LOGIK LP4HUB10, TP-LINK TL-WN722N, Logitech K260 conflict

[FishNChipPapers]

Am running with the Debian 6.0.4 OS and a combination of verified peripherals

Working configuration: Pi, LOGIK LP4HUB10 powered 4-port hub, TP-LINK TLWN722N wireless USB adapter; Lenovo KU-0225 wired keyboard, Microsoft optical wired mouse 500 V2. All peripherals connected to the hub.

Replace the wired keyboard and mouse with Logitech K260 wireless keyboard/mouse combo following clean shutdown; disconnect power; reconnect power.

Recognises hub; wireless adapter; Logitech USB receiver

Reaches point of 'Configuring network interfaces ...'. TP-LINK green light does not flash and just appears to hang.

Restart. Does not recognise the wireless adapter

dmesg shows the following:

usb-1-1.2.2: ath9k_htc: transferred htc_9271.fw, size 51272
ath9k_htc 1-1.2.2:1.0: ath9k_htc: Target is unresponsive
Failed to initialize the device
ath9k_htc: probe of 1-1.2.2:1.0 failed with error -22
usbcore: registered new interface driver with ath9k_htc
EXT4-fs (mmcblk0p2): re-mounted. Opts: (null)
fuse init (API version 7.17)

Have tried connecting the Logitech receiver directly to the Pi with wireless via hub; wireless adapter connected to the Pi with Logitech receiver via the hub.

Thoughts welcome.

[FishNChipPapers]

Having invested in a multimeter I have been testing the variation in voltage with various configurations:

1) Hub with wired keyboard, mouse and WLAN adapter - 4.65v - working configuration
2) Hub with wireless keyboard/mouse USB dongle and WLAN adapter - 4.64v - non-working configuration
3) As 1) but with no power to the hub - 4.38v - non-working configuration
4) As 1) with WLAN in 2nd Pi USB slot - 4.59v
5) As 2) but with wireless keyboard/mouse USB dongle in 2nd Pi USB slot - 4.64v

1) in comparison with 4) suggests that the WLAN does reduce voltage when connecting directly to the Pi
1) in comparison with 5) suggests that the wireless keyboard/mouse dongle does not reduce voltage when connecting directly to the Pi

Conclusions
1) First things first I need to look at the power supply to the Pi based on the information provided here: http://elinux.org/R-Pi_Troubleshooting# ... r_problems as it is below 4.75V (although it should be noted that configuration 1) works). I am currently using an iPhone/iPad adapter with UltimateAddons Extra Long 2 metre Data and Charge USB Cable for Samsung Galaxy S2 S II 2 i9100 Smart Phone. (http://www.amazon.co.uk/gp/product/B005 ... 01_s00_i00). Any recommendations would be welcome.
2) Once I have resolved 1) to eliminate power as an issue it should determine whether the issue is down to a conflict between the 2 wireless devices. The comparison of configurations of 1) and 5) suggest there is some sort of conflict as the voltage in both cases is the same but that could be because I am at the lower limit of voltage already.

[FishNChipPapers]

I performed one further test: disconnected the hub completely and just had HDMI and wired keyboard connected. 4.58v.

[FishNChipPapers]

And finally ...

Changing the power cable - not the adapter - gave me 4.78v which is within the acceptable range defined here: http://elinux.org/R-Pi_Troubleshooting# ... r_problems. However, this still doesn't allow the wireless keyboard/mouse and WLAN adapters to coexist.

[mahjongg]

Maybe 4.78 volt is "within the margin", but IMHO you need something like 4.9 Volt, just 4.78V is really stretching things to the limit. In fact what you really want to know is what the USB device receives, maybe the SoC in the PI can work with 4.78 V, but the USB polyfuses can introduce another significant dip, below what the USB device will work with. An 100mA device "should" work with only 4.40 Volt, but many other USB devices also need 4.75V, and won't get it due to the polyfuses.

measure the voltage on pin 1 of the USB connector (either on the solderpin of the USB connector, or on the side of the polyfuse connected to that pin).

[FishNChipPapers]

Thanks for the response. Sorry for what is perhaps a dumb question (I'm not a sorcerer's apprentice - never mind electronics wizard): how do I
"measure the voltage on pin 1 of the USB connector (either on the solderpin of the USB connector, or on the side of the polyfuse connected to that pin)."

[FishNChipPapers]

Maybe 4.78 volt is "within the margin", but IMHO you need something like 4.9 Volt, just 4.78V is really stretching things to the limit

With the Apple 5V 2.1A USB charger for iPad2, model A1357 (one of the verified power adapters), a Farnell-supplied " HK-US10/1 LEAD, USB A MALE-MICRO B MALE, 1M" and nothing else connected I am getting between 4.76V and 4.78V. Any thoughts on how I can get up to 4.9V? Are there better power adapters? cables?

