Logik LP4HUB10 power problem fixed


30 posts   Page 1 of 2   1, 2
by Newt_Othis » Wed Jun 20, 2012 9:30 pm
I was struggling to get my Logik LP4HUB10 powered USB hub (PC World £12.99) to work properly when I noticed that it appeared to be back-powering the RasPi through the USB socket. When I disconnected the Pi's main PSU (connected via the micro USB port) I noticed that the Pi stayed on, drawing power from the hub.

Other posts have confirmed that this is a *bad thing* as a voltage mismatching may blow the fuse or even cause damage. So, it was necessary to perform some surgery on the hub to prevent this.

***NOTE - This will render the hub unable to draw power over USB so it will permanently require its own PSU to function. Please don't hold me responsible for broken hubs or Pis if you botch it.***

Step 1 - Open the case
image1.JPG
image1.JPG (20.48 KiB) Viewed 6267 times

Thankfully the case is easy to open with a small cross-head screwdriver. Undo the 4 screws and carefully pull the case apart.

Step 2 - Cut the red wire
image2.JPG
image2.JPG (34.15 KiB) Viewed 6267 times

With some small snips, cut the red wire that comes from the board to the USB cable that trails out of the back.

Step 3 - Put it back together

You should now be able to attach the hub to the Pi safe in the knowledge that no power will flow back from the hub.

Thanks to the posters on earlier topics for the help. I wouldn't have thought to do this otherwise.
User avatar
Posts: 20
Joined: Tue Dec 06, 2011 1:12 pm
by mahjongg » Thu Jun 21, 2012 12:33 am
the real problem with hubs trying to push back power to the PI, is that if the main PSU of the PI is off, the hub tries to provide all the power to the PI, but the polyfuses in the PI's USB port will mean the PI will only get a fraction it needs, this causes stability problems, even to the point that the PI will run wild and may damage the files on your SD-card.

with a typical 6 Ohm polyfuse and an average current draw of 350mA by the PI and a hub supplying 5.0 V, the PI will be left with only 5.0 - (.35 x 6) = 5.0 - 2.1 = 2.9 Volt.
User avatar
Forum Moderator
Forum Moderator
Posts: 5752
Joined: Sun Mar 11, 2012 12:19 am
by metaljay » Fri Jun 22, 2012 2:40 pm
just want to say thanks for your post, worked perfect for me :)
User avatar
Posts: 51
Joined: Fri Jun 22, 2012 7:26 am
by Newt_Othis » Fri Jun 22, 2012 3:05 pm
metaljay wrote:just want to say thanks for your post, worked perfect for me :)


You're welcome. It feels like getting 'approved' USB hubs is a bit of a problem for some people. I've heard a few horror stories regarding some of the Belkin hubs.

The LP4HUB10 was nice and easy to obtain, not too expensive and does what I need it to do - after an easy modification.
User avatar
Posts: 20
Joined: Tue Dec 06, 2011 1:12 pm
by michthom » Sun Jul 01, 2012 10:55 am
I've done a mildly improved version of this fix; I added a diode in series with the +5V line, to block current flow up to the host but still allowing the LP4HUB10 to be used as an unpowered hub in other circumstances. The 0.5V forward drop might be an issue in some cases, but I'll take that chance!
Attachments
Diode.jpg
Diode.jpg (60.15 KiB) Viewed 6065 times
Posts: 7
Joined: Sun Jul 01, 2012 10:46 am
by recantha » Mon Jul 09, 2012 8:28 pm
I have the power problem...

I am now cutting... the red wire...

I am now hoping I don't kill my house power...

Plugging in the big USB plug... Nothing! No power! Woo-hoo!
Plugging in the micro USB power from the same hub...

Yes! We have boot!

Thank you!!!! You do, indeed, cut the red wire!
My Raspberry Pi blog with all my latest projects and links to articles
http://raspberrypipod.blogspot.com. +++ Current project: PiPodTricorder - lots of sensors, lots of mini-displays, breadboarding, bit of programming.
Posts: 209
Joined: Mon Jun 25, 2012 10:41 am
by recantha » Mon Jul 16, 2012 8:19 am
The only problem I have now is that, apart from powering the Pi in a basic fashion (i.e. it boots), there now seems insufficient power to run a Wifi dongle and Bluetooth dongle, not to mention a usb flash drive. Kind of disappointing.
My Raspberry Pi blog with all my latest projects and links to articles
http://raspberrypipod.blogspot.com. +++ Current project: PiPodTricorder - lots of sensors, lots of mini-displays, breadboarding, bit of programming.
Posts: 209
Joined: Mon Jun 25, 2012 10:41 am
by itimpi » Mon Jul 16, 2012 9:41 am
recantha wrote:The only problem I have now is that, apart from powering the Pi in a basic fashion (i.e. it boots), there now seems insufficient power to run a Wifi dongle and Bluetooth dongle, not to mention a usb flash drive. Kind of disappointing.

