N900 will not allow USB OTG! - Page 93 - maemo.org

MohammadAG | # 921 | 2010-05-20, 17:01 | Report

Originally Posted by qole View Post

My experience with USB WiFi sticks and the N800 is that they need an external power supply to run.

That's most probably the case then, I'll get an AC adapter for my hub soon.

Slightly off topic, and a bit dangerous (cause of NOLO), but does anyone have 2 N900s to try flashing the other using the on board flasher?
(or if flasher-3.5 works for the N8x0 then someone with an N8x0 could try connecting the device to the N900 and running flasher (flasher, not flasher-3.5))

efekt | # 922 | 2010-05-20, 17:08 | Report

Odd and a bit off-topic, but why when I tried scrolling back a few pages in this thread I ended up being in http://forum.internettablettalk.com, and not http://talk.maemo.org? odd stuff... Did they change the domain? because if I replace the internettablettalk address with the talk.maemo address, I still get the exact same thread.

MohammadAG | # 923 | 2010-05-20, 17:12 | Report

Originally Posted by efekt View Post

Odd and a bit off-topic, but why when I tried scrolling back a few pages in this thread I ended up being in http://forum.internettablettalk.com, and not http://talk.maemo.org? odd stuff... Did they change the domain? because if I replace the internettablettalk address with the talk.maemo address, I still get the exact same thread.

tmo was itt before N900 times afaik (they were merged - so it's completely normal)
Back on topic...

lifenexus | # 924 | 2010-05-20, 17:14 | Report

Originally Posted by efekt View Post

Odd and a bit off-topic, but why when I tried scrolling back a few pages in this thread I ended up being in http://forum.internettablettalk.com, and not http://talk.maemo.org? odd stuff... Did they change the domain? because if I replace the internettablettalk address with the talk.maemo address, I still get the exact same thread.

the original forum was forum.internettablettalk.com now maemo.org uses it

joerg_rw | # 925 | 2010-05-20, 18:40 | Report

Originally Posted by blue_led View Post

for what reason bme read an idle/unused asic ( bq24xxx ) ?
as i understand from specs in charge or boost mode for sustaining those functions, some activity on i2c bus is needed.
if NO CHARGE mode why i2c read / writes activity ?. this useless activity drain the precious battery.
i will look closer on specs for a decent reason.

if vbus sensing for mode selection pop-up is needed instead of pulling status of BQ24k vbus sensing from 4030 via interrupt is preferred.

popup usb mode selection is not in bme or mce . what code is doing this selection ?

i wish a new one with:

pc suite mode ( never used )
mass storage
host ( docking station )
printer for photos ( pict bridge )
.....

VBUS sensing is to sense attachment of N900 to a host, while 1707 PHY pullup of D(+/-, depending FS, LS) is for sensing attachment of a device to N900.
Quite obviously attachment of a device to N900 can not be detected unless N900 is in hostmode.
So any popup requester offering a hostmode option ("host ( docking station )") is rather nonsense.

The whole hassle about VBUS-supply from N900 is just because we are not in a clean plain hostmode, and tricking the OTG statemachine from B-state into A-state includes a rather complex negotiation scheme about which of both OTG devices supposed to form the session is responsible now for providing VBUS power (if any of both) - so the OTG statemachine has quite a number of useless and partially nonfunctional tests like OTG_STATE-A_VBUS_RISE etc pp (exact names may differ), and we need to do a lot of messing around with VBUS supply to make these useless tests succeed.
For a plain vanilla correctly implemented hostmode, all those tests are moot and should be disabled.

and again what I believe will become my mantra and signature for the next few dozen posts:
We need to put "echo host >mode" to work, we need to switch PHY chip to host config which is 2* pulldown R.

The Following User Says Thank You to joerg_rw For This Useful Post:
S0urcerr0r

titan | # 926 | 2010-05-20, 19:03 | Report

I can now successfully mount my USB stick and here is how I did it:
Warning: only for experts!! It may damage your USB device.
This not the real hostmode. It is an unstable hack to get some debug information.
1. install the kernel from (removed. insiders know where to find it)
(contains MohammadAG patch and hcm's i2c debug patch)
2. boot and connect N900 via USB cable to PC, select PC suite mode
3. execute the script
http://maemory.com/N900/kernel/usb/usb
4. connect your stick via USB F/F adapter to N900, wait a moment

The resulting syslog excerpt is here
http://maemory.com/N900/kernel/usb/syslog.gz
for this I enabled USB debugging 5 before connecting to PC,
so that we know what the driver does when switching to peripheral mode.

