Raspberry Pi Forums

I have two (I think identical) Pi. One of them connects to a network without issue, whether it's wired or wireless.The other just will not connect under any circumstances. Same peripherals, same distro, interface looks up, but will not get an IP address. I get the following several times during Boot and if I run ifup eth0
DHCPDISCOVER on ETH0 to 255.255.255.255 port 67 interval x
x changes each time.

Followed by
Unable to obtain lease first time, failed to bring up eth0

I have tried all sorts including starting with a brand new Wheezy install, installing a new Wheezy from scratch on the working board, getting the network running, then moving the SD card to the not-working board.
I'm struggling to accept that the board is faulty, given that the error appears on both the built in network port AND a USB wireless dongle.
Does anyone have a suggestion?

Madrilleno

Have you swapped the power supplies?

Power supply - yes. Three different ones, 1A, an ipad charger (1.6A I think) and a 3.1A beast. Also using a powered USB hub.

Reading up, I think it could be related to the buitl in USB on the Pi. I understand that the on-board wired Lan port is actually implemented as a USB device. This is one explanation which woud give me the failure on both the wired and wireless coneections, but I still have difficulty beleiving it.
Running lsusb shows all devices connected and enumerating correctly.
Attached is a screenshot of the relevant NW errors.
Madrilleno

If you have a voltmeter could you check the voltage reading between TP1 / TP2 what are you getting??

Richard S.

TP1 -> TP2 = 4.59v

Madrilleno wrote:TP1 -> TP2 = 4.59v

and on your other PI?
but that looks too low TBH

Working Pi = 4.60v

Have also tried using a static IP address. Obviously no errors about DHCP on boot up, but still no connectivity. Error message reports detination address unreachable through correct interface.

try another [see fatter or shorter] power lead - you are too low
5% of 5V is 4.75V

RaTTuS wrote:try another [see fatter or shorter] power lead - you are too low
5% of 5V is 4.75V

Took the USB lead out of the equation and powered using a 5v 2A supply onto the GPIO pins as suggested http://elinux.org/RPi_Hardware#Power_Supply_Problems.

TP1/TP2 now read 5.07v.

Still will not connect to the network.

Madrilleno

just some questions to get a better picture of the situation:

The usb ports work as you can connect and use a keyboard or a mouse to them?
The ethernet leds light up normally if you connect the cable?
When you run lsusb, you can see all connected usb devices? Is it possible to post the results of this?
When you use the same setup as your working Pi (wifi stick, sd card, power supply and supply cable),
you still don't obtain a valid dhcp on it?
Do you have other usb devices connected, besides the wifi dongle if you test wireless?

You could try to reduce the usb speed to usb 1.1 and see if that makes a difference.
add
dwc_otg.speed=1
to the end of the /boot/cmdline.txt file and reboot to achive that
You could try to disable your ethernet full duplex operation by adding
smsc95xx.turbo_mode=N
to the same file. (and also a reboot)
This aren't solutions, and the second setting shouldn't make a difference for wireless operation. The results of those tests just might help figuring out what is going on.

Yes, the USB ports work, have tried both.
Yes, ethernet lights on the non-connect board light up in the same way as the connecting board.

Results of lsusb:
Yes - same setup as connecting card, no connection on non-connecting.
The only connections to my Pi are power, HDMI out (to DVI adaptor on monitor), one USB port to a powered USB hub (keyboard, mouse & Wif dongle attached) and a Cat5 patch lead.

Have also tried plugging the patch lead into different switches to eliminate that.

Tried your USB and duplex suggestions - no change

Madrilleno

It's not easy to read, but I can't seem to find your keyboard.
I also noticed that you have the hub appearing twice. I assume it's a 7 port hub internally connected as 2 cascaded 4 port hubs?
If you disconnect the usb hub and all other usb stuff, does your wired network connection work like that?

The only other thing I can still suggest is updating the os to the latest version on the working pi, to see if it makes a difference. I have no idea what else could be causing this.

obcd wrote:It's not easy to read, but I can't seem to find your keyboard.
I also noticed that you have the hub appearing twice. I assume it's a 7 port hub internally connected as 2 cascaded 4 port hubs?
If you disconnect the usb hub and all other usb stuff, does your wired network connection work like that?

The only other thing I can still suggest is updating the os to the latest version on the working pi, to see if it makes a difference. I have no idea what else could be causing this.

Yes, it's a seven port hub. I think the KB may be the Holtek device, but I don't really know.
I have another Pi on order with Farnell. When that turns up I will give it a whirl and let you know what happens. I really am beginning to suspect the Pi board itself now.

Just to note, the 'no load' voltage of my power supply is 5.07v. It gets pulled down to 4.6v on both boards when they are running. I also have another power supply on order.

In the meantime, thanks to everyone for your posts so far.

4.6V is low for the Pi. For stable operation it should remain between 4.85 and 5.25V.
Maybe you can try to power it from one of the ports of your hub, or you might just try the other supply when it arrives.

You definitely have a problem with the USB lead. I suggest you go out and buy a phone charger cable from a phone shop. What you have is probably only designed for data and it may cause you problems in the future even with your working board.