WiFi dongles and USB Flash drives can both take more power than the Pi can provide via its USB ports. Those ports only really provide enough power for basic wired keyboard/mice although bluetooth dongles tend to also work for bluetooth keyboards/mice. The WiFi and USB Flash drive should both work via a powered hub though.
Posts: 1035
Joined: Sun Sep 25, 2011 11:44 am
Location: Potters Bar, United Kingdom
by recantha » Mon Jul 16, 2012 10:05 am
Is it possible that the Pi is trying to supply power to the hub from the USB connection on the Pi (rather than the other way around which is why we cut the red wire)?
My Raspberry Pi blog with all my latest projects and links to articles
http://raspberrypipod.blogspot.com. +++ Current project: PiPodTricorder - lots of sensors, lots of mini-displays, breadboarding, bit of programming.
Posts: 209
Joined: Mon Jun 25, 2012 10:41 am
by recantha » Mon Jul 16, 2012 10:05 am
My last post may have seemed very stupid...
My Raspberry Pi blog with all my latest projects and links to articles
http://raspberrypipod.blogspot.com. +++ Current project: PiPodTricorder - lots of sensors, lots of mini-displays, breadboarding, bit of programming.
Posts: 209
Joined: Mon Jun 25, 2012 10:41 am
by zardoz99 » Mon Jul 16, 2012 10:07 am
Anyone got any thoughts on using a Schottky diode instead to reduce the voltage drop? Any recommendations?

Z.
Posts: 157
Joined: Fri Jan 13, 2012 2:25 pm
by zardoz99 » Mon Jul 16, 2012 10:12 am
recantha wrote:My last post may have seemed very stupid...


Ideally we should have zero power transfer in either direction during normal operation.
However, the 5 volt supply on both the RPi and the hub will always be fractionally different in output voltage, so there will always be a tendency for a small (maybe VERY small) current to flow between the devices in ether direction when running. The red wire cut prevents that possibility.

So, no your post is not stupid..

Z.
Posts: 157
Joined: Fri Jan 13, 2012 2:25 pm
by Zowghi » Thu Jul 19, 2012 5:57 pm
I just did the basic method of snipping the red wire. Worked a charm. Thanks for the mini tutorial!
Posts: 2
Joined: Thu Jul 19, 2012 5:49 pm
by mxcum167 » Thu Jul 19, 2012 8:02 pm
Rather than open up my brand new USB Hub, I cut the red wire in a USB extension lead that I had spare.
Martin www.chez-cummings.com
User avatar
Posts: 41
Joined: Sun Mar 04, 2012 11:56 am
Location: UK
by drgeoff » Thu Jul 19, 2012 9:07 pm
zardoz99 wrote:Anyone got any thoughts on using a Schottky diode instead to reduce the voltage drop? Any recommendations?

Z.

Should work and be slightly preferable to an ordinary diode. It needs to be a Schottky power diode (as opposed to a Schottky signal diode) rated at a continuous forward current of at least 1 or 2 amps. AC performance is not an issue so no worries about capacitance, operating frequency etc.
Posts: 2959
Joined: Wed Jan 25, 2012 6:39 pm
by Roman0 » Fri Jul 20, 2012 5:51 am
Lucky bastards. I have a similar problem with my LogiLink UA0085, but I guess there's no way for me to cut any 'red wires', heh.

Image

EDIT:

Oh, and one more thing. Don't buy it for Raspberry Pi. I have some nasty USB problems with it. Mouse works only half the time, keyboard one time works fine, then it either doesn't work at all or has sticky keys, and Wi-Fi loses connection. So yeah, even if it's on the http://elinux.org/RPi_VerifiedPeripherals as working, doesn't mean it does.
User avatar
Posts: 4
Joined: Tue Feb 14, 2012 5:38 pm
by drgeoff » Fri Jul 20, 2012 8:49 am
Roman0 wrote:Lucky bastards. I have a similar problem with my LogiLink UA0085, but I guess there's no way for me to cut any 'red wires', heh.

Image

EDIT:

Oh, and one more thing. Don't buy it for Raspberry Pi. I have some nasty USB problems with it. Mouse works only half the time, keyboard one time works fine, then it either doesn't work at all or has sticky keys, and Wi-Fi loses connection. So yeah, even if it's on the http://elinux.org/RPi_VerifiedPeripherals as working, doesn't mean it does.

