has anyone else experienced a difference between the database codes and learned codes?

the codes im learning dont work as well, often take a few presses, compared to those in the database, which always work on the first push, for the same function.

i have tried changing the duration, but that doesnt produce a different result.

i have decoded/encoded the learnt codes with irtool and makehex.

Sounds like the code has not been learnt properly and is dirty.

i think (and hope) i figured it out. the code i was making was encoded with the nec1 protocol at 38K. i tried making it at 40K and it works a lot better. is there somewhere in the makehex program that would have told me this?

There are a small number of devices that need nec1 protocol at 40Khz. But there are a very large number of devices that use nec1 at 38Khz. Learning flaws that shift the frequency 2Khz from the true value are common enough, that an nec1 signal decoded at 40Khz is equally likely to be a signal that's really at 40Khz or one that had its frequency learned wrong.

I don't see where in the Makehex package I could have given any info about this situation, that would be noticed by the few people who need it without being a big distraction to the majority who don't (but I'm open to suggestion).

In the DecodeIr documentation for nec1 at
[Link: john.fine.home.comcast.net]
I should have dealt with this issue. Apparently I didn't.

Please reply with any suggestions you have on how to document this issue to be clear and concise, noticed by those who need it, and not confusing for those who don't.

My first cut, which I don't think fills those requirements is:

A few devices use nec1 protocol at 40Khz, rather than the typical frequency. When getting a decode of nec1, if you notice that the frequency is closer to 40Khz than to 38Khz, examine multiple learns from the same device to estimate whether the 40Khz frequency is a learning error or a true characteristic of the device. If the 40Khz is correct, there are methods in JP1, or MakeHex (whichever you are using) to reproduce nec1 at 40Khz rather than the usual frequency.

sorry, i didnt mean make hex. i meant decode ccf or irtool. looking at it now, ir tool does tell me that frequency of the code. however, i just looked back at the codes i learned and they all decode to ~38kHz. i dont know if this was a problem with learning on my remote? has anyone been able to learn codes that are 40k/s?

the warning above is sufficient, i dont know how many people have the issue that i do though. a more general mention could be that if you try codes generated at 38kHz and they dont have acceptable range and often take a few pushes before functioning, try generating the codes at 40kHz.

as an aside, i just looked at the frequency of the codes in the ccf's for harman kardons and they are at 38kHz. i dont know if other people are having trouble with the harman kardon codes posted here?

if so i can email the codes that i have.