The question above pops up quite often these days.
An example is CCTV cameras that operate on 12v DC.
Would you choose a single 12V 10A power supply to power all of them or say 10 small ones, one for each camera?
If you let aside the fact that with a single PSU, when it fails, everything fails as opposed to the small ones where a single failure results in the loss of only one load, what would be more energy efficient?
The same goes if you have a number of arduinos operating in the same environment. Would you go for a single PSU ?
The question could be extended even further: Say you have available 12V DC but you also need 5V for some TTL logic. Would you use a DC-DC converter to get 5V from the 12V or install a seprate mains PSU for 5V ?
Or in the case of POE (Power Over Ethernet 48v) devices, would you choose separate POE injectors or go for a single POE PSU / switch?
Power supplies these days are mostly of the switching type, keeping thermal losses to the minimum.
Would it be fair to assume that a bigger PSU would be more efficient than a number of smaller ones?
Or is the difference so small that it doesn't really matter?
However if you are distributing power with PoE at 50V then for the same cable size I squared R losses will be 25 times what they would be at 250V. Add to that the fact that the cable won't be the same size as the wire in cat 5 cable is thinner than typically used for mains power and you get even higher losses with PoE than with a mains fed PSU close to the load.
Cable length decides the issue - short cable runs and one PSU is likely to be the best bet, long cables and
you need to distribute power at high voltage (ie mains) to avoid spending a fortune on copper.
Imagine some wiring run that's 1 ohm in total. At 240Vac mains that can carry 12A for 5% voltage/power loss,
ie it can carry a total of 2.9kW for that level of inefficiency.
The same wiring at 5Vdc can only carry 250mA for the same percentage loss, or 1.2W total load.
So there's a very strong incentive to distribute power at higher voltages for longer distances, and
this often forces the disposition of power converter units.
If the power supply you use is rated for 10A and
the wires are rated for 2-3A, if you have a short
it will melt the wires and /or fry the IC without
reaching the current limit rating, possibly causing
the IC to burst into flames.
Been there, done that.
Bad idea.
Two 90% efficient power supplies are still 90% efficient.
Say you have available 12V DC but you also need 5V for some TTL logic. Would you use a DC-DC converter to get 5V from the 12V or install a seprate mains PSU for 5V ?
DC-DC converters are not 100% efficient so a power supply that has 12V and 5V outputs is probably more efficient, depending on the particular power supply. If you are just powering "some TTL logic" the 5V power usage may not be significant.
One large one and allow for proper wire gauges based on run length. I also would fuse each load and considering power supply cost I would have a spare on the shelf.
Maybe I should have clarified that cable length remain the same in both scenarios.
In the case of POE devices, the cat5e/cat6 lengths are the same no matter if I use a single POE switch or individual POE injectors. In both cases cat5 needs to run from the central point (which is common for bot cases) to each and every POE device. So cable losses are the same in both cases.
If the power supply you use is rated for 10A and
the wires are rated for 2-3A, if you have a short
it will melt the wires and /or fry the IC without
reaching the current limit rating, possibly causing
the IC to burst into flames.
Been there, done that.
Obviously a 10A PSU would have proper power distribution and no single wire would get overloaded.
I just wanted to focus on the efficiency of individual small PSU Vs a single big one.
If we ignore those wiring losses...
Two 90% efficient power supplies are still 90% efficient.
True!
However, do bigger PSU tend to be more efficient?
And how about quiescent current / losses ? 10xIq Vs Iq(big)
Bear in mind that not all devices / loads draw the same amount of current all the time.
Or maybe if the same PSU is loaded towards its full rating, it gets more efficient.
In general a larger PSU will be more efficient(*), in particular you consult the datasheets involved
since there is a lot of variation, and efficiency is a function of load current, not a single figure,
so until you know the loading you can't necessarily chose between supplies with identical power
rating. There's far more variation in the efficiency at 5% loading than 50% loading for instance.
(*) because thermal management is more of a design issue and the budget allows for more tricks to
be pulled (things like resonant conversion, synchronous rectification, which are more complex/costly
but bring efficiency benefits).
"Obviously a 10A PSU would have proper power distribution and no single wire would get overloaded.
I just wanted to focus on the efficiency of individual small PSU Vs a single big one."
Proper power distribution meaning WHAT ?
How exactly do you plan to prevent a short from occurring somewhere? Anything running off the 10A supply (directly) canpotentially draw 10A.
Fusing each circuit individually (preferably with
poly fuses) is the only way to prevent overloading wires.
raschemmel:
"Obviously a 10A PSU would have proper power distribution and no single wire would get overloaded.
I just wanted to focus on the efficiency of individual small PSU Vs a single big one."
Proper power distribution meaning WHAT ?
How exactly do you plan to prevent a short from occurring somewhere? Anything running off the 10A supply (directly) canpotentially draw 10A.
Fusing each circuit individually (preferably with
poly fuses) is the only way to prevent overloading wires.
Yes, fusing each individual circuit.
But I dont like polyfuses as they have a resistance that leads to voltage drop /losses which, in my experience, increases each time they operate. I prefer standard replaceable fuses.
MarkT:
In general a larger PSU will be more efficient(*), in particular you consult the datasheets involved
since there is a lot of variation, and efficiency is a function of load current, not a single figure,
so until you know the loading you can't necessarily chose between supplies with identical power
rating. There's far more variation in the efficiency at 5% loading than 50% loading for instance.
(*) because thermal management is more of a design issue and the budget allows for more tricks to
be pulled (things like resonant conversion, synchronous rectification, which are more complex/costly
but bring efficiency benefits).
Thank you.
My thoughts exactly.
PS: Reasonat conversion: There s something I am not familiar with... - Karma points given!
- Karma points given!