Strange screen artifact

It's a low-power warning. It shows up when the 5V supply goes below 4.65 if I remember correctly.

http://www.raspberrypi.org/forums/viewtopic.php?f=29&t=82373

--

Ragnar

It's a low-power warning. It shows up when the 5V supply goes below 4.65 if I remember correctly.

http://www.raspberrypi.org/forums/viewtopic.php?f=29&t=82373

--

Ragnar

Good spot!

Thank you very much Ragnar!!!!!

That's exactly what was happening.

I've had issues many years ago with crappy USB leads, so I started buying one at a time from different ebay vendors until I found one that had cables that could carry 2,000 mA and then I bought a dozen or more from him, tested them each and gave most of them to my friends who were complaining about slow charging times.

(I also bought a bunch for myself)

So, when I saw your message, I collected a few of my 5V, 2,000mA power packs and tested them (as well as 2 different cables) with the Pi2.

Voila!!!

One substandard cable in the bin later and the Pi is happy.

Thanks again,

Nick

Glad you got it sorted!

--

Ragnar

@ Nick

I don't think it is as simple as low capacity power supplies or poor USB leads. In fact, I think the leads either work or don't work. I don't think they have much chance of going high enough resistance at the sorts of current taken by the Pi to stop it working correctly.

Anyway, here's my practical experience. I was getting freezes and a loss of USB with my Pi2 and found that transferring the mouse/keyboard and wifi dongles to a hub powered from the same dual USB PSU as the Pi, solved the problem. I had noticed the rainbow square at times, but not related it to power.

Recently, I decided to try the dongles plugged straight into the Pi and it was fine. The reason was that the (cheapo)  hub supplied power over its USB ports as long as the power was connected and that fed back to the Pi and supplemented the 5V rail. It also lights the red LED even if the Pi is not powered up.

I unplugged the hub from the Pi and rebooted. It worked well, but every time it communicated over the wifi dongle (a very cheap, very short device that probably has to work at full power) the warning rainbow square faded in and out.

The key points are,

     1) The hub and the Pi are getting power from the same PSU and so it is not a problem with the power supply lowering the 5V rail under load.

     2) It is unlikely to be a high resistance power lead. I just don't accept that two leads in parallel solve the problem.

The most likely problem is that the internal power arrangements for the Pi are not able to supply the peaks of load from the cheap wifi dongle. It's well known that Wifi is a significant drain of mobile device power, but it will take a bit of pfaffing about with USB PCB sockets and soldered wiring to find out how much this particular dongle is drawing. I may try with an up-market, slightly longer dongle to see if it draws less power, but that needs the IP blacklist on my router to be modified so I am in no rush.

Hi Roy,

Thanks for the information!

Alas, in my case, I have actually found USB leads in the past that vary a lot, as far as how much current they can carry. (This was done as part of a phone-charging issue....)

It took me a while to find ones that could carry (very close to ) 2A.

In the case of my Pi, replacing the lead and 5V supply helped fix the rainbow problem. I noticed that just changing the cable made the rainbow appear for a while, fade away, come back and so on....indicating brief outages, as opposed to consistent low-voltage. Once I also changed the PS fixed, it all went away.

I am using a wired-LAN connection, instead of WiFi, so that is not causing me any issues.

If this Pi was used in a critical application, I'd have looked for a nice, regulated 5V supply but, since it is only used as a media center and it works reliably, I'll keep using the little powerpack I have,

I've got a lot on at the moment, developing software for paying clients, so my hardware itch of getting the multimeter out and measuring resistance in various USB cables, as well as output voltages from the various powerpacks, has been put on hold!

Thanks again,

Nick

OHMS LAW my dear friends OHMS LAW.

On the newer versions of the Raspbian, there is a core feature that measures the onboard voltage and if it is detected below a set point then the little rainbow square shows up on the top right of the screen, you PI may not crash but it might

USB Cables come in all sorts of grades and by that I mean how thick the wires are in the cable, under light loads the drop across the cable can be low due to a lower current, as you start spinning up applications or adding load in the form of USB devices plugged directly into the PI, the current increases and therefor the voltage drop increases, I have shown this specifically through experimentation and presented the readings in my post. THE CABLE MATTERS.

I agree that based on your observations, the power adapter is not lightly to be the issue if everything works when the hub is powered via a separate cable from the power adapter than that of the PI and if you read my blog post on the matter you will see that in all the testing I performed it was never the PSU wall wart.

Cheap USB hubs that back feed to the PI should not be used especially if the PI is also powered from its own power input, the two can interfere with each other and cause problems, use a better quality HUB.

