Newbie, RS 15-2103 and memory overflow

[gimac]

After some successfull experiments with oneforall remotes and jp1, i tried on a rs 15-2103 of a friend.
just having a protocol update + device update IR615 says that there is memory overflow and i should remove protocols or device (but it was the 1st device and 1st protocol !).
Ok, i restarted with rs15-2103 2k version rdf (previously i used the 1k version rdf), and i tought to replace the internal eeprom from 24c08 to 24c16.
I was surprised but IR tool gave me exactly the same overlow messages, and looking to the 2 files (1k and 2k version) the memory maps are exactly the same, so the same space available for protocols and devices.
Can somebody help me to understand?
thnx

[The Robman]

The 15-2103 does indeed have a very small 512b upgrade area, compared to the standard 1536b, and because of the learning section, the 2k version of the remote doesn't appear to give you any more usable memory (at least, according to the RDF).

There is a way to use the extra memory, but unfortunately IR.exe doesn't support it. I have just created a special hacked RDF for you which will give you much more memory for upgrades, but you will need to remember to do a manual patch everytime that you change your upgrades.

Here's the RDF...
https://www.hifi-remote.com/forums/dload ... le_id=2219

To use it, download your current memory using IR.exe, then copy your upgrades into it. When you are done, you should go to the Raw Data tab and copy the 4 bytes found at address $400 over to $100 and hit APPLY.

You won't need to re-do the patch until you make some changes to any of your upgrades. Whenever you do make changes to them, re-copy the data from $400 over to $100.

[gimac]

I'll try during the week-end, and i'll let you know.
Never expected such prompt answer and solution.
Great!
again many thanks
gm

[gimac]

Sorry to contact you after long time, but i just detected that i was using a wrong rdf file (1k version) instead of 2k version. few days ago it tried to load a new device+protocol and i get again the memory oveflow. Resource were available still on learned keys, but if i'm not wrong you 2k hacked version were supposed to reduce the learn memory to give more room for the rest.
So i restarted, but nothink works.
A small summary: i've 15-2103 with 2K ee (it was 1k but i replaced by a 2k). I've 3 different rdf: 1k version, 2k version, 2k.hacked version.
I keep only 1 at time in my rdf directory. The ee works fine, i can read/write all locations (also over 0400x, and 7ffx), so i can upload to remote and download to remote.
Step 1: 1k rdf
i put 2 devices it (just device upgrade and any protocol), all is OK
Step 2: 2k rdf (not-hacked)
same 2 devices, first read it says i've got a wrong checksum, ok after first write. Nothink works, i can see the device programmed with right codes, i can see the device upgrade loaded, BUT nothink works. I learned the send codes with another remote (O4All 8910) and i found the 2103 is sending a wrong protocol (??)
Step 3: 2k rdf hacked
same as before, with one more problem. In the raw data area, now there is somethink written over 0400x, so i copied from 0100x-0103x to 0400x-0403x , as soon as i change location 0400x and i do apply i got a dialog box:
Warning Protocol list not found after device list
and then the value in 0400x return to the original, so i can't change anythink.
I've some doubt about 2103-2k.rdf, and so i can't fix 2k-hacked.
Rob can you help me?
thanks

[gimac]

A small update after many trials:
the 2k version (NOT hacked so the std one) works if the rdf is modified for the checksum:
[Checksums]
^$000:$002..$7FD
to
[Checksums]
^$000:$002..$7FE
This is taken from another post in this forum. This way IT WORKS, but still
with the memory limitations for devices and protocols (the upper 1k eeprom is not used ...)
So i modified the 2k-hacked version from The Robman, putting the same checksum, but the remote still don't work, it's sending a different protocol.
It's looks clear to me that if the checksum is wrong the remote doesn't look into the eeprom, may be it's having some defaults.
How can i be shure about 2103-2k-hacked version has a correct checksum?
thanks

[The Robman]

Have you been saving your old IR files as you go along (and if not, you really should have)? If you have, have you tried just re-loading one of your old files and verified that everything still works?

I'm asking such a basic question because I'm a little confused as to what you've been trying to do? For example, if you've modified the remote to have a 2k EEPROM, why would you even try using the 1k RDF, given that you know that it's incompatible.

If the checksum gets corrupted, the remote will no longer recognize the EEPROM, so it won't use any upgrades that you have loaded.

The only way that IR will know that there's a checksum error is if the checksum in the remote doesn't match the formula programmed into the RDF.

If you have been using a wild assortment of different RDFs and have been doing other wild experiments, I would recommend that you completely clean out the EEPROM and start over. To do this, select the "Advanced" menu item in IR and then select "Initialize to $FF". This will reset the EEPROM to all $FF values. After doing this, do a 981 reset, this will reset the EEPROM to what we call a "virgin" image.

If you are using the hacked 2k RDF that I created for you, you need to remember to copy the data from $400 to $100 after you make any changes to your upgrades.

[gimac]

Thanks Rob, and sorry to ask for help, but as you said i was really confused after doing many trials.
Let me resume the situation:
the original remote had 1k ee, i changed to 2k to get more space for my upgrades. I kept all original rdfs, so i've 3 of them 1k, 2k, 2k-hacked.
At the beginning the 2k and 2k hacked was not working at all, programmed 1 simple device upgrade (an european tv set), but the remote sent wrong protocol.
I just tried 1k (even with 2k eeprom, but i said it will used half of it), cleaned the ee before and the remote was working fine.
So i moved to test the 2k, not working (IR programmed the same contents of the 1k version, but the first 4 bytes i guess checksum were different, and of course the upper memory $0400 to $07ff were filled by FFs). Reading other experciences in this forum , people changed the checksum-setup of the 2K version (see my previous post).
So i did it, and the 2k rdf version with 1 simple device upgrade worked fine. I tought i found the problem for the 2k-hacked version you modified.
So i cleaned the eeprom, and changed the checksum-setup in the rdf, and tried again.
Not working at all.
Ofcourse i always copied the 4 bytes 0400-0403 to 0100-0103 as you recomended.
I did also an additiona trial, i programmed with 2k working working version, than i changed the rdf to 2k-hacked, download from the remote (so the upgrade was after $0100), re-programmed the device upgrade (with same upgrade) and i saw the upgrade going to $0400.
The bytes from $0100 to $0142 were exactly the same as $0400 $0442,
so i tried to leave the first 4 byte to 01 37 01 3D, or change them to 04 37 04 3D. Not working in any case (and i guess it should!).
The other difference in the 2 cases is the checksum.
I've a dump (.txt) of the 2 trials: 2k not hacked working, and 2k hacked not working if it's usefull (but i don't know how to send it to you).
I just add that in the RDF folder i kept only 1 rdf at time (just to be shure).
Many thanks again for your patience.

[The Robman]

Are you saying that this never worked? I assumed when we never heard back from you 6 months ago that it was working.

[gimac]

Rob, as i apologized in my 3rd post here i tought i was using your rdf, but IR was using the 1k rdf instead, and after a message of wrong checksum it worked. Only recently i tried to add a 3rd device (device+protocol) and i went back to the memory problem.
I understand that you can be disappointed, I'm so sorry!
But i can confirm (now) that it doesn't work.