On October 2, 2006 at 08:37, leemcg said...
old B+W version could sometimes learn something
Do you still have that old Pronto?
Some of the experiments that may be needed to solve this would be easier with an old (pre NG) Pronto. Other experiments need a second PC readable IR learning device (such as a second Pronto) in addition to the Pronto being tested.
On October 3, 2006 at 02:50, leemcg said...
I'm now able to
learn something on my 7000 pronto every time.
If I understand you correctly, a short press always learns something but the learned command never works.
it actually remembers
and sends discrete codes.
I was assuming that the remote remembers the state of all the settings of the AC and each IR command is a complete specification of the entire state. What you observed supports that theory.
The samples you learned are obviously truncated. The signal continued (hopefully not much longer) but the Pronto stopped recording.
I don't know whether a Pronto can send a slightly longer signal from hand edited Pronto Hex than it can learn. I hope so.
Also notice the 0454 in the learned signal. That is a long enough gap that an older Pronto should be able to send everything up to that point as one IR signal in a sequence and then send the rest as a second signal with little enough delay between to match that 0454. So if we are lucky and there is only a little lost at the end of that signal, and we discover the lost part (see below) the older Pronto could send the whole signal.
An NG Pronto can't delay that little between steps in a sequence. Even asking for no delay you get more delay than that. But maybe the AC doesn't care about the length of that long gap. Maybe it will be equally happy with a much longer gap.
To find out the beginning (hopefully all) of the missing part of the signal, you need to learn the SAME exact signal again but somehow make the Pronto start recording a tenth of a second after the original remote starts sending. Then I can match up the begining of that learn against the end of the learn you already posted.
In learning, I assume you press some button on the Pronto as the last step of telling it to start learning. Some small fraction of a second later the Pronto actually starts learning. Some larger fraction of a second (or longer would have been OK) later you press the button on the original remote.
You would need to experiment to tweak that timing. First try pressing the two buttons together. Maybe the time an NG Pronto takes from the button press to actually starting to learn is close enough to a tenth of a second to get what I want.
If the Pronto learns the wrong chunk of the signal, you should be able to adjust a twentieth of a second in either direction by starting with one finger higher than the other and move both down pretty fast to press the buttons at almost the same time.
If the Pronto starts so late (on simultaneous press) that it misses the whole signal, you need to experiment with pressing the Pronto button then really fast but not simulataneous press the other button.
In all that it is important (and maybe difficult) to keep both remotes aimed and stationary when pressing the buttons. If using one hand for each remote and going fast makes you jerk the remote while pressing, that would wreck the results.
If you post a few samples with your guesses of where they were relative to simultaneous, I can probably match enough pattern to extend the known part of the signal.