A word of warning on using cheap Micro USB cables.

[waveform]

I ordered a longer usb cable which arrived today from ebay, which is about 1m long.

This is a picture of the exact cable:



When using it with the Raspberry PI, the screen kept cutting off, and the PI would restart repetitively.

I also noticed some SD card reading errors.

When the PI didn't restart, and asked for a login, there was no response from a USB keyboard, which indicated it was not receiving enough power.

The power source is a 2A usb plug which works fine.

I swapped it for a friends Micro usb cable, which is only slightly shorter.

And instantly the problem went away, the Pi booted into debian perfectly.

I even tried his cable with an on board usb port, and despite the 500ma limit, it booted perfectly also.

I'd imagine this cheaper cable has thinner wires inside which are not quite capable of delivering 700ma of current, or may have a slightly more resistive conductor.

The cheaper cable worked fine on my friends phone for charging, and transferring data.

So if anyone out there has a PI that's restarting

or not accepting an input from a usb keyboard,

then it's most likely the power supply or the micro USB capable is not sufficient enough.

[jbeale]

Interesting…:

The ground between the two male USB connectors has low resistance and is connected in parallel to the USB's plug shielding metal and cable shield. The problem is the +5V wire. Even though the distance between the two male USB connectors is only 22 cm the resistance is 4,5 Ohm. Apparently ultra fine wire is used, as in many cheap USB cables [...] taken from http://forum.xda-developers.co.....tcount=196

For a resistance of 4.5 ohms at 22 cm length, the wire gauge must be incredibly thin. AWG 40 wire which is hair-thin would be about 1 ohm.  That is one cheap cable (and clearly not USB spec-compliant).

[waveform]

The good USB cable my friend swapped has a Samsung sticker on the back of it, it looks identical to this, except that the sticker says U2 instead of U8:

http://www.ebay.ie/itm/GENUINE.....0813257487



Note the distinct wedge shape on the connector.

There's hong kong sellers selling the exact same shape cable also, but the colour looks lighter, more grey than black:

http://www.ebay.ie/itm/NEW-Mic.....1039405808




It might be a good idea if people start posting up links to known good cables the purchased,  and pictures of what they look like exaxtly.

[waveform]

jbeale said:


Interesting…:

The ground between the two male USB connectors has low resistance and is connected in parallel to the USB"s plug shielding metal and cable shield. The problem is the +5V wire. Even though the distance between the two male USB connectors is only 22 cm the resistance is 4,5 Ohm. Apparently ultra fine wire is used, as in many cheap USB cables [...] taken from http://forum.xda-developers.co.....tcount=196

For a resistance of 4.5 ohms at 22 cm length, the wire gauge must be incredibly thin. AWG 40 wire which is hair-thin would be about 1 ohm.  That is one cheap cable (and clearly not USB spec-compliant).



Interesting post.

The funny thing is that the diameter of the good Samsung cable is a fair bit thinner than the bad Nokia – clone cable.

That's not to say the wires inside the bad cable are any thinner though, it may just be the insulation that is thinner.

**BTW if someone could delete the duplicate post it would be great!

[bobc]

I've found various cheap cables have a thick plastic sheath but with hair fine wires inside. Plastic is a lot cheaper than copper I'll bet some of this cable was intended for things like headphones which do not need to carry current, but help by being flexible.

It is has become an expectation in the West that products we make and buy should actually work, it seems that expectation is not shared in the far East !

[johnbeetem]

bobc said:

It is has become an expectation in the West that products we make and buy should actually work, it seems that expectation is not shared in the far East !
I've usually had good experience with Japanese and Korean products.  OTOH, I've had mixed experience with [Mainland] Chinese products -- some are excellent and others terrible.  Looking at a cable on a web site or without cutting it open, it's very hard to tell.  It reminds me of John Wanamaker's famous quote:


Half the money I spend on advertising is wasted.  The trouble is, I don't know which half.

[niftyshellsuit]

So my Pi runs for a few hours, then disappears, no display, no access vis SSH, nothing.

