Hi Ron,

I looked at the port capture when I got my UV8D-A, and finally just figured out the checksums with the help of your email. Thanks! :)

The model I have is UV8D-A. It may be different than the non 'a' version. This is my ID capture:

000046: Write Request (DOWN), 12.09.2014 21:03:31.235 +0.0 (1. Device: Prolific USB-to-Serial Comm Port (COM3))
Buffer size: 0x5 bytes

 7D 80 FF 00 7F


}€ÿ. 
000049: Read Request (UP), 12.09.2014 21:03:31.240 +0.005 (1. Device: Prolific USB-to-Serial Comm Port (COM3))
Buffer size: 0x30 bytes
Status: 0x00000000

 7D 80 00 2B 4B 47 2D 55 56 38 44 2D 41 00 01 02
 62 5A 00 03 19 74 06 02 62 5A 00 03 19 74 06 00
 CC 77 C0 01 0B 06 66 00 CC 77 C0 01 0B 06 66 9E


}€.+KG-UV8D-A...
bZ...t..bZ...t..
ÌwÀ...f.ÌwÀ...fž
We may encounter some terminology 'friction', seeing as how you were likely taught proper English across the big pond from where I learned my wronglish. I apologize in advance...

On Sep 11, 2014 8:37 AM, "Ron Wellsted" <ron@wellsted.org.uk> wrote: 

A Gotcha: the first identify packet returns a bad checksum, subsequent
attempts return the correct checksum... (well it does on my radio!)

The factory software initially does two consecutive identify reads, with the only difference being the checksums. I initially suspected the dual VFO nature of the radio, and you have since discovered that the first ID read checksums incorrectly. So, I wouldn't worry about that, just double ID in CHIRP, or don't bother to checksum the ID frame.

Channel memory structure sussed out so far:
0x0900:0x477f  channel data 000:199, 0x10 length blocks, FF nulled.
0x4780:0x66bf  channel names 000:199, 0x8 length blocks, 00 padded.

My send read request looks like this (and the breakdown):
7d 82 ff 03 09 00 40 cd chan001:004data

 7d 82 ff read req

           03 count

                09 00 mem location

                        40 bytes requested

                            cd checksum
And the reply:

 7d 82 00 42 09 00 ... 1d chan001:004 data

7d 82 00 read ack

             42 count

                  09 00 mem location

                          ... data payload

                               1d checksum

When I get around to learning Python, I might be able to make something useful happen in CHIRP.

-Dana