long time to cool. Lights and other non essential items are powered from a separate 12V PSU. So when power fails, the 24V to the boards is maintained and derived from the batteries but the bed is not heated and any non essential things like lights turn off. The UPS has relay outputs, the contacts of which are connected as triggers on the Duet board. I use these to set the bed heater fault detection parameters to silly high numbers when power is lost and restore them to sensible numbers when power is restored. I could also slow down the print which might prolonged battery power for longer.
I haven't tested these latest batteries but estimate that, just to maintain the temperature of the hot end, run the hot end cooling fan and a couple of steppers, it should keep the printer going for at least an hour. That suits the majority of power outages we get in this part of the world. Of course, bigger batteries would last even longer. -
@deckingman, that sounds like a very workable solution. I gather you have a removable build surface so you don't have to wait the hour or two for your bed to cool down in order to remove a print ....
Two of my printers are bed slingers so I can't increase the bed mass too much but insulation would help the bed keep it's heat.
One thing that just crossed my mind .... . As has been pointed out, average power for the bed, once up to operating temperature, is much lower than rated power. With the slow switching frequency of the Duet and the fact that the heater is a resistive load, I believe the peak current demand is still very high even though average power output is reduced to a large degree. Since the UPS has to be rated for the peak current that the heater draws, it would need to be a high power UPS. If one could somehow limit the current to the heater, one should be able to get away with a much lower power consumer grade UPS. I wonder if such a current limiting device is available off the shelf.
I will have to look into that. -
@jens55 they make them up to 40A so you will be more than ok. Phoenix Contact, Schneider electric and Mean well are all ones I have played with. Not all go that high however
-
Schneider Part Numbers:
Phoenix Contact
https://www.phoenixcontact.com/en-nz/products/uninterruptible-power-supplies-ups/dc-upsThere are too many to list them all
-
@jens55 You are correct that I use a removable build surface (6mm float glass in my case).
If you get a DC UPS with additional relay operated contacts (my Meanwell one does, others probably do too) then you could use the contacts that change state when power fails as an external trigger as I do. In my case, the bed power is mains so cuts when power is lost. In your case you could use the trigger macro to turn off the bed heater (or set it to zero) and another trigger macro that will restore it when power resumes.
If you want to power a DC bed during power outages, then your biggest outlay will likely be the batteries needed - much more than the UPS (at leat they are here in the UK).
If you do it properly, and want to protect batteries from over discharging, as well as keeping them in balance if you have more than one, then it can get a bit more complicated. Funnily enough I documented all this stuff in part 8 of my printer rebuild which you can see here https://www.youtube.com/watch?v=74u2AluoZgo&t=1s
-
@jens55 I have found that the 468MP adhesive that holds the PEI on the aluminum usually lets go after 2-3 years of heavy use, including spending a lot of time at 100C for printing ABS. When it lets go most of the adhesive stays stuck to the aluminum.
Keenovo sells heaters with the 468MP adhesive attached. Their manual says to seal the edges of the heater with silicone after it's been mounted on the plate. I'm about to replace the 468MP on my printer's PEI/bed and will try sealing it that way to see if it lasts longer. We'll know in a few years.
-
@mrehorstdmd, sealing the heat pad to the aluminum with RTV is my standard procedure now although I do not know if it makes any difference. Call it a 'belt and suspenders' installation
I am not sure how you could 'goop' on the RTV on a PEI print surface. Maybe only put the RTV to the edge of the PEI material but I kinda doubt that this would make any substantial difference.
The PEI material I have used was maybe 2mm thick which could possibly build up internal stresses enough to lift the edges. I have not tried to go with a much thinner sheet to see if the same thing happens. It wasn't worth the risk of wasting more money. In my case, I ran into curling almost immediately. -
@jens55 I use a 0.7 mm thick sheet of PEI. It covers the bed plate right to the edges, as does the heater. I'm going to seal the edge with silicone once I get the PEI reglued. The only time I've had curling of the PEI is when the adhesive let go a couple years ago (a large print pulls up the PEI) and again very recently, hence the regluing that is about to happen.
When the 468MP fails on a heater you get air bubbles between the heater and plate. That results in poor heat transfer from the heater to the plate and burns the heater. After the last failure I mounted a new heater using high temperature silicone instead of the 468MP. It's been on there for at least 3 years with no issues. Keenovo sells heaters with and without adhesive.
The Rino 30:1 worm gear reducer I use in the Z axis keeps the bed from moving when power is lost. In theory, it can resume a print after power is interrupted. I'll have to put a gauge on the printer and see how much the bed jumps when power is restored. The bed moves 20um/full step of the motor, so I suspect the jump will be small. It might produce a visible discontinuity in the print (one layer too thick or too thin) but the print should finish. I usually print in 0.5mm layers with a 1 mm nozzle, so a few microns jump won't matter much. If I were printing in 100 um layers it would certainly be more of an issue.
