View previous topic :: View next topic |
Author |
Message |
matzh9
Joined: 21 Aug 2004 Posts: 12
|
Posted: Tue Aug 24, 2004 5:22 pm Post subject: |
|
|
Hi again Jon!
You are very close. If I fire the IRs of from C-Max as fast as possible, approx 1 per second, all codes eventually gets recognised. Once recognised it is always right, meaning 10=0 and 19=9. I guess this means that the codes indicated in the dumpfile is right and your LIR is 99%
As I was struggling with learning codes I managed to get one that worked 1 of 100 times when I used 54kHz. The button was power. I tried your codes with 55 as first byte but that did not help.
I feel ashame to ask you if you could have another look.One or two working codes would help me a lot. Now I'm conviced there is a solution.
Thanks,
Matz |
|
Back to top |
|
|
jon_armstrong Expert
Joined: 03 Aug 2003 Posts: 1238 Location: R.I.P. 3/25/2005 |
Posted: Wed Aug 25, 2004 8:43 am Post subject: |
|
|
Are you saying:
1.some commands work the first time, or that
2.none will work without multiple sends?
If it's 1. please tell me the ones that work. If its 2. I think I will eliminate the multiple repeats. _________________ -Jon |
|
Back to top |
|
|
matzh9
Joined: 21 Aug 2004 Posts: 12
|
Posted: Wed Aug 25, 2004 9:25 am Post subject: |
|
|
Hi Jon!
It's 2. I feel like a very spoilt child so while your at it maybe you could add INFO which is Hex 00 1B in IR.EXE (EFC 218). I think that would help as a reference for decoding the other buttons as its ouside the 0-9 sequense.
Thanks again for you patience,
Matz |
|
Back to top |
|
|
jon_armstrong Expert
Joined: 03 Aug 2003 Posts: 1238 Location: R.I.P. 3/25/2005 |
Posted: Wed Aug 25, 2004 4:53 pm Post subject: |
|
|
I just uploaded a new lir file that has all commands (I think). I tinkered with some of the timings and I think this should work better. Same link as before. Open force.lir in IRMax to find the commands that are labeled. They start at 10 (numeral zero) and go from there. _________________ -Jon |
|
Back to top |
|
|
matzh9
Joined: 21 Aug 2004 Posts: 12
|
Posted: Wed Aug 25, 2004 6:37 pm Post subject: |
|
|
Hi again Jon!
I'm sorry but it is like before, 2. The good thing is that every single code works accurately after a few (5-15 randomly) repeated sends.
When you alter timing is that then the pulse width or the time between the repeats.
It is very hard to decode the binary values from the data/graph as the pulse widths differs all the time (OFF). Shuld I read them as in IR.EXE 0=00 1=01 and so on? Is there anything I could try on my side. I can now grasp the Pronto to LIR conversion or the x, -y, z format if that is easier to discuss.
I ask, not because I'm in any hurry, but i feel you are spending loads of your private time on my problem. It is sure appreciated.
Matz |
|
Back to top |
|
|
jon_armstrong Expert
Joined: 03 Aug 2003 Posts: 1238 Location: R.I.P. 3/25/2005 |
Posted: Wed Aug 25, 2004 7:10 pm Post subject: |
|
|
Matz,
Are you using any sort of IR distribution system (I assume you are).
Or how is the signal getting from the Leopard to the Force DSS receiver? is it wired wireless etc. I am thinking this may affect what is going on. If possible can you wire the emitter to the Force directly from the Leopard and test it directly?
As far as what I am doing I have created a lir code generator in a spreadsheet. I basically ran several tests to see what the actual pulse lengths are vs the description. The timing of the Off pulses are skewed particularly at the short end. I fixed those and made a minor tweak to the gap between frames. So I can fix the timing easily and the changes are automatically made to all the commands.
If I can't get closer, if you have Excel than I can send you my spreadsheet and you can tweak. Right now, I'll go back to just sending a few numerals until we get closer. The problem is pasting in the lir format commands into IRMax is a little tedious.
BTW, do any of the commands in the last batch work better than the others? _________________ -Jon |
|
Back to top |
|
|
matzh9
Joined: 21 Aug 2004 Posts: 12
|
Posted: Thu Aug 26, 2004 4:22 am Post subject: |
|
|
Hi Jon!
