[linux-dvb] NOVA-S SE (1016) help

Michael Uemminghaus uemmingh at cablenet.de
Sat Mar 12 09:27:29 CET 2005


I have been playing with my NOVA-S SE low profile card a lot, but now i am 
stuck. I hope at least to give somebody some additional input.

My findings so far are:
- The budget (not budget-ci) driver matches the saa7146 part quite well if the 
appropriate PCI ID is added.
- After opening the top of the tuner box, the PCB matched the picture earlier 
in a thread concerning PCI ID 1016 and the datecode is only 1 day later.
- Looking up the two possible demodulation chips mentioned in that thread, i 
now am very sure that the two chips are identical or at least register 
compatible (the P1010 was developed (for|with) samsung).
- The sh1420 datasheet tells me that all reset values are OK so there is no 
setup table needed.
- The driver detects the i2c serial eeprom at i2c device 0x50 and can read the 
MAC address.
- I could change the PCI ID of the card by writing to i2c device 0x50 and 
address 0 (and back)
- I determined that the saa7146 GPIO 2 is equipped with a pull down (600 ohm 
measured) and is routed into the frontend box. The I2C traces go parallel to 
the serial eeprom and the tuner box.
- I have a good chance to get 18 Volt output at the F connector by simply 
setting the GPIO 2 to OUTHI.
- The chance is about 100% if I try to read from all possible i2c devices 
afterwards. This is weird also.
- According to the sh1420 data sheets the device can have i2c id 0xa0,0xa2, 
0xa4 or 0xa6 which translates to linux i2c addresses 0x50 to 0x53 if i did 
not get it all wrong.
- I am 80% sure that the GPIO is the inverted reset input of the sh1420
- BUT the sh1420 should output 13V according to the default reset values.
- running the saa7146 dump on my brothers XP machine could not list the saa 
registers, possibly because i specified 0x50 as i2c device the first time 
around and did not log the output. (serial eeprom might be overwritten??) I 
have not yet tried to plug it back into my machine, it even might be unusable 
now. On the XP machine the output is a bit above 19 Volt.

Weird things:
- The serial EEPROM and the sh1420 seem to be at the same i2c device address 
by default!  (both 0xa0 and following)
- when scanning all possible 256 device addresses of the i2c bus i see that 
the upper 128 device addresses mirror the result from the lower addresses.
This and the fact that the serial eeprom datasheet say the device is at 0xa0 
and is detected at 0x50 tells me that the device address for linux is shifted 
one bit down.
- scanning address 0 gives a rc code of -5 (should be a device broadcast 
address?)
- scanning (linux) i2c address 8 gives very strange results:
when reading back single bytes, the return is always the address with the two 
lowest bits set to 0: (device 8, starting at address 0)
00 00 00 00 04 04 04 04 08 08 08 08
When reading back more bytes at a time:  (device 8, starting at address 0)
00 00 00 00 00 00 00 00 00 00 00 00 
or whatever start address is given without the lowest two bits. This seems a 
bit strange for a serial bus.
The rc of the i2c transfer is always OK (2)
- Apart from that there is NO indication of the presence of the sh1420, all 
other addresses result in rc -121


Did i miss something here?
Has somebody seen something like that before?

 Michael




More information about the linux-dvb mailing list