User avatar
Ed Raket
Posts: 164
Joined: Mon Apr 30, 2012 4:28 pm

Re: USB Port Current Boost (solved)

Wed Sep 12, 2012 9:59 pm

jbeale wrote:Just for clarity, can I confirm the problem device is the "EDIMAX EW-7811Un" as for example this ?
Yes, that's our "testcase" adapter ;) Verry nice example you lay out here, it shows the voltage drop on the Pi's is rather drastic and explanes the PCB resetting.

toxibunny
Posts: 1383
Joined: Thu Aug 18, 2011 9:21 pm

Re: USB Port Current Boost (solved)

Thu Sep 13, 2012 2:37 pm

alexeames wrote:I just "did the deed" on one of my Pis. Glad to say it survived the heat treatment, although it ain't pretty. Look at those huge gobs of solder :lol: ...
Image

The sticky keyboard issue (that only arose after switching to Raspbian and overclocking to 900) stopped immediately. 8-)
I just did that too, and mine is a hell of a lot uglier than yours. Raspi seems to have survived, though It didn't do what I hoped, which was to allow me to plug in my USB HD without a powered hub. Your photo is very misleading though - that soldering is FIDDLY!

Edit: damn, should have read the thread more carefully! USB HD grumble grumble...
note: I may or may not know what I'm talking about...

User avatar
mahjongg
Forum Moderator
Forum Moderator
Posts: 9402
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands

Re: USB Port Current Boost (solved)

Thu Sep 13, 2012 2:55 pm

Yes, due to the input polyfuse that limits the total input current to something between 0.75 and 1.5A, (depending on ambient temperature, at room temperature it will probably be slightly over one Amp) you cannot feed a harddisk directly through the PI, even if your PI's power supply can handle the current, and you use a rev2 board, or a board that have the USB polyfuses bridged.
The main reason is that during spin up the harddisks motor may use in excess of one Ampere!

About the hot-plug issue, note that having a small resistance large enough to prevent all and any hot-plug problems might also prevent successful back-feeding of the PI by the hub. And note that when you have a powered hub you can hot-plug devices in the hub, as the hub should prevent hot-plug issues.

GeorgeIoak
Posts: 27
Joined: Thu Aug 30, 2012 1:50 am

Re: USB Port Current Boost (solved)

Tue Sep 25, 2012 12:22 am

I've read through this thread and thought about everyone's assumptions but I'm surprised that nobody has hooked up a scope and actually measured the power rails to see what is actually happening rather than just talk about theories. Did I miss any actual measurements in some other threads perhaps?

I've got a "newer" Rev 1 Pi with the zero ohm resistors in place and can confirm that when inserting a USB thumb drive which reports only requiring 100mA that the board will reboot most of the time. I've used a 5V/3A supply as well as using a bench top power supply with the same results. What I found interesting is that if you power the board through the GPIO header you will take F3 out of the circuit. I got the same results powering from the header so I don't think F3 is the "main" cause of the problems.

I did add a 100uF Al Electrolytic across C32 and that seems to eliminate the reboots on USB insertion but about 50% of the time the USB thumb drive is not enumerated. I don't recall the USB spec details but I do know that it states a maximum amount of capacitance on the USB_5V but I also know I have a few designs with more than 2X the spec and never had any issues.

If nobody has taken any actual measurements I can hook up the scope and see what I find but it seems odd if this hasn't been done since it should be the first thing to do to find the root cause.

r_leyland
Posts: 3
Joined: Thu Sep 27, 2012 12:51 am

Re: USB Port Current Boost (solved)

Thu Sep 27, 2012 1:10 am

We tried overclocking our RPi and got the famous reset problem.

I measured the voltage on the +5 rail, and watched it drop to 4.7V when we fired up some graphics. At which point the CPU reset. The RPi was drawing 0.46 Amps.

I measured the resistance of the polyfuse F3 at 1.5 Ohms, which seems awfully high. What should the resistance be? (when cold, i.e. no current flowing through it)


thanks,
Robert.