Last edited by titan; 2010-05-20 at 19:50.

The Following 6 Users Say Thank You to titan For This Useful Post:
clasificado, egoshin, Marshall Banana, MohammadAG, nightfire, S0urcerr0r

hcm | # 927 | 2010-05-20, 19:16 | Report

now, I have logged the I2C communication to BQ24150 for a whole charge-discharge-cycle.
I deleted lots of unnecessary lines from the logfile, but kept some stuff which is not directly related to i2c.
All i2c entries are just to device 0x6b (ensured by if-statement in the kernel sourcefile, logging all i2c-messages would be way too much to analyze)

I found out the following times which might be interesting:
- 00:29:07 started charging
- 02:16:23 battery is full (after that: sometimes charging, sometimes not => patented battery management

)
- 09:07:11 unplugged charger
- 19:28:19 shutdown because of empty battery

the whole day, nothing was plugged into the usb port (I only played lots of videos to discharge the battery quickly

)

attached you find an image from batterygraph showing the charge/discharge. the log is too large, therefor its here (2,5MB)

Attached Images

The Following User Says Thank You to hcm For This Useful Post:
joerg_rw

javispedro | # 928 | 2010-05-20, 19:20 | Report

Originally Posted by joerg_rw View Post

We need to put "echo host >mode" to work, we need to switch PHY chip to host config which is 2* pulldown R.

Fully agreed. What is needed:

Set up a special global variable that is asserted after otg_set_host
When this global variable is asserted, disable all power saving stuff on the PHY
Also make otg_set_host configure 1707 appropiately via Nokia's MUSB ULPI helper functions. Also ensure you power it before configuring.
Ensure the 1707 remains configured (to check if item 2 was done properly)

What is missing:

List the places where the 1707 PHY is being reset, powered down, reconfigured..
Understand why the way the 1707 is currently configured doesn't match what we'd expect for normal peripheral operation.
And of course, what would be the correct 1707 configuration for host mode (might not be what the datasheet says, as per the above).

Note that I'm not sure what I'm talking about when I say "powered down." It might be low power ulpi protocol, it might be really powered down (but 1707 is connected to vbat?), ...

Last edited by javispedro; 2010-05-20 at 19:23.

The Following 3 Users Say Thank You to javispedro For This Useful Post:
egoshin, joerg_rw, S0urcerr0r

blue_led | # 929 | 2010-05-20, 19:54 | Report

Originally Posted by joerg_rw View Post

VBUS sensing is to sense attachment of N900 to a host, while 1707 PHY pullup of D(+/-, depending FS, LS) is for sensing attachment of a device to N900.
Quite obviously attachment of a device to N900 can not be detected unless N900 is in hostmode.
So any popup requester offering a hostmode option ("host ( docking station )") is rather nonsense.[/COLOR]

joerg you are partially wrong. from your point of view is nonsense because you want only host vanilla , cherry or any host flavor you want
BUT
i want a docking station with 7 usb A receptacle capable to power N900 while is in host mode . I want charge battery because the charging hardware is independent from usb controller

i can write a module doing what i want as appendices to a n900 in plain host mode with all vbus code removed
i don't give up on this
as a proof of concept i post a picture with a ACA with mini A receptacles for data and charger . this device can power and charge a device while this device is in plain host . in fact i don't know if this board can handle otg voltages & states and i don't care

2 month ago i tried on n900 this adapter but F..bme was in place

http://talk.maemo.org/showpost.php?p...5&postcount=17

The Following User Says Thank You to blue_led For This Useful Post:
S0urcerr0r

egoshin | # 930 | 2010-05-20, 20:17 | Report

Originally Posted by joerg_rw View Post

We need to put "echo host >mode" to work, we need to switch PHY chip to host config which is 2* pulldown R.

I double checked that - both pulldowns are ON.

ID PULLUP may be missed sometime but DM and DP are pulldowned.

The Following 2 Users Say Thank You to egoshin For This Useful Post:
javispedro, S0urcerr0r