It is very possible that your two boards are differently sensitive to the reduced voltage. But that doesn't explain your problem, given that it still fails with power supplied to the GPIO pins and 5.07V between TP1 and TP2. It sounds to me like it is a faulty RaspPi.

Have you tried using the SD card from your working board in your non-working board? That would be the only remaining test that I can think of. There's other tests that I could do, such as monitoring the network traffic and watching the DHCP packets, but that is not something to be done by someone who doesn't understand networking protocols. Anyway if two identical boards behave differently with identical software then they are not identical boards -- one of them is broken. So if swapping the SD cards doesn't get you working then I think you should return the board as faulty.

rurwin wrote:You definitely have a problem with the USB lead. I suggest you go out and buy a phone charger cable from a phone shop. What you have is probably only designed for data and it may cause you problems in the future even with your working board.

It is very possible that your two boards are differently sensitive to the reduced voltage. But that doesn't explain your problem, given that it still fails with power supplied to the GPIO pins and 5.07V between TP1 and TP2. It sounds to me like it is a faulty RaspPi.

Have you tried using the SD card from your working board in your non-working board? That would be the only remaining test that I can think of. There's other tests that I could do, such as monitoring the network traffic and watching the DHCP packets, but that is not something to be done by someone who doesn't understand networking protocols. Anyway if two identical boards behave differently with identical software then they are not identical boards -- one of them is broken. So if swapping the SD cards doesn't get you working then I think you should return the board as faulty.

I'm inclined to agree that it's a board problem. I am au-fait with packet monitoring, but to be frank, I haven't bothered. I can guess that the traffic will show either incomplete outgoing packets, or ignored replies. Either way, it doesn't take me any further down the road to a diagnosis - I know that the problem doesn't lie in the connection between the patch cable, my network and my DHCP server since it works with all other devices.

I take your point about data vs charger, and will probably give it a go anyway, but given the results with power on the GPIO pins directly, with no voltage drop under load, it's unlikey to alter the result.
I have tried the SD card from my working board in the non-working one early on in the test process; I must have forgotten to mention it. Same result - no connection.

New board arrived from Element14 this morning. Plugged in, network connected, test voltage 4.50v. As you can see, the DHCP lease wasn't taken at the first bite.
New power supply will arrive this afternoon. Will test that and post result.

New power supply. Pull down voltage now 4.73v-4.75v so marginal, but better. Same result on non-connect board.

To close this post:

Another power supply received this morning, advertised at 5.2v, no load voltage was actually 4.8v, and dropped to 4.4v when the RPI was running with minimal peripherals plugged in. All I had was HDMI to a monitor, NW cable and a powered USB hub. The KB and mouse were pugged into the hub.

These are the power supplies I have tried:-

Best so far
Runs at 5.0v no load, 4.75v loaded
http://www.amazon.co.uk/gp/product/B008 ... 01_s00_i00

Take care with this one, it's advertised at 3.1 amp, but it's split into 2.1A and 1.0A sockets. Ran at 4.8v no load, 4.4v loaded
http://www.maplin.co.uk/3.1a-usb-power-supply-629936

This one also ran at 4.8/4.4v
http://www.maplin.co.uk/micro-usb-power-supply-393067

I would be glad to hear from any UK users who have found a power supply which actually supplies better than my best one so far, and is obtainable in the UK (i.e. don't have to wait four weeks for shipping from the likes of Hong Kong).

I said I was closing the issue, but here I am, posting again.

As a test, I have removed the USB connector from the end of my latest power supply, and connected it directly to the 5v and ground pins on the GPIO connectors. I now get 4.70v under load on the RPi rather than the 4.4v I got with the same power supply through the micro-usb connector. The result is that I am loosing 300mv in the circuit from the USB connector to TP1/TP2. I suspect this is probably the polyfuse since it seems to be consistant across all three RPi which I have tested.

Next, I'm building a regulated 5v power supply from scratch to see what that gives me.

I used to get that with my Realtek RTL8188. I updated Debian and all of a sudden it began to work. It's annoying, but it worked, so no complaints there.

So here http://tiny.cc/dbpjlw is my prototyped 5v power supply based on a 7805 voltage regulator. The design is modified from this page : http://www.eleinmec.com/article.asp?16 leaving out the rectifier circuit since I used a 12vdc 1.25A source.
No load voltage was spot on 5.00v, and running under load gave a rock steady 4.90v with 500mA.
My no-connect board is still a no-connect board, and I beleive I have now tested sufficiently to say it's faulty.
Even tried an older Debian release as suggested by Genius23008 - same result.

Revceived this power supply this morning http://www.amazon.co.uk/gp/product/B008 ... 02_s00_i00

Runs at 4.40v.

It seems you are loosing about the amount of voltage I would expect if you have a non recovered F3 polyfuse. Please measure across the polyfuse (F3 on the back of the board) to see if you are losing a good part of a volt over it. It would explain why you cannot get the voltage level between TP1 and TP2 to a reasonable level, that is at least 4.85V.