Power supply advice please



  • I'm looking to source a power supply for my "build in progress" printer. It'll be 24v. The bed heater will be 240V AC via a SSR so no need to worry about that. I'll have potentially 8 stepper motors ( 1 for X, 1 for Y, 3 for Z, and 3 extruders feeding a diamond hot end). So, my simple layman terms logic says 8 steppers potentially all drawing 2 amps = 16 Amps (unlikely but possible?) Then say a couple of amps for the hot end heater (40W) and a couple more amps for fans and so forth making a minimum of 20Amps. With a bit in reserve I was thinking along the lines of 25Amps (600W) power supply. Is that reasonable? Too big? Too small? Thanks


  • administrators

    Too big. With 24V power, the current taken by the stepper motors will be much lower. For example, if the stepper motors need 2A each @ 4V when idle, to a first approximation the current drawn by the drivers will be 4/24 times 2A each. It will be a little higher than that because the drivers are not 100% efficient, and the voltage needed for a motor will be higher when an axis is accelerating or an extruder is pushing filament. But still very much less than 16A. If the Diamond has only one hot end heater then 10A @ 24V is probably sufficient. My delta uses a 24V 5A power supply but it has only 4 stepper motors and one 30W heater.

    EDIT: removed incorrect 'times 1.414' from the stepper motor PSU current computation.



  • Thank you. I owe you a beer or 3 for the money you keep saving me.



  • Respectively question the numbers David after all the stepper motor constant current discussion.
    Based on that discussion the voltage applied across the stepper = V supply ie 24v and the current rises to the stepper drive Imax setting (lets say 2A) in time x.
    In that case the peak power = 24v x 2A divided by stepper driver efficiency (say 95%)
    So peak input power becomes 48/.95 = 50.5W

    Now i'm speaking and dealing only with the peak value as this is important when using switch mode power supplies as they must be able to deliver the peak power (ie current) required. Cheap ones will often go into current limit or even turn off output at 105% overload, better ones will deal with 110%.

    Now the the steady state numbers could be lower as I understand you reduce current to the steppers when idle.

    I would also advocate the 80% rule in which you allow 20% overhead at all times on your power supply.

    When my DuetWiFi board arrives i'll wire in a voltage/current/power monitor between the SMPS and the Duet to see if I can get some real world readings. acknowledging that the sample time on it will not capture the peak draw.



  • Aussiephil - the on time for these stepper drivers is very short - think microseconds. The time the FETs are on is not nearly long enough to trigger SMPS overload detection.

    Proper power consumption for steppers would be:

    (rated voltage * drive current) / efficiency

    so:

    2.7*2 = 5.89w -> 56w from the 8 steppers, 40w from the hotend heater, probably do a 150w or 200w PSU.



  • See all the big capacitors on the board next to the driver chips? Those are why you don't need to size the PSU for the full on-phase current draw. The bulk capacitance smooths it out pretty well during the extremely brief high-power-draw transients like startup and attempting to accelerate against a stall. Then the PSU has a bunch more capacitors. The stepper driver PWM is high enough frequency compared to the total available capacitance that you only need to consider the AVERAGE power draw, as elmoret and dc42 have noted.

    The heatbed is very different – the 10 hz switching frequency is so slow, and heatbed loads are so large, that you absolutely need to size the PSU and wiring+connectors for peak current draw at 100% power.



  • I'm going to bow out until I do some real world measurements as none of the figures add up to me. Whilst I haven't worked with steppers and stepper drivers, I've done a lot in SMPS field and in constant current drive of LED's so not a stranger to power supply theory and practice.



  • It's an inductive load being push/coast/pull PWM'd through an H-bridge at something like 30khz. The motors can push power back into the V(in) rail during part of the PWM cycle and when decelerating from high speeds. How do you factor that into your PSU sizing?

    You often have weird shit going on like one coil giving energy to the other coil within the same motor. Not to mention the complexity of the phase relationship between back-emf and current due to load angle, and how that affects power draw. Or the power factor issues that arise at high speeds when the motor is inductance-limited so drive voltage and current go out of phase.

    It's goofy as hell from an energy balance standpoint. Unless you REALLY want to get into numerical simulation or o-scope measurements, all you can do is work in averages.

    • The bulk capacitance largely covers the transients during the PWM cycle, and you have a minimum of 8 coils all in different push/coast/pull PWM phases that will partially cancel out each other, so you can pretty much ignore the PWM and just look at the coil current and how much AVERAGE voltage is required to sustain that current. That's the nominal power per motor.

    • Motors use more power than nominal when accelerating, and use less (or even give back) power when decelerating. The best way to figure this out is the conservation of energy… you're changing the kinetic energy of the moving mass, and that change (plus/minus some inefficiency losses) is the difference in the power the motor draws compared to baseline. (You can also calculate this via load angle and back-emf, but that's a bitch.) Nobody ever bothers calculating this though, because it usually isn't very big for 3D printer style machinery. And most accel/decel work occurs within ~0.1-0.2 sec of a roughly equal-and-opposite accel/decel period, so this also somewhat cancels out at the time scale that over-current protection functions at.

