First, a bit of background. I have a PVR-150 card. I got the LIRC stuff going with the IR pickup that came with the card. But, it only works with the Hauppauge supplied remote. I know in theory I could probably get it to work with other RC-5 remotes, but none of the codes I tried on my universal remote seemed to work.
I tried making my own, but the voltages weren't right and I just didn't have the time to iterate on that thing.
So I bought an IRA-3 from www.home-electro.com thinking my days of screwing around were over. Sadly, I was wrong. The IRA-3 uses the IRMAN protocol, BTW.
I hooked the thing up and tried to get it to work with the version of LIRC already installed on my machine (0.7.1). No dice. I couldn't seem to get anything. It kept giving me an error in irman_init(). Crap. So I got the source to LIRC and tried building it. No dice, same error. Because I could initialize the device on my Windows machine, I didn't understand what the problem was. I got the latest versions of LIRC and libirman and tried to sort things out.
After a considerable amount of digging and learning far more about IRMAN, IR, LIRC and serial ports than I ever wanted to learn, I eventually managed to narrow down the problem. While the IRMAN protocol specifies 9600 baud, 8 data bits, no partity, no flow control and 1 stop bit, the port initialization fails to actually specify one stop bit (although it doesn explicitly handle all the other parameters). I added one line to the initialization routine in libirman (to set the stop bit to 1, something it should have already been doing even according to the comments in the code) and low and behold...it worked!!! My IRA-3 initialized and I started getting data from it in the expected 6 byte chunks. Hooray!!!
At this point, I thought I had finally conquered this stupid issue. I rebuilt libirman and LIRC using the new initialization code and I was well on my after running irrecord and building up a .conf file for my remote.
Then...disaster struck. After trying to use Myth with the new remote it seemed like it was ignoring some of the key presses. I went back into raw mode (using test_io in libirman to output the codes) and it turns out that the stupid thing is spitting out multiple codes for the same key. I don't mean it repeats itself (which it does, but that is a different issue). I mean it gives a different code sequence! God Damn IT!!!!
The problem seems to be related to the history of key presses. As I said, a single keypress send a duplicated code but that doesn't seem to cause any problem. The real problem is when the codes change. For example, running 'test_io', these sequences are for the same key:
[29][14][7d][f7][5f][ff]
[29][14][7d][f7][5f][ff]
[2a][e8][4e][f7][5f][3f]
[2a][e8][4e][f7][5f][3f]
[29][14][7d][f7][5f][ff]
[29][14][7d][f7][5f][ff]
WTF? I find that if I hit the same key repeatedly I pretty much get consistent codes. But, if I jump around (e.g. LEFT, RIGHT, Channel-UP, LEFT) that the same key (e.g. LEFT) will have different codes.
My hope is that somebody has some simple remedy (e.g. something simple like turning on flow or changing the parity or something). By the way, the Hauppauge remote works fine. But it only send 2 bytes?!?
Any suggestions?!? I was so close. After having hacked may way through all this to have it fail to read remote key signals reliably is so aggravating!!!!!
Any help would be greatly, greatly appreciated.
--
Mike
_______________________________________________ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