-- at 1.5 Ohms, and 0.45 A, that would drop the voltage 0.72V and of course the RPi would reset!

GeorgeIoak
Posts: 27
Joined: Thu Aug 30, 2012 1:50 am

Re: USB Port Current Boost (solved)

Thu Sep 27, 2012 3:04 pm

I'm not convinced that it's the 5V rail that causes the CPU to reset. I can't find any details about how the BCM2835 uses the 5V rail but the connections are labeled VDD_BAT1 - VDD_BAT4 so I would think that it's just monitoring this rail for when it's used in a battery application and a drop from 5V to 4.5V shouldn't assert a reset condition.

I'm leaning towards perhaps the LAN9512 doing something on the 1.8V rail which back feeds into the BCM2835 causing the reset. The 1.8V rail is tied to PLL_1V8 of the BCM2835 and typically the PLL circuits need to be really clean.

I didn't see anything on the scope but I might take a closer look at this. I haven't seen any details on the changes for the Rev 2.0 boards so I'd be curious to know what they did and if it changes any behavior.

I hope in the Rev 2.0 boards they removed R36 & R37. I see how they added those for when they don't populate the LAN9512 but adding that option is going to cause havoc with the controlled impedance D+/D- lines. Since they're on the "host" side it should impact all communication but I wonder if the B port has more problems than the A port for people reporting poor USB communications.

Is this the right forum to post technical type questions? I'm a bit surprised that my initial questions haven't been commented on yet.

User avatar
Burngate
Posts: 4861
Joined: Thu Sep 29, 2011 4:34 pm
Location: Berkshire UK
Contact: Website

Re: USB Port Current Boost (solved)

Thu Sep 27, 2012 5:54 pm

GeorgeIoak wrote:I hope in the Rev 2.0 boards they removed R36 & R37. I see how they added those for when they don't populate the LAN9512 but adding that option is going to cause havoc with the controlled impedance D+/D- lines. Since they're on the "host" side it should impact all communication but I wonder if the B port has more problems than the A port for people reporting poor USB communications.
R36 & R37 aren't fitted on model B. They're 0 ohm, and required for model A.
GeorgeIoak wrote:Is this the right forum to post technical type questions?
Yes
(as far as I'm concerned - what other people think I've no idea!)

r_leyland
Posts: 3
Joined: Thu Sep 27, 2012 12:51 am

Re: USB Port Current Boost (solved)

Thu Sep 27, 2012 5:56 pm

@George - what is the resistance of the polyfuse (F3) in your RPi ?

The 1.8V controller is just an MCP1117 so it has no extra lines to trigger a reset. Are you thinking that something else outside the BCM2835 is forcing a reset?

Is the actual datasheet for the BCM2835 available? I found the "software" description, but not an electrical level spec.


thanks,
Robert.

GeorgeIoak
Posts: 27
Joined: Thu Aug 30, 2012 1:50 am

Re: USB Port Current Boost (solved)

Thu Sep 27, 2012 6:46 pm

@Burngate: I know that R36 & R37 are not mounted but they still create a stub on the controlled impedance lines which will result in reflections. I personally don't always follow the USB spec's every detail but nevertheless this is something that will cause problems, it's just a matter of whether they are enough to result in end users having issues. I recently sold my USB analyzer so I can measure the numbers of NAKs thrown during communication but I'd be interested in seeing if there are more than normal.

@r_leyland: I have one of the "newer" Rev 1.0 boards which does not have polyfuses mounted but instead zero ohm resistors. Others have mentioned seeing some improvement when adding a small resistance or a higher current rated polyfuse and I've ordered some parts for testing on my own. What these would do is help limit the inrush current when a USB device is plugged in but again I wonder if even that condition would warrant a CPU reset.