    • Driver heat losses and iron losses from attenuating the high-frequency current waveform components are pretty easy to capture with a lumped efficiency fudge factor. If you want to get rigorous, you can look at the RDS(on) of the driver H-bridge, but we're really not talking about a lot of power here. Just add some safety factor and move on with your life.


  • administrators

    As Ryan and others have said, the switching frequency of the drivers is around 30kHz. That's a cycle time of about 33us. If the motor needs 4V and you use a 24V supply, then the on-time of the motors will only be a little higher than 4/24 * 33 = 5.5us at standstill. Meanwhile, your SMPS also probably has a switching frequency around 30kHz, so the feedback time constant in the SMPS must be a lot greater than 33us to prevent the SMPS being confused by its own ripple voltage. So the SMPS won't see the peaks at all, even if all the stepper drivers happen to take their peak current in sync, and will respond to the average stepper driver current.



  • i think i am having power issues i have a 12v 25amp PSU i turn bed and hot end on and can here my hot end fan slow down just a little
    now i tuned my hot end with just hot end turned on so now i get error heating to slow if i try and heat the bed and hotend at the same time
    so i believe my PSU is two week would you agree
    also if it is i read some where about running two PSU 1 to run the heated bed and anther to run the duetwifi is it safe and a viable option are am i just splitting hairs and need to re-tune my machine with everything turned on


  • administrators

    I run a dual extrusion printer with an approx. 120W bed heater from a cheap 12V 25A LED PSU, with no problems. The fan slow down when the bed turns on is barely noticeable. So I suspect that either you have too much resistance in the wires and connections to the power supply, or it is indeed too weak - perhaps because you have a more powerful bed heater than I do.

    Try measuring the voltage at both the power supply output terminals and on the Duet VIN terminals, with and without the bed heater turned on, to see if the power supply voltage really is dropping significantly or the problem lies in the wiring.



  • do you know i did not even consider the amount of power draw from my heater bed i got a 12V 280W 300x300 SILICONE HEATED BED
    so i would need a12v 300w 15 amp psu to run the bed alone

    and in the mean time can i reduce the power consumption of my bed and wait longer for it to heat up



  • i just looked and i got a 300w 25amp 12v pwu



  • anyone like to recomend a good 12v psu needs to run 280w hotbed e3d hotend 2 fans and 5 motors running at 800mw each


  • administrators

    See https://duet3d.com/wiki/Choosing_the_power_supply. But consider whether an upgrade to 24V would be beneficial, see https://duet3d.com/wiki/Choosing_stepper_motors#How_to_work_out_the_power_supply_voltage_you_need. Of course, if you do switch to 24V then you also need 24V heaters and fans.


  • administrators

    @cap:

    do you know i did not even consider the amount of power draw from my heater bed i got a 12V 280W 300x300 SILICONE HEATED BED
    so i would need a12v 300w 15 amp psu to run the bed alone

    and in the mean time can i reduce the power consumption of my bed and wait longer for it to heat up

    You could turn down the power supply voltage a little if it has a voltage adjustment potentiometer.



  • well if my math is right i need 350w now add 20% extra 470w that was rounded up never down
    i want to stick with 12 volt i no it means things take longer to heat up but i have a stock pile of fans heater cartage

    so if i get a 500w PSU that is going to out put about 40amps is that not going to kill the duetwifi board
    i mean thats some power supply also what size wire do i need for that


  • administrators

    Yes 450 or 500W sounds about right for your printer.

    Can you confirm that you are using a SSR to control the bed heater? The Duet WiFi is rated at 15A bed heater current, but your heater will draw 23.3A.



  • no im not whats a ssr

    edit

    ok i did say im no engineer but i do love this guy i think if you sent this guy a board asking him to do full review and setup blog it would help your sale to low tech peeps like me and boost your sales
    https://www.youtube.com/watch?v=TiEwNf1H_Tc
    https://www.youtube.com/watch?v=TiEwNf1H_Tc

    thanks for the help good job ive only got low hours with my prnter plugged in ill get this sorted with out going stupid money what would be a good ssr to get



  • A solid state relay - a device to switch a large current using a small one. Used when you power your heatbed from mains power with a DC-AC SSR, or if you buy a DC-DC SSR from a second DC power supply. Usually they isolate the low current and high current side optically. A good idea if your print bed is larger than the usual 200x200mm as to achieve the desired 0.5W/cm2 requires a larger heater which draws more current than most controller boards can handle.

    They can handle low frequency PWM for more precise control of bed temperature.

    It is also possible to use a mechanical relay, but without PWM using bang bang mode.


Locked
 

Looks like your connection to Duet3D was lost, please wait while we try to reconnect.