Re: IR signal not as stable as original remote

[Bengt M. <bu...@be...>]

On 08/07/14 10:49, Jiechao Li wrote:
> Hi, everyone
>
> I have been using lirc on my raspberry pi to control Haier A/C,and it
> works fine now.
>
> The strange thing is that, when I use lirc to send IR signal, it's not
> that stable. Sometimes it works, sometimes I have to try multiple times
> to make it work. But with the original IR remote, it works perfectly
> almost everytime. The question is why mine is much less stable than the
> original one?

With the information you give, it is impossible to go into details, so I 
will be general. The most common cause of unreliable operation is 
"dirty" learns. An IR signal is a sequence of durations, and "learning" 
records these. As all physical measurements, these have errors, and when 
you transit your learned signal, you are sending the slightly 
erroneously learned to the device. The receiver has a certain error 
tolerance, but sending and receiving adds yet more errors, and at one 
point the signal is not recognized any more.

Generally speaking, the best thing you can do is to identify the learned 
signals as being from a well-known protocol (e.g. NEC1, RC-5, Sony, 
etc.) and replace the learned signal with a generated signal. The latter 
then contains NO learning errors, and in many cases is just as reliable 
as the original remote -- or even more reliable, in the case of 
stronger/better sending hardware.

Unfortunately, not all IR signals identify as a known protocol. In 
particular, air conditioning units (is your "Haier A/C" one such?) are 
known to have long and "hairy" IR signals.

I have written a tool, IrScrutinizer 
http://www.harctoolbox.org/IrScrutinizer.html which may help you to 
identify, analyze, and re-generate your IR signals.

Greets,

Bengt