Graham,
> > I never intimated that you were stealing.
>
> Well, whatever word you use instead of stealing, it amounts to the
> same thing. I'm either doing something Atmel can justifiably object
> to or I'm not.
I was only making you aware of what I've been told by Atmel. If you
need clarification, speak to Atmel.
> > If, however, you ship the device already
> > programmed then there may be a problem.
>
> I would prefer to ship it blank, which is why I think it's good that
> it's very quick to load up. If Olimex ship their clone programmed,
> this quick load is not an advantage to the end user, though. Even
> so, if Atmel specifically allow AVR Studio to be redistributed and
> this includes the software for the JTAG-ICE, I don't know if they'd
> care whether that software was redistributed on a CD-ROM or in an
> ATmega16's flash memory.
I suspect the horse has long since bolted and they can't put the genie
back into the bottle, to mix some metaphors. However, they are
certainly aware of the situation.
> > ... I have heard from people within
> > Atmel: the Mk II isn't compatible
> > with the Mk I because of 3P cloning.
>
> (3P -> third party -> someone other than Atmel or the end user, that
> is, someone like Paul and his compiler or me and my JTAG-ICE clone.)
>
> Well, this could be from the horse's mouth or from some other
> orifice.
Hmm, it's from the person responsible for the JTAGICE.
> The Mk II has a USB interface on the PC/host side and
> supports DebugWire for the newer ATmegas as well as JTAG. The
> internal MCU has got to be different (bigger/faster?).
It has two Mega128s.
> What do we
> mean by isn't compatible, anyway? Won't run software written for the
> Mk I?
That about sums it up.
> Isn't a software upgrade to the Mk I hardware? Atmel need a
> new ICE for DebugWire. There are PCs out there with no serial ports,
> only USB. Hence the Mk II. Obviously, it won't work with Mk I
> software and its software won't work with the Mk I. I can't see third
> party clones as a driving force for this. The clones give on-chip
> debug access to people who might not shell out $300 to get it. This
> puts the ATmega chips in a better light and could win business for
> Atmel. This may sound as if I'm pooh-poohing your input, but I value
> it greatly so please don't take my comments as other than spirited
> discussion.
Hey, I purchased two Olimex JTAG ICE clones because we expect to have
customers that use them. I also have JTAG ICEs and a JTAG ICE Mk II.
We want to ensure our software works on the widest range of hardware
possible. I certainly understand the difference between $40 and
$300--almost an order of magnitude, and it won't be lost on customers.
> > ARM certainly [make money out of tools] ...
>
> Ah, but ARM don't make chips. Their business model is to license the
> ARM architecture and the development tools. If Atmel wanted income
> from the tools, why would they give AVR Studio away?
They give it away in *exactly* the same was as TI give (or gave) away
their tools for the MSP430: there are customers who want assembly
development only and don't need compilers. Microchip do the same, as do
Dallas. On another note, it's the only software (to my knowledge) to
support the ICE{3,4,5}0 as Atmel keeps the way to drive them under
wraps.
My take on all this is that it doesn't make any difference to me whether
you make your own JTAG device or not. Now, to make it really useful,
shove an FTDI USB<->RS232 chip on the end of it and you have a (slow)
USB JTAG solution. That's a big value sold on convenience.
I totally agree that Atmel needs is a *cheap* JTAG solution; TI's FET
for the MSP430 is $10 from Olimex and the ARM one is $19 IIRC, whereas
the AVR one is $40. $300 for the same from Atmel is nonsense to be
sure, so I would actively support low-cost cloning if I were they--but
I'm not.
I don't expect the Mk I to get debugWIRE and I don't expect the Mk II to
be easily clonable.
--
Paul Curtis, Rowley Associates Ltd http://www.rowley.co.uk
CrossWorks for MSP430, ARM, and (soon) Atmel AVR processors