I have built my IR booster but I hope it does not scare you of. I have'nt had any problems with my other devices Sony, Panasonic (VCR), XBOX, Nokia (previous DSS), InFocus. It is a very simple circuit connected directly to Leopards IR-out. I took the stringest doid I could find and sacrifised some angle. If I remember right its 20 degrees wide. I also have an emitter with short reach so I could try to use that glued on the Force if you suspect my circuit is slow or "uneven".
If any of the commands works better than others I would say 7 and TEXT.
I do understand that IRMAx is very heavy to use and the sequence is not short either. Did you know that the LIR file format is just blocks of 256 bytes with IR commands stapled after each other. You could actually dump the 256 bytes directly to a file and use CMAx to download. CMax stops after the last full 256 bytes block, much faster... If you could dump just 7 (or better a few variants of 7) to a file I would then get it as IR#0. Another option is a Pronto hex and let me do the tedious work. I guess if you represent the Pronto with another frequensy 25kHz you could achieve the 40us resolution.
I'm travelling today but I'll read your response should they have WLAN at the hotel. Back home tomorrow.
Thanks,
Matz |
|
Back to top |
|
|
jon_armstrong Expert
Joined: 03 Aug 2003 Posts: 1238 Location: R.I.P. 3/25/2005 |
Posted: Fri Aug 27, 2004 5:07 pm Post subject: |
|
|
I think I found the problem. I had entered the fixed data as 0D 0F 44 0D rather than 0C 0F 44 0D. Try the file at the same link again. _________________ -Jon |
|
Back to top |
|
|
matzh9
Joined: 21 Aug 2004 Posts: 12
|
Posted: Sat Aug 28, 2004 5:10 am Post subject: |
|
|
Jon you got it!
Now its 100%. The only little problem is that the code is interpreted as 2 button presses (most of the time).
If I take away the last repeat its always sees a single button press but then it instead misses every now and then. The same behaviour if I delete both repeats. I guess the 2 repeats are required and I should instead play with the time between the burstsgroups.
I'll play around for a while and let you know.
Thanks a million,
Matz |
|
Back to top |
|
|
jon_armstrong Expert
Joined: 03 Aug 2003 Posts: 1238 Location: R.I.P. 3/25/2005 |
Posted: Sat Aug 28, 2004 7:31 am Post subject: |
|
|
Matz, do you have Excel? I can post my spreadsheet and you can make those corrections globally. I did go back and check the gap between the frames and it's pretty close to the OEM remote (assuming the device upgrade is correct), but usually when you get multiple commands in a protocol that has a repeating style command, the gap between frames is too long, and it treats the successive command as a new command. So you might experiment with one command by shortening the gap between frames by 1 mS until you get it right. _________________ -Jon |
|
Back to top |
|
|
matzh9
Joined: 21 Aug 2004 Posts: 12
|
Posted: Sat Aug 28, 2004 10:54 am Post subject: |
|
|
I do have Excel and your welcome to post the spreadsheet. I'll start playing with the gaps now.
Matz |
|
Back to top |
|
|
jon_armstrong Expert
Joined: 03 Aug 2003 Posts: 1238 Location: R.I.P. 3/25/2005 |
Posted: Sat Aug 28, 2004 2:12 pm Post subject: |
|
|
Here is the spreadsheet . I protected the cells that are calculated but didn't put a password so you can unprotect it. You can change frequency, gap, mid frame gap and OBC. _________________ -Jon |
|
Back to top |
|
|
matzh9
Joined: 21 Aug 2004 Posts: 12
|
Posted: Mon Aug 30, 2004 6:55 am Post subject: |
|
|
Thanks again Jon!
I have tried with different gaps between burst groups but it does not help a lot. Then I started to examine learnt variants. The learnt code has init plus TWO repeated sequences and is garbage. However the time between bursts is very close to what you put originally.
I think the problem for Leopard, as you said before, is that the IR signal has so short burst pairs that the Leopard IR detector can not cope with them and as a result puts them with wrong timing and length. The groups are separated by a longer interval and therefore that time space is more accurately represented.
Now I’m running your variant but I skip the last repeat. This is what works best. One missed button out of ten but no doubles. I’m happy with this a suspect that the missed buttons are my IR booster and/or distance to blame.
My conclusion is that your code is accurate but Force looks at total of three bursts groups.
BTW I will try altering the carrier frequency slightly. That is one thing I forgot to try.
Matz |
|
Back to top |
|
|
|