Adequate Power Supply Critical to Pi Stability

I was able to pull a mind blowing 3 amps through the Micro USB connector (with wires bridging F3/D17 directly to the USB port power pins), so yep, it can handle pretty much whatever you feed it.

By back powering one of the USB ports I was able to pull 6.5 amps out of the other (with the USB power pins bridged), but then the USB cable started to melt

There are lots of ways to get more power into and out of your Pi

-Jamie M.

I was able to pull a mind blowing 3 amps through the Micro USB connector (with wires bridging F3/D17 directly to the USB port power pins), so yep, it can handle pretty much whatever you feed it.

By back powering one of the USB ports I was able to pull 6.5 amps out of the other (with the USB power pins bridged), but then the USB cable started to melt

There are lots of ways to get more power into and out of your Pi

-Jamie M.

Heh, sounds like you had fun

I'm actually not plugging any USB devices into the Pi, so that will only be drawing 700ma-1A at most but even so I need to make sure if I use the micro USB that sufficiently thick cable is attached to it, not some cheap single strand rubbish. If it looks at all questionable, I'll just use the GPIO.

The hub might draw 2A but that'll be through decent cable to a barrel connector so no worries there.

So I've hooked up a 16v/4A PSU to my UBEC, which is feeding the RPI via a 2A fuse into the GPIO and my 7-port hub directly, which has plugged into it a Dual DVB-T USB tuner and SD->USB adapter holding a 8GB Sandisk card (I'm waiting for a normal USB stick to be delivered).

I measure 4.87v across TP1-2, 4.89v with the hub data cable disconnected and 4.93v with the hub power disconnected.

Doesn't seem great when others report they get a steady 5.25v using the UBEC.

I measure 4.87v across TP1-2

TP1-TP2 voltage is after the polyfuse F3. I bypassed/bridged the polyfuse F3 (solder a wire from one side to the other) in order to stop so much voltage drop to the Pi. I think some of the poly fuses have less resistance from the factory then some

Touch a paperclip or something to each side of F3 to temporarily see how much your TP1-TP2 voltage improves.

-Jamie M.

I measure 4.87v across TP1-2

TP1-TP2 voltage is after the polyfuse F3. I bypassed/bridged the polyfuse F3 (solder a wire from one side to the other) in order to stop so much voltage drop to the Pi. I think some of the poly fuses have less resistance from the factory then some

I'm going in the GPIO header though, so bypassing the polyfuse (hence why I'm using a 2A fuse between the UBEC and the GPIO header).

I'm going in the GPIO header though, so bypassing the polyfuse (hence why I'm using a 2A fuse between the UBEC and the GPIO header).

Ahhh, so GPIO is on inboard of F3 polyfuse? Didn't know that Nice.

Sucks about your low voltage then

-Jamie M.

I plan on using a Kindle charger (4.9V, will post amperage when I remember) for my Pi's power supply. Does anybody know if that's an adequate supply for it as well as my keyboard/mouse and a USB drive?

I plan on using a Kindle charger (4.9V, will post amperage when I remember) for my Pi's power supply. Does anybody know if that's an adequate supply for it as well as my keyboard/mouse and a USB drive?

USB thumb drive maybe, USB powered hard drive, only if you do serious modifications to your Pi (plus minimum 2amp/10watt USB power supply and super good quality MicroUSB cable with 24gauge power wires)! http://raspberrypi.stackexchange.com/qu ... hrough-usb

-Jamie M.

Okay, so what about for just the Pi itself and a keyboard/mouse? Will a standard iPhone charger do the trick?

Okay, so what about for just the Pi itself and a keyboard/mouse? Will a standard iPhone charger do the trick?

Yeah, mouse and keyboard are very little draw. The older iPhone chargers are 1000ma and with a descent MicroUSB cable will do the job perfectly. The newer chargers are even more powerful and work great of course as well.

-Jamie M.

