On 2/11/2001 at 1:49 PM Peter Graf wrote:
>On the Q40 I minimized the availability problems by restricting myself to
>components especially chosen for that reason. But I had a price to pay:
The
>chips are not cheap (because I use 4 of them) and so small that it was
>endless work to squeeze the logic into them. And I had almost no
>flexibility for changes left. (For example Lattice has an example design
>for a Video controller, but they use two big chips were I need only one
>small chip.)
GoldFire requires the logic to be in one or at most two chips, for one main
reason: restricted space. Also, there are several wide busses that actually
take a lot of space as tracks on the printed circuit, as well as pins on
the various chips. However, if they are all routed within one larger chip,
they don't take any extra space as such. It's just a matter of lateral
thinking. It takes a bit of time to figure out all costs but I actually did
that.
>That you live in Eastern Europe makes hardware development more difficult.
>But even here in Germany I would have problems to get some of your chips,
>for example the MCF5102PV40.
I currently live in the US, actually. It's not a problem generally getting
any part - as long as you want 50-100 of them! That would not be such a
great problem if it didn't imply a lump sum of several thousand $, which
even if I did have to spend, would be spent on matter of a higher priority.
I went through the loop between the manufacturer telling me to get samples
from their distributors, who don't have any and don't keep a stock, who
want you to go back to the manufacturer or wait until someone else buys
some and has left-overs so many times I'm abut ready to be sick when I even
think about it any more.
>>Problem 2:
>>There is simply no way I can do the work on the initialization software
and
>>QDOS/SMSQ modifications involved in getting the GoldFire to work.
>>From my own experience I am sorry to have to say: There is probably no
>other way but to write the initialization and some simple driver software
>on your own.
Well, then, prepare to wait for a couple more years, provided I don't give
up before that.
>>This is NOT easy. Unfortunately, this also means that the design
>>itself is really obsolete when it's finished. The MCF 5102 has already
been
>>superceeded by two newer CPUs and we can't use it because it's such a
>>problem writing and modifying the necessary software - in fact, it's
>>unlikely that most of the features I've lost a whole lot of time thinking
>>out and designing will ever be used.
>
>I can only underline this. Many users think: If a hardware feature is
>integrated, it will take time, but eventually there will be software for
it.
>It isn't true. Several possibilities of the Q40 will remain unused under
>SMSQ.
Well, i supose it wouldn't be too hard to take out several parts of the GF
design such as second/ipgrade CPU support, and just design a simple
interface to SDRAM.
This means that for a newer ColdFire CPU about half of the logic would have
to be redesigned, as well as, of course, the PCB. The PCB is modulairly
designed - everything else would still be the same, so that would cut the
developement time of a potential successor considerably. I'm willing to
take votes on this.
> With a complex sound chip we
> probably still would have no SMSQ/QDOS sampled sound support at all.
Well, the complex sound chip is not really TAHT complex, fortunately. But
in any case, if no new features are added with every, increasingly rare new
product, then we might just stop bothering doing anything at all.
>>Yet, on the other hand, people complain about not having the features
>>on the QL they have on the PC.
>
>And not only this. Many expect PC prices for QL hardware. No joke.
Seconded. But this is, quite frankly, impossible.
>Hard times for QL hard-ware these days.
Well, it's really been that way for years now.
Nasta
