Theory.

Yes, there are better solutions - a table of fixed baudrate dividers (as in DS89C450 - there's a limited number of baudrates and crystals frequencies in practice, but the solution is somewhat rigid); a better autobauder calculating the average across both "positive" and "negative" (SPACE and MARK) pulse and several of them, rounding the result rather than truncating (that's how the original SST bootloader works, but AFAIK SST did not "burn" the bootloader into their chip so it was up to the user); having a command to explicitly load the divider, allowing to do a "sure" autobaud at low baudrate and then switch to higher (P89C51RD2, but my memory is fading, I may remember it wrong).



Practice.

If it is only for you in a given product, a fixed baudrate would do.

There are two ways to accomplis this.

One is to patch the original bootloader. The easies thing is to get the v7 bootloader with "updater" from esacademy, that in the first 8k (except the first 3 bytes) contains the image to be burned into block1; patch that, and then run it exactly as intended. This might be something I would do given I've played with this before (provided that I manage to recall the details and the place where I stored them... :-( ).

You can write your own bootloader, too, residing together with the firmware in the "user" space in block 0, using the original one only for the first firmware programming and then as the IAP "machine". And if you upgraded to version 7 - the "real" one, published through esacademy, not the "crippled" one which is now apparently burned in factory by NXP, you can also disable further entry into the "original" bootloader. But you can also device your own method of entry and leave the original as a backup plan.



A cautionary note.

As you are surely aware, for years I was preaching that the P89V51RD2 are rebranded SST89E516 (originally they were SST89E564, but they were exactly the same as far as the user's view is concerned) with bootloader burned in factory. SST has been bought by Microchip somewhere a year ago. Go to Microchip site, look up the SST89E516, and have a look at the top of the datasheet. It's really not surprising what you can see there.

Now go to NXP site and listen to the guy in the CortexM3 promo video I referred to already.

Draw your own conclusions...

:-(

Jan