GTA04 Block V4

[Ian Stirling]

Andy Green wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Somebody in the thread at some point said:
> | Werner Almesberger wrote:
> |> Ian Stirling wrote:
(of the stm32)

> |>> I think you can always mass-erase the chip though, so a USB bootloader

> | If I read it correctly - you can secure the code - p22 of
> | http://www.st.com/stonline/products/literature/pm/13259.pdf says you
> | can't overwrite pages 0-3 in read-protected mode, without mass-erasing.
> 
> We've been through this with the NOR.  We really don't like write-once
> stuff either for us having to make a golden implementation for the
> factory or for the user not being able to update it in a package.  If
> malicious code mass erased this bootloader section we are bricked.
> 

Fair enough.

It occurs to me that another solution that isn't quite ideal might be 
possible.
Make a 'debug board' available, but much, much cheaper.
This would be something like (in basic form)

A 20ish way 0.5mm FPC connector, replacing the existing debug connector, 
which - for most users - they simply connect with a basic debug board 
comprising only a 9 pin serial socket, a FPC connector, a couple of 
resistors and zener diodes, and a jumper.

Plug the jumper in, you get bootloader mode on the STM. Don't plug the 
jumper in, and you get serial-console or something, from the STM32.

Supply with a picked USB-serial adaptor that can live with 3.3V thresholds.

This would solely be for recovering from a bricked STM32.

The advanced debug board is either an existing one, with a converter 
board, or a new one.

As a possible plus, this connector could also support add-ons, rather 
than the fragile expensive to make existing one.

$5 for a serial-USB converter. $4 BOM, $1 PCB, $1 assembly (in moderate 
quantity) - plausible to sell at $20.

(for production programming, I'd imagine a bed-of-nails, with power, 
ground and rs232 might be more convenient than the FPC.)