I think the blame for those problems should not be laid at the door of that USB hub. There is mounting evidence that the RPi's USB code has some bugs. There are many posts on this forum and see eg https://github.com/raspberrypi/linux/issues/29.
Posts: 2959
Joined: Wed Jan 25, 2012 6:39 pm
by JRT » Sat Jul 28, 2012 11:51 am
Just wanted to say thanks for this guide! This hub is cheap, powerful enough and importantly - small!
Posts: 28
Joined: Tue Jul 17, 2012 2:44 pm
by Montala » Sat Jul 28, 2012 12:20 pm
Out of interest, what is the rating of the PSU for the Logic LP4HUB10 powered USB hub please?

Also, is the power connection to the hub a 3.5 mm o/d 'barrel' plug, or something else?

Thanks! :)
User avatar
Posts: 638
Joined: Mon Mar 05, 2012 11:54 pm
Location: Herefordshire (U.K.)
by JRT » Sat Jul 28, 2012 2:55 pm
It's plugged in right now (I'm typing this on my pi!) so I can't check, but I'm fairly sure it's 2A. I'm almost 10% on that. Also it's a barrel type, yes - I'm not up to spec on power supplies so I couldn't tell you the exact type, but it looks like the one you said.
Posts: 28
Joined: Tue Jul 17, 2012 2:44 pm
by HUD Engineer » Mon Aug 06, 2012 4:54 pm
I bought a similar L4THUB10 (with one port on the top face) this week (£7.99, Currys), which does have 2.0A @5V PSU. This is listed somewhere as one of the ones that works. Mine doesn't. Before introducing the Hub, I had a Dell branded wireless Logitech keyboard and mouse running on one USB port, and a Belkin FD5750 (Vn4xxx) (ZD1211 chipset) in the other, on Raspbian 2012-07-15. The Hub instantly introduced problems, whether powered or unpowered.
e.g. multiple line error reports "eth0:kevent 4 may have been dropped".
or:
keyboard still works through it, and wlan0 in upper slot of Pi does not: "failed to bring up wlan0"

After cutting the power wire on the incoming USB lead in the hub, it was not willing to work as a hub for any peripheral that I have tried.

It did work with a regular notebook PC before I cut the red wire, but not any more.
Posts: 7
Joined: Wed Jul 18, 2012 10:25 pm
by drgeoff » Mon Aug 06, 2012 5:58 pm
@HUD Engineer. Did you verify that the red wire is the power feed or did you assume that because the power wire on a completely different piece of equipment was red, then yours would be red too?

The 'may have been dropped' errors you had are not the fault of the hub per se. See the post by gsh at viewtopic.php?p=141987#p141987.
Posts: 2959
Joined: Wed Jan 25, 2012 6:39 pm
by Nosavvy » Mon Aug 06, 2012 7:33 pm
#HUD Engineer.
Logik L4THUB10.
Wish you had posted a few minutes earlier.
I read the previous posts and trotted off to CUT THE RED WIRE. Returning to the monitor I now see your post. :lol:
I can CONFIRM that it does not now work. Oh well! off now to do a bit of soldering. :D

NS.

Success, it is now working again. I have also noticed that the PSU is 5.0V 2.0a but the unit itself says 5.0V 500mA so I gather that is the reason it does not work with the Pi.

NS.
Posts: 2
Joined: Thu Jul 12, 2012 6:54 pm
by HUD Engineer » Wed Aug 08, 2012 8:12 am
(Revised submission):
I can confirm that in the case of the Logik L4THUB10, Red still equals power (marked R, J1). (Renaults are good training that you should assume nothing from wire colours, but I accept my first post could have been more concise.) The design has evidently evolved from that of the LP4HUB10:

In the L4THUB10 the USB lead from the root hub powers the Genesys Logic GL850G (U1) directly, drawing just about 69mA in my observations (when plugged into a Raspberry Pi), with no other downstream devices connected, [rising to 72mA with a bus powered USB 1 Hub added downstream of the powered Logik Hub, and rising to 85mA with the Logik Hub unpowered]. The voltage from either USB port of my Pi seems to be a lowly 4.3V (4.75V on GPIO), when the other port is supplying 3.3V to my Wifi. If I unplug the wifi, I can see about 4.5V on the USB (and 4.84V on GPIO), but voltage doesn't seem to be an issue for any devices I've tried so far. My Pi PSU (I've tried several sources) is generally 700mA or better.

For comparison, using a Compaq Mini: 68mA with no other downstream devices connected, [rising to 74mA with a bus powered USB 1 Hub added downstream of the powered Logik Hub, and rising to 90mA with the Logik Hub unpowered]. The Compaq Mini USB Root Hub voltage is 5.1V

On the L4THUB10, all 4 ports are powered from a common 5V rail, which draws power from the higher of the voltage of the external supply, or the root hub voltage minus the Schottky diode (D1) 0.4 V forward voltage drop. Potentially, a Hub power interruption can suck a lot of current from the Pi.