Looking at the cables for powering the Pi, I now wonder why, if 24 AWG is sufficient to carry a couple of amps at 5v, why audio cable needs to be 32*0.1=3.2mm=8 AWG like this one or 156/0.06=9.36mm=greater than 0 AWG like this one when it only carries 0.7-1v at milliamps? Even the cheap stuff like this is 0.7mm=21 AWG.

why audio cable needs to be 32*0.1=3.2mm=8 AWG like this one

That audio cable is 0.1mm "Stranded: 2 x 32 x 0.1mm ", which is a super tiny 38 gauge!

I think you're looking at the outer cable diameter, which includes all of the conductors, their insulation, plus any shielding if applicable, and the outer cable insulation (which seems super overkill on that cable).

8awg cables for audio are generally car audio power cable, good for up to 500watts amplifier RMS (usually pulling around 50amps of current). 4awg is good for up to 1000rms/100amps and 0 gauge is good for "more"

Here is some of my car audio wiring, the purple wire is 0 gauge, the brown wire is 4 gauge


LICK FOR HIGH RES!

-Jamie M.

why audio cable needs to be 32*0.1=3.2mm=8 AWG like this one

That audio cable is 0.1mm "Stranded: 2 x 32 x 0.1mm ", which is a super tiny 38 gauge!

I think you're looking at the outer cable diameter, which includes all of the conductors, their insulation, plus any shielding if applicable, and the outer cable insulation (which seems super overkill on that cable).

8awg cables for audio are generally car audio power cable, good for up to 500watts amplifier RMS (usually pulling around 50amps of current). 4awg is good for up to 1000rms/100amps and 0 gauge is good for "more"

I don't get it Surely the 32 x 0.1mm means 32 strands each 0.1mm thick totalling 3.2mm?

The overall diameter of the cable is 9.6mm so each of the outer sheaths is about 4.8mm and the core is obviously quite a bit smaller than that but no way it's only 0.1mm.

EDIT: This 8 AWG car audio cable is 700 strands of 0.12mm which I'd think would add up to 84mm but obviously that would be a massive cable and the specs state the overall cable diameter is only 7mm, so clearly I'm missing something here! I'm just trying to find some flexible cable suitable for carrying 2-3A at 5v. Would this speaker cable with the core at 79/0.2mm be OK? http://www.maplin.co.uk/hi-fi-loudspeaker-cable-50

Actually, that's too thick at 6mm x 3mm as I need to get two wires for the power and two for the USB data into a DIN plug like this which can take a maximum of 5mm diameter in it's cable grip.

Another option, this cable gives the Nom. conductor area as 0.055mm2 which equals about 20 AWG according to this chart

On the Overview tab it states "Stranded cores, 7/0.1mm copper conductor" so I think it's safe to say this cable I bought, which states "Stranded cores 7/0.12mm copper conductor" will be slightly thicker and 20-19 AWG.

Well whilst the cable I bought at 7/0.12mm should theoretically be around 20 AWG, it seems thinner than the cable coming out of my UBEC, which is roughly the same thickness as the cables attached to the PC headers I have (used to connect front panel power/HDD LEDs, power/reset buttons to the motherboard), which are marked as 26 AWG.

I think the 7/0.12mm cable will be OK for the USB data but I'm not comfortable using it to carry 2-3A to my hub, so I'll have to just use two cores from the 4-core lead for the USB data and the PC header cable or something else if I can find it for the UBEC->Hub power.

Another option, this cable gives the Nom. conductor area as 0.055mm2 which equals about 20 AWG according to this chart

Ah, misread that chart. 0.5mm2 equals about 20 AWG. 0.055mm2 is more like 30 AWG. Not sure if I should even use that for my USB data cable now.

This page http://www.rbeelectronics.com/wtable.htm suggests a 3ft cable carrying 6v/3A needs to use 18 AWG, although that's probably overkill as it says that 18 AWG can be used up to 20A http://www.rbeelectronics.com/wtable.htm. I doubt the cable attached to my UBEC is 18 AWG anyway as it looks more like the 26 AWG PC header cable I have.