The 1.8V regulator is supposed to power the SDRAM as well as provide 1.8V to the BCM2835. Since the LAN9512's 1.8V output was incorrectly tied to this rail any disturbance on this rail could cause problems for the BCM2835 which results in a reset. Typically 1.8V rails are specified to have tight tolerances and should have very little noise. Another test would be to lift the 1.8V output pin on the LAN9512 which would be similar to what others have done by cutting a trace on the board.

My problem/concern is that I have 30 of these boards which we were going to use for a project part of which was to use the USB port to update files. I incorrectly assumed that if we used a 10W (5VDC/2A) power supply and fed it through the GPIO pins that we would be able to use the USB ports like in any other "embedded" board.

As others have mentioned, the cost of 2 polyfuses is about the same as adding in an active USB controlled switch and that could have prevented limiting how the USB ports are/can be used. Either way, I don't think my project is going to work without some modification to the RPi.

r_leyland
Posts: 3
Joined: Thu Sep 27, 2012 12:51 am

Re: USB Port Current Boost (solved)

Thu Sep 27, 2012 7:29 pm

Hi George,

My boards are also later RPis with 0-Ohm resistors at the USB side

What I am seeing is a reset condition when I step up the clock rate, and draw more current. This appears to be caused by either the SMPS detecting an overcurrent situation, or the polyfuse on the power input side causing a voltage drop. (it measures 1.5Ohms) - which will cause a pretty significant voltage drop if the current drawn either by the SoC or downstream USB devices goes up.

We have about 10 RPis for a project. Our USB device draws about 220mA so we are concerned about this too

cheers,
Robert

GeorgeIoak
Posts: 27
Joined: Thu Aug 30, 2012 1:50 am

Re: USB Port Current Boost (solved)

Thu Sep 27, 2012 8:00 pm

Hi Robert:

Can you try to power your board from the GPIO header rather than the micro USB. This will bypass the main polyfuse and take that out of the picture.

George

GeorgeIoak
Posts: 27
Joined: Thu Aug 30, 2012 1:50 am

Re: USB Port Current Boost (solved)

Thu Sep 27, 2012 9:58 pm

BTW, before someone calls me a dummy, I'll do it myself. Kind of hard to lift a leg on a QFN package! Whoops, I forgot the LAN9512 was a QFN...

p4trykx
Posts: 127
Joined: Wed Jan 11, 2012 2:55 pm

Re: USB Port Current Boost (solved)

Sat Oct 27, 2012 10:50 pm

Just want to report my experience with 3G modem Sierra Wireless 307. After shorting F1 and F2 it was working only when plugged in in from the beginning. Hot plugging causes LAN LEDs go off. So I added 22uF electrolytic capacitor to C32 but it didn't help. Currently I don't have 1 Ohm resistors but instead I connected the modem through 1,5 meter cable so the inrush current would be limited. It kinda works like those 1 Ohm resistors.
The 3G modem seems to work fine however it's not using full TX power. I will have to test it in worse signal conditions.

Looking at the schematics C32 seems to connected after the polyfuses(1Ohm resistors)
USB---resistors--C32
Wouldn't it be better to add some caps between the USB and the resistor?

USB --aditional cap -- resisotrs - C32

p4trykx
Posts: 127
Joined: Wed Jan 11, 2012 2:55 pm

Re: USB Port Current Boost (solved)

Wed Oct 31, 2012 8:42 pm

As advised I soldered 1 Ohm smd resistors over the polyfuses. I got 1206 which are a little to short but these where the biggest avaliable in my local shop.
I also have added ceramic capacitors directly under USB connector. The are directly connected and connected to the rest of the board through 1 Ohm resistors. I used 47nF ones.
As mentioned earlier I added C32 electrolytic capacitor to C32 but alone(no resistors) it was not enough.