Your experience with the WIFI adapter proves my point, as you start to use things  like the WIFI, power consumption goes up (WIFI adapters and most other devices to not consume the same power all the time, it goes up and down depending on how much your asking them to do)

Powering the PI through a Keyboard USB outlet of from the standard PC USB port may work for low loads but these USB interfaces are nearly always limited to 500mA, not even close to that needed by a busy PI

Technically two leads in parallel could solve your problem, although impractical the technical result of two resistors in parallel is lower resistance and therefor lower voltage drop for the same current flowing through. In your case it is not really two leads in parallel, one lead is powering the HUB and the connected devices, the other lead it powering the PI, if you have a non-compliant HUB that back feeds to the PI then you will get a partial parallel effect but this is not why it is working better. It is because lower current is flowing in each cable and therefor lower losses before the power gets to the device (PI or HUB)

You need a Power adapter rated at 1.5A or more for the PI, especially if your adding loads of USB devices, the overall load on the PSU does not change if you separate the load over to cables, you just reduce the risk of dropping too much volts to the PI

PI 2 will require more power when up and running due to quad core etc. etc. If you still have keyboard, WIFI, and other devices attached, consider a HUB for them and use a 2A Adapter with GOOD quality cables

I said it at the top and Ill say it again “THE CABLE MATTERS”

Peter

Yeah, yeah, yeah: whatever!

So, I decided to do some measurements.

• No USB Devices and with the Raspi idling             Vout = 4.904V
• With Mouse/Keyboard USB Dongle                        Vout = 4.879V
• With above and idle WiFi Dongle                            Vout = 4.85V
• With active web-page download                             Vout = 4.764

• Voltage Drop of idle Raspi2        = 0.27V   @ 0.6 Ohms                  Current = 450 mA
• Voltage Drop of m/Keyboard     = 0.025V  @ 0.6 Ohms                  Current = 41.7 mA
• Voltage Drop of keyb +idle WiFi = 0.029V  @ 0.6 Ohms                  Current = 48.3 mA
• Voltage Drop of WiFi surfing       = 0.083V  @ 0.6 Ohms                  Current = 138.3 mA

My phone definitively likes the 10cm micro USB cable better than the 25cm one. (When depleted, it will charge a little, try to boot, and shut down repeatedly unless the charger is my 20000mAh battery pack and the cable is the 10cm one....Everything else repeats the loop.)

I have new information to add on the cable front.

I purchased 1m 'Charging and Data' cables before my previous test. The new cables were wimpier than the old ones despite the illustration suggesting otherwise, but actually worked somewhat better. After the test, I purchased 0.5m cables and 0.25m cables all sold as 'Charging and Data' cables.

I have now tested their loop resistance. The values are at the limit of my tester and so are somewhat of a guide.

Old 1m cable    0.6Ω

New 1m cable   0.6Ω

0.5m cable        0.2-0.3Ω over a number of measurements

0.25m cable      0.2Ω

The wire gauge of the old cable was not shown, but all three of the new cables were marked 28AWG. I suspect the cable came from the same cable factory even if the assembly factories were different.

The quality of my short-circuit was not great (a screw-driver gently inserted into the micro USB plug). I know that this could easily have added up to 0.05Ω if not more and so the accuracy of the results drops quickly with the length of the cable. All the measurements were taken several times and only the 0.5m cable showed any variation.

The conclusion is that a 1m USB power cable is likely to be on the verge of causing the rainbow square voltage warning when wireless and wifi USB dongles are used. Use a powered hub if using a cable of 1m or above, although you might just get away with a good 1m cable with some devices.

I am using a 0.5m cable without any signs of problems and without a powered hub. You might well be able to use all four USB ports as long as power hungry devices are not involved. A 0.25m cable does not give a significant advantage over a 0.5m cable, but might still be useful in some circumstances as when running off a mobile device charging battery. As the battery pack discharges and the output voltage drops, the shorter cable might give you a useful extension in operational time.

If you look carefully at USB cables you will find that they are marked with the gauge of wire they use. Although they vary slightly it is usually in the form of something like: 28AWGx1P+26AWGx2C.

This means they are using 28 AWG wire for the data connections ( 1 pair) and 26 AWG wire for the power pins. One thing I have noticed, and it seems to be fairly consistent is that longer cables (2 m)

tend to use smaller gauge wire (30AWGx1P+28AWGx2C is typical) than shorter (0.5 m) ones. This means that shorter cables have two obvious advantages when they are used for providing power, they

use larger wire which can handle more current, and they have less resistance because they are shorter.