Still no idea how convert something like 32/0.1mm into mm2 though. This page shows that 16/0.2mm equals 0.5mm2, which is 20 AWG. http://www.rapidonline.com/cables-conne ... 2mm-62322/

[rurwin]

in 16/0.2, the 0.2 is the diameter of the strands. The area of the cross-section is simply pi x r^2. So halve the diameter, square it and multiply by pi.

0.2/2 = 0.1mm radius
0.1 x 0.1 x pi = 0.0314mm^2 each strand
0.0314 x 16 = 0.5024mm^2 overall

so 32/0.1 would be: 0.05 x 0.05 x 3.14 x 32 = 0.2512 mm^2

When you halve the diameter, the area drops by a factor of four.

Ah thanks for putting me out of my ignorance rurwin

Now I can work out that my 7/0.12mm cable is actually 0.0789mm2, so around 28 AWG, which is probably too thin to carry 2-3A at 5v. Mind you, perhaps the 26 AWG (0.129mm2) PC header cable I was going to use is as well, so to be safe I might get some 16/0.2 (20 AWG) or 32/0.1 (23 AWG) for that instead.

I need to run a cable from the USB data socket about 1ft to the other side of the case, near the micro-usb end of the Pi, where it will go into a 5-pin DIN socket. Also attached to that will be two short wires from the nearby UBEC to carry +5v and GND. Then a short cable attached to a DIN plug will take all four lines (USB Data +/- on pins 4 and 5, +5v and GND on pins 1 and 2) to the hub.

My concern is shielding the cables so they don't cause/receive interference. The data cable is probably most vulnerable to interference, which is why I wanted to use the 28 AWG 4-core shielded cable I bought, both inside the case and in the external lead. If the data leads were shielded, do you think it would be safe to run the +5V/GND through the unshielded 26 AWG PC header cables (or some thicker cable) alongside it? Or am I being overcautious and would it be safe to run the data through a pair of unshielded 26 AWG PC header cable as well? I could use 24 AWG cores from a CAT5e cable inside the case, it's just the external lead that I can't as it needs to be flexible, not solid-core.

I checked the cable attached the hub's 5v/4A PSU and that's 18 AWG and I found that the cable attached to a PC ATX power lead is 18 AWG (some of the wires are thinner, so you need to check, I found the red and white wires were 18 AWG) so was going to use that but comparing it to the leads coming out of my UBEC, those are a lot thinner, maybe half as thick, so using 18 AWG is clearly pointless and if I can use something like the 26 AWG PC header cable I've got, that will be a lot easier to solder to my DIN plug and socket. If that's too thin to safely carry 5v/3A, I guess I could just run two lengths in parallel (4 in total for +5v and GND), which would still be thinner than 18 AWG and thus easier to solder.

I've already whined about the F3 and PSU issues here:

viewtopic.php?f=24&t=25910

IMHO a standard DC barrel jack would be much preferable to the current Micro USB.

But if you're looking for a good PSU, I'd recommend the Nokia AC-10E. 1200mA and has a good quality PCB with a regulator IC and a heatsink inside.

Thanks but I've already got a 12v/5A PSU which is feeding my UBEC (via a barrel connector fitted to my case, which as you say is the common connector on PSUs).

I posted my thoughts over on that thread as well anyway.

Thinking about using two PSU's - one for the Pi only, another for a HDD and a USB hub, mounting everything in a single case. Not sure of the wiring, but I have a scheme in mind. Can I poost a drawing for someone to look at? How do I post a gif?

Can I poost a drawing for someone to look at? How do I post a gif?

Upload it on tiny pic and it'll give you the URL of the image, and you can just put the URL between the IMG tags (there's an IMG button on the top toolbar when you go to post).

If it shows up small when you post it, don't worry, that fine, we just right click the picture and go "view image" or whatever, and it'll make it full size.

-Jamie M.

Can I poost a drawing for someone to look at? How do I post a gif?

Upload it on tiny pic and it'll give you the URL of the image, and you can just put the URL between the IMG tags (there's an IMG button on the top toolbar when you go to post).

If it shows up small when you post it, don't worry, that fine, we just right click the picture and go "view image" or whatever, and it'll make it full size.

-Jamie M.

Hope this works -- will this wiring approach work?