EEVBlog Suggestion: What happens when you put bog-standard PSU's in Parallel

[Avacee]

Edit: Thread was originally titled "in Series" - I have a brain-fart as I meant to suggest "in Parallel".


This suggestion is inspired by this thread: https://www.eevblog.com/forum/projects/2-dc-power-supplies-in-parallelseries-operation/  (chosen as most recent) and several other threads where people ask how to combine one or more power supplies into one mega-supply.

This question seems to be asked a lot but I couldn't find a decent video blog about why it's not so simple and what happens when it goes wrong.
There are quite a few videos about how such and such whizzy PSU, whizzy controller, whizzy perfectly matched components on a whizzy perfect PCB can do it.

Whilst my google/youtube-fu may be weak there don't seem to be any high-quality in-depth educational demonstrations showing what happens with bog-standard PSUs (ie lacking all those whizzy features).. with DaveCAD illustrations explaining the fundamentals.

ie why those 12V 1A and 5V 1A supplies don't magically make a 8.5V 2A
or why that 12.0V 6A and 12.0V 4A supplies are actually ~12.01V and ~11.99V and don't make 12.0V 10A with perfectly balanced 60/40 current.

I think there's also an excellent opportunity for the Magic Smoke to be released during the video and who doesn't want to see that
Edit: I'm not suggesting Dave blows up several hundred dollars worth of PSU's - just some cheap chinese wall warts

[SL4P]

An interesting idea, but there are a number of fundamentals that make it quite hard to demonstrate (without cooking supplies)...  isolation, shared grounds, output stage design expect.

It’s a great case of finding the weakest link, and perhaps developing a set of tools for psi output protection.
It could certainly be a worthwhile whiteboard discussion.

[Yansi]

Bog standard PSUs will become bog standard series connected.

[SL4P]

Bog standard PSUs will become bog standard series connected.

If i’m reading your post correctly, please take a video when you do this!

[Yansi]

Well I would like to do more videos (have a channel already), however having not any reasonable camera being the main lilmitation currently, apart from my "not wanting to speak to much".

Regarding the series connection: There is not much of a magic there.

1) Be aware of PSUs having sometimes grounded negative output terminals.
2) Be aware of PSU rating of maximum voltage to earth-ground, if the output is floating. You can't float a bog standard PSU at few kV (connect hundred 24V PSUs in series), obviously.
3) Be aware of that series connecting (any) PSUs may work under normal load conditions, however at overload (short circuit),  PSUs with lower output current limit threshold will start decreasing voltage sooner and may become reverse polarized by those PSUs that still put out a enough volts. This may result in damage. In that case anti-parallel diodes can help to protect the PSU.

Yes, (too) short explanation, you already know all of that. Just to remind some of the bascis.

[james_s]

I've connected power supplies in series lots of times, so far it has always worked fine that I can recall. Is it a good idea? Probably not, I certainly wouldn't design it into a commercial product but my own experience suggests that it will be fine if you're just screwing around.

[Avacee]

Actually, I've just realised I meant to say Parallel in the title 

The examples I gave in the original post were for parallel :p  eg people thinking the voltage and/or current would magically auto-share.

What made me think of this suggestion was w2aew's Youtube video #233 showing power supplies in series. 
I google'd for parallel examples but found none hence the suggestion.

Though, ... Series or Parallel .. there are lots of "Traps for Young Players" either way
Maybe 2 videos - one each

Edit: Changed title at this point to say Parallel as that's a more accurate representation of the suggestion.

[james_s]

Well the server PSUs can certainly be run in parallel, they're designed to be used that way for redundancy. They have a load sharing pin you connect so that they can work together.

For other types in many cases it will work, but again it's not really recommended.

[IanB]

Often you can connect laboratory bench power supplies in series or parallel. See the user manual for instructions.

For example, here is an extract from the user manual of a TTi power supply:

Series or Parallel Connection with Other Outputs

The outputs of the power supply are fully floating and may be used in series with other power supply units to generate high DC voltages up to 300V DC.

The maximum permissible voltage between any terminal and earth ground is 300 VDC; the maximum permissible voltage between either terminal of one output and any terminal of another output on the same supply is also 300 VDC.

WARNING! Such voltages are exceedingly hazardous and great care should be taken to shield the output terminals for such use. On no account should the output terminals be touched when the unit is switched on under such use. All connections to the terminals must be made with the power switched off on all units.

It should be noted that the unit can only source current and cannot sink it, thus units cannot be series connected in anti-phase.

The unit can be connected in parallel with others to produce higher currents. Where several units are connected in parallel, the output voltage will be equal to that of the unit with the highest output voltage setting until the current drawn exceeds its current limit setting, upon which the output will fall to that of the next highest setting, and so on. In constant current mode, units can be connected in parallel to provide a current equal to the sum of the current limit settings.

http://docs-europe.electrocomponents.com/webdocs/0ce7/0900766b80ce71b9.pdf

[Richard Crowley]

You should probably suggest this to those YT guys who make videos of things blowing up and maiming themselves, etc.       

As others have said, there may be SOME power supplies that are safe to connect in parallel, although the percentage is likely very low.
Else, they will likely fight with each other and they may both (all) lose in that battle.
There are so many variables here, it would be impossible to make a general case/answer for this question.

[Brumby]

You should probably suggest this to those YT guys who make videos of things blowing up and maiming themselves, etc.

There are enough idiots doing that.  So much so, that I prefer to keep any such ideas to myself.

My concern lies more with the impressionable uninformed who might think it's a fun idea to try themselves - without understanding what's going on and don't take appropriate care.

[Richard Crowley]

There are enough idiots doing that.  So much so, that I prefer to keep any such ideas to myself.

Better for the OP to watch a professional idiot do such a stupid thing than to risk it himself.

If the power supply does not explicitly state that it is designed for parallel use, you can assume that it is NOT safe to do such a thing.  That will likely cover 99.8% of all cases.

[Brumby]

There are enough idiots doing that.  So much so, that I prefer to keep any such ideas to myself.

Better for the OP to watch a professional idiot do such a stupid thing than to risk it himself.

There is that....

[james_s]

It depends on what sort of power supply we're talking about. I was thinking mostly small low power stuff like wall warts and cord lumps, the sort of thing where abusing them might kill the power supplies but probably isn't going to result in anything dangerous.

[Brumby]

.... Then you need to be a bit more specific in defining the subject of the discussion.  Preferably in the first post, so that discussion doesn't head in a dozen directions.


Having said that - the question is quite valid.  Plug packs, wall warts or whatever you want to call them are often a cheap source of power for many as they start out.  Looking for higher current is not an uncommon ambition.

[Richard Crowley]

There are certainly tried-and-true techniques for combining power sources such as ballast resistors and/or using diodes to prevent cross-feeding power sources, etc.  But simply hard-wiring even ostensibly "identical" power supplies in parallel is unlikely to have the desired effect without destructive side-effects.

Ref: http://www.ti.com/lit/an/slva250/slva250.pdf