I swapped out the micro usb cable to the one that came with my phone and that improved it, but it still gave up by the following day.

I think the problem must be shoddy cables. I researched the power adapter and got a good one, but didn't really consider the cable too much so just got a cheapo £1 from ebay…

As I have spares, I decided to chop this one up to see how fine the wires were inside…

[hippy]

niftyshellsuit said:

I think the problem must be shoddy cables. I researched the power adapter and got a good one, but didn't really consider the cable too much so just got a cheapo £1 from ebay…
The thing is though, if something is deteriorating over some period of time, that would not seem to me likely be the cable deteriorating.

A power supply deteriorating over time would be understandable, components getting hotter or becoming unreliable the longer it is used, but a good quality supply shouldn't do that.

A poor quality supply with a poor quality cable can fail sooner than a poor quality supply and a better cable but it still really comes down to the supply. Some indication of what supply have would allow people to assess if that's likely to be the problem.

[jbeale]

niftyshellsuit said:

As I have spares, I decided to chop this one up to see how fine the wires were inside…




That photo shows wires that look OK to me. If you had one of the very-high-resistance cables (4 ohms category) you'd see just one ultra-thin strand of wire for the power lines. Of course there still might be a bad connection somewhere; either inside the cable or at the connectors on the power supply or RasPi side. There could also be a software issue although that should not kill everything (at the very least you'd still see the power light, since that is hardwired to the power supply). There's a resettable input fuse but you shouldn't be able to overload that unless you're drawing current from the GPIO pins, since the USB connectors have small fuses of their own that should (in theory) trip first.

[thexman]

i woul point out the standard usb cable is REd +5v black 0V green sd1 and white sd0 yours doest even have the universal colour coding thats just wrong.

[BertM]

To all people that are suspecting a micro USB cable: measure the voltage between TP1 and TP2. If it's at least 4.9V, you can be pretty sure the cable is not faulty.

[niftyshellsuit]

Thanks for all your comments  - I don't know much about this stuff so it's a bit of a learning curve trying to get it all right!

I am beginning to think it is software rather than hardware, but I am still suspicious as it did last much better with a different cable... I have myself a multimeter to play with tonight so I will post some measurements later.

[Chris.Rowland]

I've seen a power issue that looked exactly like a software error. The failure happened in exactly the same place, in exactly the same way and was completely reproducible - until I changed the power supply.

What was happening was that the software operation caused a power surge at exactly the same place every time and so the same memory was corrupted as the power dropped out, causing the same fault.

[n3tw0rk5]

niftyshellsuit said:


So my Pi runs for a few hours, then disappears, no display, no access vis SSH, nothing.

I swapped out the micro usb cable to the one that came with my phone and that improved it, but it still gave up by the following day.


Thats whats happening to me on a daily basis, but i"ve swapped the original 700ma psu and cable I was using for an 850ma one that came with my sony ericsson arc s, so they should be half decent, but it still freezes.

dmesg and logs show nothing, so dont know if it could still be a power issue or a memory hog issue etc.

[niftyshellsuit]

Borrowed a multimeter and wow, there's some wildly different results!

They went from the cheap cable with the Belkin plug (I bought specially for the Pi, grr) which fluctuated wildly between 3.86 and 4.46. This is the combo it was running on originally before I swapped the cable..

I swapped it to the Belkin plug with an old Nokia cable, which was much better but not brilliant, giving me a nice stable 4.75.

I tried out all the other cables and plugs I got til with varying degrees of success and found that the winning combo was my ZTE Blade plug with a Sony Ericcson cable, which gave me a nice stable 5.03.

The next best was the same plug with the Kindle cable which was 4.86. Turns out anything I stuck in the Belkin plug was way low, not one of them got over 4.60. The Sony Ericcson was not much different.

We will see how it goes anyway, I'm happy I don't have a power issue any more at least...

[niftyshellsuit]

Got to work today having left the Pi on overnight and it was gone

So it must not be a power issue. Back to the drawing board.

n3tw0rk5, if you find a solution please be kind and share! Sounds like we are having the same problem...

