*khem, khem* So now, I come with real mystery. It may be not directly related to HEN as-is, but charging while hostmode was "introduced" here by joerg, so problems with hostmode is even less good thread for that (cause hen is working here like charm).

Apart from powering with Y-cable, i tried using external-powered USB 2.0 hub, to not only deliver power to USB peripherials, but also charge N900 while hostmode.

It worked like a charm, but one thing is beyond my understanding. When i deliver power to hub using stock N900 charger (through a homebrew adapter, especially for having less chargers to carry when travel), N900 is hapilly charging through hub, at 300 mA rate + power need to operation with peripherials, its about 1500 mA at all (so current_now is -300 mAh, cause battery "get" 300 mA more than is pulled from it)

but, when i power my hub through modified ATX power supply - much more powerful than N900 charger - N900 is barely charging at all, with current_now fluctuating from -50 to +50 (so, part of the time it isn't charging at all, power pushed into battery is little less than used by N900).

I already confirmed that N900 itself is not using more power at this moment - exact same conditions as with powering hub by N900 charger. For some reasons, power delivered to charging N900 is less.

I tested (using multimeter) both N900 charger and ATX power supply current delivered through hub - that is, tested current at USB cable connecting to N900 (of course, tests were made UNCONNECTED to N900). N900 charger was delivering 1,5A, as expected. ATX power supply was delivering 2.0 A - HUB do its job by limiting current going through it to exactly 2A (4 ports X500 mA).

I tested it 5 times for each combination, restarting HEN, connecting, disconnecting, etc. The result is, that when N900 have 1.5A available power at USB cable, thanks to charger.sh it uses it all. When it have 2A available, it uses only about 1A.

Someone able to explain that phenomena?

// edit

Main reason why i posted it here, is because i hope that this may help in development of new HEN version with charging during hostmode - behavior, when N900 charge less when more power available is strange, to say at least. And probably related to some stock components (mbe? just guessing) so maybe jrbme will use that additional info...

You considered that battery might simply be completely charged? check battery voltage, if it's at 4150mV or above, then there's little reason for it taking more charge, no matter how much the power supply can deliver

cheers
jOERG

hello guys i don't know if this problem have been mentioned in the thread i can't read 80 pages anyway
my tranfer speed between n900 and the pc with a usb cable is about 13mb/s but after i use h-e-n (well actually turnig on the vbus and using any usb flash) and after that (shut down the h-e-n) then transfer a file to pc with a usb cable the transfer speed goes down to 4 mb/s
is that a bug in h-e-n?
a bug in kernel-power-v47?

so i need to reboot n900 to get the full transfer speed
please somebody confirm
Thanks

Thank for giving ideas. Yea, that was also what i thought, but i tested it by switching sources (restarting hen etc) 5 times in row, on the same conditions. Always more powerful current possible to draw from source (mentioned 2A) resulted in N900 drawing less. That was with battery @ about 85 %

Also i tested it again with 3,815 voltage (check before tests), with exact same results. With 1,5 A available, N900 used everything, when 2A available, barely charging @ the same rate as power pulled from N900. I got no idea

Of course i also did it with the same cables, checked cable integrity etc... My first thought was that my cable going from ATX power supply was broken, thats why tested if it can give 2A @ end of USB cable (going through HUB etc), that plug connected to N900. Very bizzare for me.

// Edit

Also, i don't want to put pressure but if You could kindly look @ that posted in last page:

I had something similar happen with an external pack I built. In the end I attributed it to losses in my usb cable/port...

However, if you inject the power into your contraption, at the same point, and you're able to measure the voltage at that point, and see the same voltage in both places... Then it's starting to get strange

It turned out a proper replacement of bme needs kernel tweaks (driver providing /sys nodes for bq27200 and bq24150), a rewrite of hal-addon-bme (to report battery state to hal and make things like normal battery applet work), among others, and those patches inevitably would cause bme to not work at all or at very least need severe messing around with kernel etc - in the end you have either a system with bme or a system working with bme-replacement but breaking several APIs for the way maemo works now, but you hardly can build a system that can do both.

([edit:] there's a kernel driver providing bq27200 sysfs API, and loading this driver causes bme to break, because the way bme is accessing that chip (direct I2C) is incompatible to what the driver does to kernel (exclusive allocation of the I2C resource, so bme can't do that).
Also I changed my mind recently towards a proper implementation of a bq24150 based charging controller shouldn't be done in userspace at all, like bme is done (Nokia probably chosen this approach only because for a kernel borne "bme" they couldn't keep it closed due to GPL). Rather it should become a comprehensive kernel module that exposes a sane mix of low level sysfs nodes to access chip registers/properties directly, and high level sysfs nodes to control what the kernel based charger does - there's a still half done spec for that I stopped to finish as nobody been interested in implementing it, obviously for the reasons mentioned above: it's no longer maemo but an incompatible kernel and APIs that needs rewrite of a lot of mostly closed source things like battery applet, and/or hal, dialer (yes!), et al [/edit])

My present ideas are more along a augmented h-e-n that implements charging hostmode in a kernel independent way. For jrbme probably shadowjk's charging script is almost as good as it gets.

[edit2] http://mg.pov.lt/maemo-irclog/%23mae...09-30T14:44:46 ff has some of the more recent stuff going on or planned regarding all that[/edit2]
sorry
jOERG

Having been busy with many other things, I have just finally tried charging with hostmode, using a very cheap powered hub and, following the instructions to the letter it works exactly as described. Thanks!

My rather naive question is now whether there is any way to insert and remove devices from the hub and have the N900 recognise them. I've tried a few combinations but the only one that seems to work is to start everything from square 1, ie disconnecting the hub, shutting down h-e-n, inserting the device, restarting h-e-n, etc. etc. If this could work, then the hub would work as an N900 docking station, which would be very nice.

@ joerg_rw:
Thanks for explanation, and there is no need to sorry You did awesome job... And, thank You for detailed information about that. First, i thought about cooperation with pali's kernel-power, but even my very low understanding of details provided by You make me feel that achieving proper working kernel-incorporated solution is close to impossible now.

Anyway, just clear information is what people here "wanted" - so again, Thank You for that. For me, shadowjk's script is quite good, and if it is most of what we can get now, what is stopping HEN team from releasing new version of HEN, incorporating Your conclusions? With upgraded version of booston/boostoff and GUI integration of "new features" like charging while hostmode.

Again, I'm not "demanding" anything - just would like to get actual info on "Work in Progress" status. Thanks in advance.

For me, most of the time when i insert new device into USB HUB, I just need to click "enumerate" again - without disabling vbus etc. - to get it "discovered" by N900. Sometimes it fail, for unknown reason - then, indeed, i just start from scratch.

Just got my adapter cable! Yay! Plugged in a USB key and selected "high speed", "VBUS", "enumerate", "mount" but it's not mounted. I'll play with it more later because my battery is dieing now. There are some error-like things in dmesg. Are there problems with some USB keys (this one is actually an SD card reader)?

So, I tried a USB mouse and another USB key I found around and both worked! yay! Any ideas on why my SD card reader would not work with the n900?