[therealeasterbunny]

Can also report I have a LOGIK LP4HUB10 and this puts through 4.79v (or 4.78v with a mouse, kbd and USB to serial converter attached. Pi runs fine though. Not noticed any issues. However if I do, I will be taking it back to Currys as 4.8v is not 5v as it says on the tin. Nevertheless, £12 for a LOGIK LP4HUB10 which seems to work for me is OK. I don't run wifi but wired ethernet.

Also used to use my wife's HTC desire charger which supplies 5.08v to the Pi.

Tested on TP1 and TP2 testpoints using digital multimeter.

[FishNChipPapers]

The LOGIK hub works fine for me with USB keyboard (Lenovo), USB mouse (Microsoft) and wireless adapter (TP-LINK) and iPad power adapter. Replacing the USB keyboard and mouse with the Logitech wireless keyboard/mouse combo prevents the wireless adapter from functioning. Given that I am only getting 4.78v with the iPad power adapter with nothing else connected, I wonder whether replacing the power adapter with the HTC power adapter which you have shown suppliers over 5v.

[mahjongg]

Thanks for the response. Sorry for what is perhaps a dumb question (I'm not a sorcerer's apprentice - never mind electronics wizard): how do I
"measure the voltage on pin 1 of the USB connector (either on the solderpin of the USB connector, or on the side of the polyfuse connected to that pin)."

With a multimeter with the minus lead connected to testpin TP2 (the GND pin) measure on each side of one of the polyfuses (the green components between the LED's and the USB connector), one side will give the same reading as TP1 (5V) but the other side will give a lower reading, (but only if its a fuse that is for an USB connector that draws some current otherwise both sides will measure the same, F1 is for the upper, and F2 is for the lower connector of the two).

Its also possible to measure between the two pins of a polyfuse, that will show directly how much "dip" the fuse generates. Lastly with the PI unpowered, and nothing plugged into the USB ports, you can measure the resistance of the polyfuses with a multimeter set to resistance measurement. The polyfuses (F1 and F2) should measure a few ohms, if you are measuring about 5.0 Ohms, its "normal" (normally unacceptable for an USB polyfuse, but normal for a PI) but some people report values as high as eight ohm, which will give a dip of 0.8 volt when a device draws 100mA, leaving only 4.2 volt for the poor USB device, even when the voltage between TP1 and TP2 is exactly 5.0 Volt (SNIFF ).
Polyfuse F3 (on the back near the power connector) should be a much more reasonable 0.1 Ohm or so.

[FishNChipPapers]

Thanks once again. Here's what the testing reveals:

TP1 voltage: 4.78/9V

Nothing in the USB
F1: 4.79V
F2: 4.79V

Powered USB hub in the top USB slot
F1: 4.79V and 5.04V
F2: 4.79V and 4.79V
F1 between the 2 pins: 0.25V
F2 between the 2 pins: 0V

Powered USB in the bottom USB slot
F1: 4.79V and 4.79V
F2: 4.79V and 5.04V
F1 between the 2 pins: 0V
F2 between the 2 pins: 0.25V

Resistance
F1: 7.2 Ohms
F2: 5.4 Ohms
F3: 1.3 Ohms

These (apart from the 1.3 Ohms measurement for F3) are broadly consistent with the measurements you describe. So, what are my options?

[mahjongg]

Two things:

The hub tries to "backfeed" power to the PI, (that is why there is 5.25 V from the hub on the polyfuse) which isn't bad, as long as its not the only thing that is powering the PI, due to the polyfuse the RPI will then receive a very marginal power. and the polyfuse (F1 or F2) might even start to trip. This might lead to SD-card corruptions. This "backfeed" discovery also means that this hub will try to extract its power from the PI.

At first glance your PSU seems far too weak, outputting just 4.78 volt. But this clearly has something to do with polyfuse F3, with a typical current of 350mA a 1,3 Ohm fuse will give a dip of 1,3 x 0.35 = 0.45V ! That means that the PSU is actually supplying 4.78V + 0.45V = 5.23 Volt!

Perhaps you can check that by measuring the voltage across F3.

It seems that during your experiments the F3 polyfuse has tripped, and hasn't fully recovered.
This perhaps could have happened when you powered on the PI sooner than the hub, and so the PI tried to also power the hub, which could have overheated/tripped F3.
Recovering from being tripped can take days, and in fact it will never completely recover to its initial low state, but good enough if you are patient.

Also especially F1 is fairly high, probably because it has heated up a bit due to the hub attempting to backfeed the PI.

My advice, let the PI rest in a cool place for a couple of days, after that if you use the PSU plug it in a switchable extension cord, together with the PSU of the hub, so that you switch on both at the same time. Alternatively power the PI from a port of the hub provided the hub can deliver at least 500 mA to the PI.

[FishNChipPapers]

Wow. Thanks for the comprehensive and very helpful response. I shall do as you advise and report back the results.