[n3tw0rk5]

niftyshellsuit said:


Got to work today having left the Pi on overnight and it was gone

So it must not be a power issue. Back to the drawing board.

n3tw0rk5, if you find a solution please be kind and share! Sounds like we are having the same problem...



Damn nearly bought another psu today until I'd noticed mine had been up over 24 hours, it's actually powered on 24/7.

Dont know if it's a buffer issue, etc. These are a bit different to what i'm used to but it's a good learning curve.

Not sure what can be played with as to memory resources etc.

If I find anything out, I'll shout up.

[n3tw0rk5]

My raspi keeled over again so I've been to maplins and got a high spec psu and a quality cable.

Will try them tonight and see what happens but if it still freezes after this it cant be a power issue. At least then I can concentrate on looking elsewhere for the cause.

[Smartybones]

I've been to maplins and got a high spec psu and a quality cable.

Be aware with maplins. quite a few of the products they sell are the same cheap products you can buy from Chinese ebay shops but place in fancy packaging.

also be careful when thinking because its a big branded product that it may be better. Many of these companies make a lot of profit in after sales parts like PSU's . Most of the time, the part of a PSU that fails will be the secondary winding or the bridge rectifier. so why make an expensive product when the cheapest components might fail? a lot of the regulation and smoothing will be done internally on the device the psu is intended for. (which is why they make it clear only to use psu that is intended for the product, so it matches the spec required for the regulators). Apple make a lot of money out of replacement PSU's when a genuine apple PSU costs £20 for something that should really only be £5

when looking for a psu, don't buy on price expecting the more you pay to be better. My PSU is one I got from morrisons for a fiver and the cable i use is the one from my HTC desire and works perfectly. my Pi has been left running for a few days now.... (just for the LOL's, i am going to compile OpenELEC on it !!)

carefully check out the specs of the psu, check out what its tolerances are. a good psu should state it on its spec sheet. look for something that is better than + or - 5%

[Montala]

Having just read through the previous posts in this thread it has come as quite a surprise to me just how much voltage is being 'lost' when using a cable with with small internal cores.
Although I haven't received my Pi yet, I have already purchased a multimeter, so that I can check the actual voltage which is being delivered to the board.... and I also have a few alternative PSU's and cables to experiment with!
Unfortunately, as others have said above, a higher price doesn't necessarily mean higher quality.

[dolson14]

I've been having good luck with this power supply, for what it's worth:

http://www.microcenter.com/single_produ ... id=0375186

[Steady_Bear]

Thin copper wire in cheap cables? Lucky you.
A lot of the cheap cables go for steel (copper doesn't rust, so I know it wasn't that ). Brittle, high impedance, but really cheap. Hence the issues.

Also, don't assume that it's copper just because the strands look like copper on the outside, it may still be any cheap old tat with a light plating of something brown and glossy.


Buy cheap, buy twice! Although the earlier post about ignore costs is far too true. Read the reviews before buying anything, or make sure you're in a position of no-fuss returns.

[Jancis]

yeah, some cheap ebay cables are really bad quality. bough 5 microusb cables for my smartphone and it wouldn't charge properly overnight. If i am using and charging at the same time, phone would pop up message saying that current is too low to charge. then i spent 5$ at local store and bought cable that actually works.

point is - it can actually be true that your raspi acts weird because of bad quality power cable. and, you should think twice about buying them cheap off ebay.

[doveman]

So where can I buy 24AWG USB-micro USB cables in the UK.

I need to cut the USB end off and wire that to a socket on my case. I did buy some micro USB bare connectors with the intention of wiring the power lines to that myself, but the pins are tiny and very close so I've given up on that idea!

Alternatively I could wire from the case socket to a 2-pin header going into the GPIO (via a 2A fuse) but I'd still need to find 24AWG wire and a way to get that attached to a header (I'm not going to solder direct to the RPi and invalidate my warranty). Is this 24AWG (the 24/0.2 stuff)? http://www.maplin.co.uk/buy-cables-and- ... pment-wire