On 05/05/2013 10:52 AM, "Ing. Daniel Rozsnyó" wrote:
I have started to compare TPI and PDI - seems very similar. The TPI
has a dedicated reset wire, so PDI looks like a subset of TPI, with no
modification required.
last year i considered implementing PDI on FTDI but i gave up after
learning that FT232R is unable to produce long and stable sequences of
clock signal because of its hardware bug - the chip spits out all
received data very quickly instead of buffering the packets and clocking
output pins according to the configured frequency - see "TN_120 FT232R
Errata Technical Note" - "3.1.2 BitBang Mode variable Pulse Width"
http://www.ftdichip.com/Support/Documents/TechnicalNotes/TN_120_FT232R%20Errata%20Technical%20Note.pdf
and this is very unfortunate because PDI requires at least 10kHz clock -
ideally we would configure the output clock to 20kHz (or more) and send
an USB packets of 01010101 - but they screwed it and the result is a
quick burst of clock followed by a pause (and pausing the clock
disconnects PDI) :-P
so i gave up and decided to implement PDI in USBASP that is so
ubiquitous and cheap
http://szulat.blogspot.com/2012/08/atxmega-programmer-for-050.html
but of course the FTDI implementation would be so much better (if it is
really possible) and i wish luck to anyone trying to accomplish this goal!
sz.