Bicephale
join:2005-09-24
 ·TekSavvy Solutions..
1 edit | reply to markuswarren
Re: GNet BB0060B ADSL Modem/Router issue
Hi MarkusWarren,
I'm sorry, «Loader Checksum ... Passed» sounded so
promissing but i should have recalled the details:
*WARNING* TFCfg.bin + firmware upgrades, Jay, 2003-Oct-13
After this message, your BootLoader is supposed to
self-extract then 'UMon' would get activated, then
the FLASH chip would be verified and, finally, the
display would say something like "TFTP/UMON Server
Started ..." once the checksum test fails but that
never happens. I was about to recommend 'TFTPD32'
to see what its "Log viewer" tab looks like but it
seems you get no TFTP support so lets forget it!!!
Anyway, in case you need it later, i've found on a
Portuguese page that the MoDem may also be reached
at 10.1.1.1 (in addition to 192.168.1.1)... Maybe
other GNet owners in a better position can try it.
Since you need to build the maintenance cable, i'm
going to finish commenting about it... The cached
text contains information i completely forgot that
could be useful in clearing some of the confusion.
First, i'll just quote the passages to begin with:
2005-Sep-18 8h23, Bicéphale
So far, pin #1=SI, pin #2=SO and pin #5=+3V3 if
pin #6 is Ground.
I get about 70 mV A.C. on pin #3 while there's no
A.C. signal on pin #4 and the later pin drops all
the way down (to below 50 mV) during boot time...
Usually, pins #3 and #4 both remain close to +3V3.
2005-Sep-20 16h49, Bicéphale
`LoadFi' launched from a command-line
describes the v1 cable ("-c1" flag by default) as:
D5 (Data-5) -> CS (Chip Select)
D4 (Data-4) -> Clk (Master Serial Clock)
D3 (Data-3) -> SI (Serial Input)
D2 (Data-2) -> Rst (Reset)
~S7 (?) <- 100Ω <- SO (Serial Output)
Gnd25 (Gnd) -> Gnd (Ground)
Then i got this while reading the Atmel #2224 .PDF
documentation (see "Pin Descriptions", page 8/33):
1) Write Protect can be tied high if unused, 2) an
internal power-on reset circuit is present and the
Reset line can be tied high if unused... Page 24:
i also read that SPI mode is 3 by default and i've
also found that the Ready/Busy* output pin must be
terminated with an external pull-up resistor since
that's an open-drain stage and it's low when busy.
Oh, and somewhere it says that "Serial Mode" is 4.
I can't garantee this will help in the near future
but knowing there are half-output stages with some
load resistors finally put me at ease, somewhat...
2006-Feb-16 19h42, Franc Zabkar
For anyone wanting to try the "parallel port"
technique for recovering from a failed flash, here
are the pinouts of the header (JP2) adjacent to
the Atmel EEPROM in a D-Link DSL-302G modem/
router.
JP2-1 -> Atmel EEPROM pin 13 & GS8120 (CPU) pin 91
via 680 ohm resistor
JP2-2 -> Atmel EEPROM pin 14 & GS8120 (CPU) pin 90
via 680 ohm resistor
JP2-3 -> GS8120 pin 53 & VIA VT6103 pin 48
JP2-4 -> Atmel EEPROM pin 11
JP2-5 -> Atmel EEPROM pin 12
JP2-6 -> ground
I'm guessing that JP2-3 is some kind of reset pin
for the CPU and VT6103. It probably prevents the
CPU from talking on the SPI bus. This would ensure
that the external device (eg PC parallel port)
would have total control of the EEPROM.
According to this, it appears LPT-4 (Data 2) might
be connected as planned to "Reset", after all: it
seems like JP1-3 isn't fixed at +3.3 Volts but can
vary over time (it would be coherent with the fact
that the master Reset line might also happen to be
active low just like the U8-2 slave Reset line)...
I can't tell with much certainty if Zabkar's cable
for a D-Link DSL-302G and that of the Chineese are
the same mechanically speaking but his comment has
focussed my attention on something of interest: i
begin to think that JP-3 effectively operates as a
master Reset signal through U10 via pin U10-53 and
a discrete transistor labelled U9, using U10-40 as
the line which also happens to control U8-12 (SCk)
via a 68 Ω resistor... The U8-2 (Rst) pin then is
a slave Reset line, i guess. So, if JP1-3 is OKay
then that leaves JP1-2 and i bet it can be cleared
too, lets see... JP1-2 connects LPT-11 (*S7/Busy)
to U8-14 (Serial Output) through a resistor so i'm
tempted to conclude that 'LoadFi' is reading FLASH
data via the *S7/Busy printer signal! This is not
what i'd expect in a perfect world but why not!...
I may still be mistaking but at this stage i'd say
this maintenance connector is going to work with a
GNet BB0060B device as was described previously...
About the 6 pins header, i found one snapshot that
makes me think you probably have enough room above
the FLASH chip (U8) to line up two 2 x 3 IDC blocs
as those used in old Hard-Disk controller cabling,
or pehaps one single 2 x 6 or 2 x 8 IDC bloc which
was used for old Game/MiDi off-board connectors...
In short, i'll say you're booked for a flea market
trip: you might have to wait for the week-end!...
I bet the RS-232 custom adaptor won't be essential
here: it's probably for monitoring purposes only.
In any case, good luck while playing guinea pig!!!
N.B.:
When working on the parallel connector be careful:
i've seen mentions about the possibility of errors
relatively to the Chineese picture being reversed!
Addendum
When you're done repairing your spare GNet BB0060B
MoDem/Router i'd like you to try this tweak below:
DMT - US/DS frequency bands/Tx Bin Adjust: [ ]
TelNet - modify dsl config txbinadj enable
DMT - Upstream (TX) Bins: 6 - 13
TelNet - modify dsl config txstart 6
TelNet - modify dsl config txend 13
--------------------------------------------------
DMT - US/DS frequency bands/Rx Bin Adjust: [ ]
TelNet - modify dsl config rxbinadj enable
DMT - Downstream (RX) Bins: 14 - 248
TelNet - modify dsl config rxstart 14
TelNet - modify dsl config rxend 248
--------------------------------------------------
DMT - ec fdm mode: FDM
TelNet - modify dsl config ecfdmmode fdm
--------------------------------------------------
DMT - Tx Power Atten.: -1 dB
TelNet - modify dsl config txatten 1
--------------------------------------------------
DMT - Coding Gain: 7 dB
TelNet - modify dsl config gain 7
If you prefer to use my tweaks based on CLI TelNet
commands instead of those for 'DMT' make sure that
you didn't forget to type "commit" and "reboot" at
the end. This tweak seems to reshape the UpStream
(1-70) and DownStream (10-255) bandwidths, it once
raised my Local SNR Margin by about 2.5 dB while i
got a Line attenuation level which was improved by
3 dB... One likely explanation is that an Upsream
DSL signal generates noise in any DSL device owned
by the customer and this is injected back into the
DownStream signal locally (reduced bands translate
as less power and hence less interfering noise)...
 |
 .JPG/thumb.jpg "461300 bytes")
%20.JPG){kind=link}
%20.JPG){kind=link}
%20.JPG){kind=link}