[siriusfox]

I hate to oppose the thought process here, but I'm inclined to think that this is not a problem with voltages at all. I'm running a similar setup (TP-Link WN722N with a K300 keyboard) on a different brand of hub, and I'm experiencing basically the same problems you are. Thus-far your errors match mine exactly.

Notably though I'm running an Astrodyne power supply, and my supplied voltages are up in the 4.9v range as needed, and I'm seeing no change in the voltages when the devices are plugged in. The first device I connect will properly detect and work, but the later will spit some sort of error into dmesg.

That being said, this likely isn't a core Linux issue as I can use these two devices simultaneously on other computers without any problems. I am inclined to think this is a problem with the specifics of the usb software stack we're using, as until the recent firmware updates that came with the Debian wheezy image, my hub brought the Ethernet chip to a halt.

[FishNChipPapers]

I am inclined to think this is a problem with the specifics of the usb software stack we're using, as until the recent firmware updates that came with the Debian wheezy image, my hub brought the Ethernet chip to a halt

Thanks. Am I right in reading your response that the firmware updates provide a working configuration? If so, how do I get hold of those firmware updates?

[siriusfox]

Thanks. Am I right in reading your response that the firmware updates provide a working configuration? If so, how do I get hold of those firmware updates?

Sorry but no. The firmware update that comes bundled with the latest Debian release on the web site (posted two days ago) fixed by usb hub issues, but has had no effect on conflicts between the wifi adapter and the keyboard.

[FishNChipPapers]

My advice, let the PI rest in a cool place for a couple of days, after that if you use the PSU plug it in a switchable extension cord, together with the PSU of the hub, so that you switch on both at the same time. Alternatively power the PI from a port of the hub provided the hub can deliver at least 500 mA to the PI.

I waited 4 days; used a switchable extension and powered on the Pi and the hub simultaneously. With the hub in the top USB slot, the voltages are as follows:

• TP1/TP2 - 4.78V
  F1 - 0.26V
  F2 - 0.00V
  F3 - 0.10V

The wireless adapter and wireless keyboard/mouse still can not operate together.
Resistances are as follows:

• F1 7.3 Ohms
  F2 5.5 Ohms
  F3 1.1 Ohms

[FishNChipPapers]

Sorry but no. The firmware update that comes bundled with the latest Debian release on the web site (posted two days ago) fixed by usb hub issues, but has had no effect on conflicts between the wifi adapter and the keyboard.

Thanks for clarifying.

[greypower]

@FishNchips

I would suggest you run rpi-update on your existing Debian. Getting the latest updates cleared my problems with a wireless adaptor similar to yours.

Cheers
Richard

[FishNChipPapers]

Getting the latest updates cleared my problems with a wireless adaptor similar to yours.

I will give rpi-update a try. Were you problems related to the coexistence of a wireless adaptor and a wireless keyboard/mouse? My wireless adaptor works without any problems if I am using a wired keyboard and wired mouse; the problems I am encountering are specific to the all wireless combination.

[siriusfox]

Getting the latest updates cleared my problems with a wireless adaptor similar to yours.

I will give rpi-update a try. Were you problems related to the coexistence of a wireless adaptor and a wireless keyboard/mouse? My wireless adaptor works without any problems if I am using a wired keyboard and wired mouse; the problems I am encountering are specific to the all wireless combination.

Correct. When I go wired keyboard my wifi problems vanish.

[FishNChipPapers]

Correct. When I go wired keyboard my wifi problems vanish.

I'll give the rpi-update a go and see whether it fixes my problems

[FishNChipPapers]

Two changes:

1) Utilising an HTC power adapter rather than Apple iPad provided a minor boost in voltage:

- TP1/TP2 - 4.84V
- F1 - 0.19V
- F2 - 0V

2) Ran rpi-update

However, this has still not resolved the issue. With the wireless keyboard and mouse and wireless adapter in the Logik powered hub receive

smsc95xx 1-1.1:1:1.0: eth0: Failed to write register Index errors on startup
INFO: task ifplugd:499 blocked

So, still no joy. Looks like I'm stuck with something wired: network or keyboard/mouse.

[siriusfox]

Just wanted to close this up as best I could. I updated to the latest official Raspbian image today, and all of my conflicts were resolved. I can not use wireless and the keyboard together without any problems!

[FishNChipPapers]

Thanks for the update.

I ran rpi-update without any errors. Unfortunately, I am still not in a position where I can use the wireless adapter and the wireless keyboard/mouse together (irrespective of which I connect to the Pi and which to my powered USB hub or if I take the powered USB out of the equation all together). As soon as I have both connected, the wireless adapter ceases to function.

siriusfox - I believe you have the TP-WN722N wireless adapter but a different keyboard (K300) - is this wired or wireless - hub and presumably mouse. I wonder if this accounts for the differences between our experiences.