description for new board added

This commit is contained in:
King Kévin 2014-05-02 21:19:55 +02:00
parent 4779b30a0b
commit 3997ecd531
1 changed files with 50 additions and 42 deletions

View File

@ -5,14 +5,11 @@ the motoserial is a phone controller
it is used to power, flash, and communicate with Motorola C1XX phones
the purpose is to use it for osmocomBB
the USB interface will offer 4 UART ports, do communicate with 4 phones
provide external power (4.5-5.0 V, ~2 A) so to be able to power the phone, but also be able to flash them
purpose
-------
I used the osmocom baseaband software on Motorola C118 phones to monitor GSM networks
to communicate with them, I used USB to UART converters (CP2102) on the 2.5 mm TRS jack
to communicate with them, I used USB to UART converters (PL2302, CP2102, …) on the 2.5 mm TRS jack
to flash the phone with the osmocomBB firmware, remove and reinstall the battery, start osmocon (or leave it running), and press on the ON button to start the bootloader
but sometimes the software gets stuck, and for monitoring GSM networks, an automatic way to reflash the phones had to be found
@ -22,10 +19,11 @@ technology
instead of the ON button, a pin on the back of the phone can be used
but I noticed that plugin the charger had the same effect
and instead of the battery, a 5V power source can also be used
so the trick is to use 2 transistors: 1 to simulate battery re-plug, and 1 to simulate charger replug; by switch off and on both power sources
so the trick is to use 2 transistors: 1 to simulate battery re-plug, and 1 to simulate charger replug; and switch off and on both power sources
since they share the ground pin, p-MOSFET on the positive side had to be used
actually the order and switching timing is not too important
thus one transistor, switching both the battery and charger power sources, is enough (and a n-MOSFET can be used)
thus one transistor, switching both the battery and charger power sources, is enough
this way a more common n-channel MOSFET can be used, but the groung pin of serial should then also ne switched
the battery can be replaced by a regular 5 V power source, the same as the one used for charging
if the voltage is above 4.2 V, the battery should not be charged
@ -45,53 +43,63 @@ hardware
files
-----
motoserial-phone:
the battery replacement
provide 5 V (~1 A) input power
the other port is to forward this power to the charger
the PCB will replace and provide power instead of the battery
a diode is used to prevent revers current when charging
capacitors are necessary to cope with the peak power usage when the phone transmits
motoserial-spare:
this is the USB to 4 UART + power controller board
it uses spare parts from my component stock (mainly the expensive FT4232H)
motoserial:
after several prototypes (see versions), this was the best solution
the PCB replaces the battery
I had too many missing or dead batteries (which will start charging only after 30 minutes, and only lasts 10 minutes)
large capacitors compensate for the large but brief power surges during transmition bursts
connect the PCB to the computer through a USB to UART converter with the following pins: 5V, TX, RX, DTR, GND
a good USB to UART converter would be the Silicon Labs CP2102 (or CP2104)
they handle baudrates slightly different from standard values (which the Mototola C1XX have) better than Prolific PL2303, but not as good than FTDI FT2232R
they are more expensive than Prolific PL2303 converters, but cheaper than FTDI FT2232R, and work quite fine
good and cheap converters are BAITE (betemcu.cn) BTE13-007, CNT-003, or B75937 (solder pin DTR)
the DTR (or RTS) signal is used control the power switch
CP2102 resets this signal when the serial connection is closed
FT2232R can only reset RTS once the option is set in EEPROM, but DTR goes and stays low the first time the serial port is opened
CP2102 also includes a 700 mA resetable switch on the 5V pin which will prevent draining too much power on the USB port
connect the UART port to a 2.5mm TRS stereo jack and plug it in the headphone to be able to flash and communicate with the phone
connect the CHG port to a jack barrel (1.1mm inner diameter and 3.0 mm outer diameter) and plug it in the charger port to able to start the phone
some phones also provides these pind on the back, but not all
only though hole components have been used because it could not be all SMD (large capacitors are very expensive, and header connections to weak)
this allows to have a single sided PCB which can also be produced in amateur labs
version
-------
0:
working version
current working version
E:
now motoserial-phone version prototype built (also provides a UART port)
bugs:
- resistors are in the way of connections
- pull-up/limiting resistor misplaced. the nMOS gate oscillates around Vgs
D:
4 motoserial-phone prototypes built to work with the motoserial-cape
bugs:
- the central tab is a bit too long. it's hard to remove the battery
C:
second motoserial-cape prototype built
bugs:
- one ground copper connection in the plane cut because too thin
- 100nF dim the UART signals and should be removed
B:
first motoserial-cape (for BeagleBoard BeagleBone Black) prototype built
bugs:
- I used nMOS instead of pMOS, which won't work because the power connection on the phone shares ground with UART
A:
first scrape prototype built using CircuitPro + ProtoMat S63
first motoserial-scrape prototype built using CircuitPro + ProtoMat S63
bugs:
- the hole for the USB and header pins are too wide. I hope it would create some through hole plating thanks to capillarity, but it didn't
- the hole for the USB and header pins are too wide. I hoped it would create some through hole plating thanks to capillarity, but it didn't
- wider vias annulus would be a good idea
- DTR it set (to low) when the serial port is opened, but is no reset (to high) when closed (FT4232H specific)
- RTS can be reset to high on close, but that need to be programmed in the FT4232H's EEPROM
- the 3.3 V on DTR can not drive the pMOS because it's switching 5.0 V, but V_GS < 1.1 V
- decoupling capacitors on the serial lines should improve the quality
- try if with less feritte beads and decoupling capacaitors PLL still works
B:
first cape prototype built using CircuitPro + ProtoMat S63
bugs:
- I used nMOS instead of pMOS, which won't work because the power connection on the phone shares ground with UART
C:
second cape prototype built, using CircuitPro + ProtoMat S63
bugs:
- one ground copper connection in the plane cut because too thin
- 100nF dim the UART signals and should be removed
D:
4 phone prototypes built to work with the cape
bugs:
- the central tab is a bit too long. it's hard to remove the battery
E:
now phone version prototype built
bugs:
- resistors are in the way of connections
- pull-up/limiting resistor misplaced. the nMOS gate oscillates around Vth
- Silabs CP2108 would be cheaper and easier to use than FT4232H