Re: [avrdude-dev] new USB-based programmer with Igor USB

[Jan-Hinnerk Reichert]

On Wednesday 26 November 2003 23:50, Jeff Epler wrote:
> On Tue, Nov 25, 2003 at 08:17:28AM -0600, Jeff Epler wrote:
> > Because "it could be done", I tried out the avr-based USB device
> > code written by Igor Cesko, and quickly had a very versatile,
> > USB-controlled device. (http://www.cesko.host.sk)
>
> [...]
>
> I've improved things further with related changes to the USB
> firmware and avrdude.  By cutting as many corners as possible,
> including using timings that I only know are safe due to the long
> packet times of USB compared to the times in the data sheets
> (tWD_PROG, etc), I am able to program at about 100 bytes per
> second.  The 270-byte program I mentioned earlier takes 3 seconds
> to erase + program + verify, and a 2k firmware takes 16 seconds. 
> Verify accounts for fully half that time, so it takes 7 seconds to
> erase + program a 2k firmware.

It just came back to my mind, that there seems to be a much better 
solution to programming via USB.

The new revisions of the USB-chips from FTDI have a bit-banging mode. 
These chips do pretty much everything one can dream of ;-)
- full-speed USB
- level-shifters at the I/O (3.0-5.0V)
- driver for linux at http://ftdi-usb-sio.sourceforge.net/
- bit-bang mode

Take a look at http://www.ftdichip.com and especially at
http://www.ftdichip.com/Documents/AN232-01_BitBang.pdf

The bit-bang mode sounds real cool:
You set a data-direction-register and program the clock rate at which 
data should be set/read.
Then you sent a packet of data and it gets latched out at the 
specified rate (as long as there is something left in the buffer).

However, the application note is a bit unspecific about read back 
(e.g. it does not tell if readback goes on when there is no data to 
shift out)

And now the usual disclaimer: I am in no way affiliated with FTDI Ltd.

/Jan-Hinnerk