The Genesys Logic GL850G datasheet (apart from expressing Power Dissipation in units of mA) indicates that the current varies as a function of how many ports are active and whether they are Full Speed or High Speed. It should be noted that for almost all permutations and combinations, the current of the GL850G is stated as more than 100mA, and a maximum of 180mA. This appears to be the current to generate the comms, not the current to power any downstream devices.

An extract from the datasheet reads
"5.2.3 SELF/BUS Power Setting
GL850G can operate under bus power and conform to the power consumption limitation completely
(suspend current < 2.5 mA, normal operation current < 100 mA). By setting PSELF, GL850G can be
configured as a bus-power or a self-power hub."

Given how critical to Pi operation the difference between drawing 100mA or 180mA max, it begs a question: Does the GL850G only achieve a normal operating current of below 100mA when configured for Bus Powered, or is that a basic current draw before comms are added in?

The datasheet suggests automatic switching can be performed between bus powered and self powered modes, although how this affects behaviour (or current draw) is not clear to me. However, in the L4THUB10, the PSELF signal seems to be pulled to (an internally generated) 3.3V, suggesting the device itself is set to Self Powered, but the GL850G is evidently Bus Powered from the way it gets power from the root hub. It is only the downstream devices that are drawing from the Hub PSU.

Would there be any harm in rewiring the L4THUB10 so that the GL850G is powered solely from the Hub PSU rather than from the Pi? Would the GL850G need to reset when the Root hub wakes up, in order to conduct whatever protocol is used to establish the best comms? Having read that the Raspberry Pi has polyfuse protection, and that the unpowered L4THUB10 will sink current into any connected loads at the time (presumably no more than 100mA each, until permitted to draw more), the L4THUB10 appears to need using with care near a Raspberry Pi. I'm actually tempted to leave the
Red wire cut, and bypass the diode so that the GL850G can only take power from the Hub PSU, and so that the other ports on the L4THUB10 cannot suck power out of the Pi when the L4THUB10 is unpowered. Or reverse diode D1 so that the GL850G can power from the Hub PSU (if available) and add a diode in the root hub power wire, so that the GL850G can power off the Pi if it needs to, but the Pi cannot sink current from the Hub PSU? Can someone with more wisdom advise on that?

Although the L4THUB10 hub sports a label indicating the Input is 5V, 500mA, I see no reason why each port cannot draw that unless the GL850G is programmed to limit the collective units of power to each intelligent downstream device. I'm not familiar with how one can determine how many units each device is requesting or being granted.

Paying closer attention to what combinations work, or otherwise, I see that I can have the Hub connected and I can run my wireless keyboard and mouse through the hub OK (although barely tested). Or I can ditch the hub and have the wifi, but the wifi will not work through the hub or alongside it (on the other Pi USB port, either way round) , and eth0 gets disrupted also.

So, having covered the power details, I suspect the USB issues pointed out by drgeoff above will prove to be at the heart of this, but if the Pi USB ports are limited to 100mA each then the L4THUB10 is taking it towards the red zone as more downstream devices are plugged into it, particularly so for High Speed devices which can potentially push the L4THUB10 to draw 180mA from the host, even when it is powered.

Andy
Posts: 7
Joined: Wed Jul 18, 2012 10:25 pm
by obcd » Wed Aug 08, 2012 8:37 am
In a normal situation, an USB port is suposed to be able to provide 500mA to an usb port.
A bus powered 4 port usb hub divides this current into 5 * 100mA. The hub electronics shouldn't use more than 100mA, and it still leaves 100mA for every downstream hub port. When a device is connected to an usb port, it will tell the usb host during it's enumeration how much power units of 100mA it needs. If it needs more than 1*100mA, and it's connected to a bus powered hub, the pc will tell you that you need a self powered hub to use the device. Since at that moment, all downstream ports can again deliver 500mA each, such a hub supply should be able to provide 2 Amps.

It's obvious that many devices aren't designed properly. A 50 euro 4 port usb hub will probably have a better supply than a 10 euro usb hub.

If the usb hub is also providing power to it's upstream usb port, than it's a very bad design.

Another rule of design is that only low power usb clients have the usb connector connected to the device directly. high power usb clients ( up to 500mA) should have an usb type b connector so that they can be connected with an usb A - usb B cable.

Usb memory sticks and usb harddisks are very power hungry devices. Specially at startup, harddisks can exceed the 500mA specification. That's why some have 2 usb connectors. One for power and signals, and one for power only. (Y cable)
Posts: 890
Joined: Sun Jul 29, 2012 9:06 pm