Power Supply Question

Question

Good day,

Can anyone recommend a good power supply to power 4 or 5 model 3 Pi's? Or can anyone point to a hack with a computer power supply wiring to multiple usb jacks?

Any info would be great. I don't want to run a single power bar for just power bricks for my Pi's.

Thanks

Terry

clem57 · Accepted Answer

Not sure where you are going with this, but take a look at Raspberry Pi gets a Blade!

rsc · Answer

I've found that my Pi-3 needs about 5.2VDC to operate properly with a WD pi-drive and regular sized keyboard and mouse. The supply I built for testing this has an adjustable 5V output. It's a Mean Well RD-50B dual output 5V@4A and 24V@1.4A http://www.onlinecomponents.com/mean-well-rd50b.html?p=11952931 I added a fuse and current meter to monitor what's happening. Here's a pic Scott

dougw · Answer

Some power supplies need a load resistor on the power output to be stable, but I have re-purposed 3 PC power supplies using a custom PCB with banana sockets, fuses and switches. The card I designed has room for power resistor loads, but in every case the output voltages were accurate and stable without any load. The power switch just controls the power control signal. Maybe I will post a blog about it....

rew · Answer

A "cheap" hack, requiring some "DIY" is to use a standard (small?) ATX powersupply. The "5Vstandby" is enough to power one pi (i.e. silent!), you need to turn the supply on to power more than one. I would personally not hesitate to make a harness that connects to the GPIO pins, so that I can bypass USB cables to be sure that I don't run into bad ones (i.e. having a large resistance).

On my desk, for testing, I have an ATX powersupply that is "ON", but I have patched the fan to run off 5V instead of the normal 12V. I'm guessing that this reduces the total allowed power output from the 500W it is rated at to below 100W or so, but it is currently powering two raspberry PIs and a couple of other projects...... (say 5W for the pi's (on average, 10W peak each) and another 5W for the rest....)

So.... Another option is to do what I did: get that ATX powersupply, and route the 5V through thick cables to a big enough (cheap! (*)) USB hub, and just use the hub to distribute the power.

(*) The hub needs to be cheap enough that it simply passes on the power instead of being intelligent about it and only delivering the required 0.1A until the client negotiates more....

Workshopshed · Answer

The kit versions for adapting ATX to bananna plugs often have a space for a high current load resistor to place across the output. I believe it will help stabalise the output voltage as most of these supplies are switch mode. Best not short the outputs though!

Workshopshed · Answer

You can twist green and GND together or have a switch if you want. The load is from 5v to GND. http://www.electronics-tutorials.ws/blog/convert-atx-psu-to-bench-supply.html

rew · Answer

Yes, for both "switching powersupplies" and "atx power supplies" stories are going around that they require a lot to operate correctly. One of the devices where you wouldn't expect it are some MEANWELL power supplies. There it seems to me that they no longer guarantee that the output will remain within 5% of the nominal value if the output is unloaded. Not really a problem if you can tolerate say +10%. Just being able to cope with 13.2V is much cheaper in the long run than adding a load resistor. Similarly, some old ATX power supplies would give weird results, like the 5V rising to 6V or more when only the 12V was loaded. I haven't seen such "weird" behaviour from any supply made the last two decades. (I used to say I had never seen it, but that is no longer true. A few years ago, I grabbed an older powersupply from the pile of "old spare powersupplies" and it indeed behaved weird and required a load to be stable. That powersupply was built in 1992. ) In practice, it isn't a problem or only happens with very old/odd equipment.

dougw · Answer

Anker 10-Port 60W USB 3.0 Hub can supply 2.4A on each of 3 power ports.

beacon_dave · Answer

Not sure if it is still the case or not but a number of SMPS designs used to require a minimum load on the output in order to operate correctly. I'd hazard a guess that the space for the load resistor may be for this purpose, especially if being used as a bench supply where the load may not always meet the minimum load requirement of the SMPS.