Image
https://dl.dropbox.com/u/96422972/raspi/DSC01305.JPG

Image
https://dl.dropbox.com/u/96422972/raspi/DSC01306.JPG

hcjehg
Posts: 9
Joined: Mon Jan 09, 2012 8:56 am
Contact: Website

Re: USB Port Current Boost (solved)

Tue Nov 06, 2012 1:37 pm

p4trykx wrote:As advised I soldered 1 Ohm smd resistors over the polyfuses. I got 1206 which are a little to short but these where the biggest avaliable in my local shop.
I also have added ceramic capacitors directly under USB connector. The are directly connected and connected to the rest of the board through 1 Ohm resistors. I used 47nF ones.
As mentioned earlier I added C32 electrolytic capacitor to C32 but alone(no resistors) it was not enough.
And with this modification, how is you pi doing with higher load usb devices? Working perfectly?

p4trykx
Posts: 127
Joined: Wed Jan 11, 2012 2:55 pm

Re: USB Port Current Boost (solved)

Tue Nov 06, 2012 1:58 pm

I only checked with 3G modem it consumes a lot of energy(it's hot ) when connected and downloading. I was running it for 24h and continuously downloading something and it runs stable.
However I'm using a free internet provider Aero2 so the bandwidth is limited to 512kb/s.

dobieg
Posts: 2
Joined: Fri Sep 07, 2012 12:21 am

Re: USB Port Current Boost (solved)

Sat Apr 20, 2013 1:06 am

Fantastic! - been trying to get a noname mini-USB wireless dongle working on my RPI for weeks, but always thought it was my software config that was wrong, I really didn't want to go down the path of a 'powered hub' as that would almost certainly involve two plugs.

The 'red wire' solution worked for me, using a 1000ma power supply, all I wanted to do use it for was a 'headerless' board remotely connecting from my PC, so with no HDMI or keyboard/mouse connections it now allows me to stick the Pi anywhere in range of my wireless LAN.

husaini
Posts: 5
Joined: Thu Mar 14, 2013 4:51 am

Re: USB Port Current Boost (solved)

Sun Apr 28, 2013 11:20 am

what can i do/mod or prevent when i insert wifi stick while the raspberry pi running it will reset? im using rev 2 board

dom
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 5091
Joined: Wed Aug 17, 2011 7:41 pm
Location: Cambridge

Re: USB Port Current Boost (solved)

Sun Apr 28, 2013 11:38 am

husaini wrote:what can i do/mod or prevent when i insert wifi stick while the raspberry pi running it will reset? im using rev 2 board
Using a powered hub is the obvious solution.

Without a hub, using a usb extender cable may help.

User avatar
abishur
Posts: 4478
Joined: Thu Jul 28, 2011 4:10 am
Location: USA
Contact: Website

Re: USB Port Current Boost (solved)

Sun Apr 28, 2013 6:11 pm

husaini wrote:what can i do/mod or prevent when i insert wifi stick while the raspberry pi running it will reset? im using rev 2 board
placing a nice capacitor over by the USB ports should help if I'm remembering discussions from that time frame correctly :-)
Dear forum: Play nice ;-)

p4trykx
Posts: 127
Joined: Wed Jan 11, 2012 2:55 pm

Re: USB Port Current Boost (solved)

Sun Apr 28, 2013 9:52 pm

In my case a extender ~1,5m cable did the job. I also added capacitors(on electrolytic and two smaller caramic directly to usb port) but it wasn't enough.

andrewr
Posts: 5
Joined: Sat Dec 21, 2013 4:48 pm

Re: USB Port Current Boost (solved)

Sat Dec 21, 2013 4:54 pm

I have an older RPI with the polyfuses. I'd like to bridge them with 1 ohm resistors to use with an edimax wifi and also usb hard drive (external power). Can someone mention where they purchased 1 ohm resistors? I can seem to find them anywhere. Also, is this still the best accepted solution? Thanks

jbuehl
Posts: 29
Joined: Fri Jun 08, 2012 2:49 pm
Location: Los Angeles

Re: USB Port Current Boost (solved)

Sat Dec 21, 2013 6:12 pm

You could try newark.com.

Return to “General discussion”

Who is online

Users browsing this forum: Bing [Bot], Heater, jopitrepls, Lomax and 19 guests