On Wed, Jan 25, 2006 at 11:24:16PM +0100, Damien Bally wrote:
I'm quite happy with the hauppauge nova-t pci (conexant model).
More exactly, it is the DVB-T 90002 REV C176. I just bought a second one. The circuit board design has not changed at all since the new Nova-T was introduced, except that some contacts on the newer unit are not solder-plated (they're bare copper).
Despite the broken aerial pass-through, this is good piece of hardware that works well with 2.6.13+ kernels and vdr.
I don't know about the pass-through yet (what is broken in it?), but I can confirm that the card works with 2.6.13+ kernels out of the box.
The IR receiver circuit could be better. I've developed a wakeup-by-remote circuit for it (http://www.iki.fi/~msmakela/electronics/worc5/). The circuit seemed to drop the range of the remote control. I recently measured a little with a memory oscilloscope and ended up wondering why the IR module outputs a '1' signal of about 4 volts. It turns out that the cx23882 (whose pin GPIO[16] is connected to the IR signal) has 3.3 volt I/O. Apparently, the bundled IR receiver module does not work with 3.3 volt supply. So, the Hauppauge guys chose to drop the 5 volts from the PCI bus by R47 (49.9 ohms) to something like 4.1 volts, so that the cheap IR receiver will still work and not kill the I/O pin on the cx23882.
Next step for me is to get a 3.3 volt capable IR receiver and connect my hardware to the 3.3Vaux and PME# lines on the PCI bus. The 3.3Vaux is directly connected to the LED on the board, but PME# (A19) is unconnected. Then I should be able to get rid of the Wake-on-LAN cable, i.e., the modification can be installed as a daughter board on the PCI card.
Marko