>> There still appears to be a fair market for expansion interfaces for
>> the QL.
>> I realise that many users are holding out in hope that the GoldFire (or
>> whatever it is eventually called) will soon be released, but I
>> appreciate that this is still several months away, if it ever reaches
>> the market (sorry Nasta).

No apologies necessary - at this stage I really cannot make promises,
except that I will fight to make it real to the bitter end. Of course,
hopefully the end will not be bitter :-)

>> Surely there is still a market for a small batch of Super Gold Cards to
>> be made up, or am I correct in thinking this may now be impossible due
>> to lack of availability of chips??�

> Indeed.  The EP1810 has proved impossible to find... these chips are
> OTP (one time programmable) so are not to be found on the reclaimed
> chip market.

They can be found on the 'old stock' market though, but there are three
typical problems:
1) The dealers have no idea what it is and are stocking returned programmed
parts - useless.
2a) They stock the 1810A which is the one that does not work, and they want
lots of $ for it - useless
2b) They stock older 1810 that does work but want $50 a piece, no guarantee
- effectively useless
That being said, it is not entirely impossible to redo the SGC logic in a
different, more easily obtainable chip, provided one has access to the
source, or just plain designs from scratch. SGC is barely feasible done
that way, assuming it is sufficiently expanded (for instance, with 16MB or
RAM). GC is more or less a futile effort, even if it is possible.

>It would be an expensive exercise to build. Usually it is not economic
>to think of less than 50 - certainly for pcbs. I think Nasta should chip
in here.

Economically, another batch of SGCs or SCG clones falters at three
components:
1) floppy controller
2) real time clock
3) Logic chip (assuming an exact clone is made)
and possibly: 4) CPU - not a lot of bang for the buck, fortunately, it is
available as ;old stock' or 'refurb' so there is a way around that last
one.

>When Goldfire arrives, it is likely to release secondhand SGCs into the
>market. Also I would not like to think of a new SGC production damaging
his
>sales.

It is difficult to say wether it would damage GF sales (and I'm glad to see
someone still thinks of it in terms of 'when' rather than 'if' :-) ). With
respect to speed and extras, anything SGC like can hardly compare with the
GF. The speed of the GF will roughly equal that of the Q60/66, except for
floating point since a 68EC060 CPU (without FPU on it) is used. It's speed
is about 13x that of the SGC. It is also doubtfull that a SGC (clone) would
actually be able to compete in the matter of price! The reason is that all
of these older chips tend to cost quite a lot because they are so rare now.
Suppliers are more than happy to hike the prices up seeing you desperate to
have the parts.

I would not be fair, however, not to mention a few possibilities. It would
be possible to construct a 'sort-of' SGC clone under some conditions:
1) No floppy interface
2) No RTC
3) Works with Aurora only
On the other hand, both 'defficiencies' could be offset by extra features,
such as:
1) System is CF card based (floppy would be an expansion, and not really
cheap, either - although it is entirely possible to make older style
interfaces work!) with CF interface on-board
2) 8301 on Aurora gets a special 'riser' socket implementing a reliable
battery powered clock (similar to that part of QIMI).
3) Has more RAM (16M) and knows how to do memory shadowing.
4) Reasonably compatible with 'older style' peripherals.

The problem is still the price - about half GF.

Another possibility that I investigated, and actually talked about some at
the US meeting, was building a portaQL around the 68SZ328 DragonBall CPU,
known as the heart of newer 68k based Palm PDAa. This is the highest
frequency 68000 core in existance (as opposed to 68020+ cores). It works at
66MHz, and is fully 68k compatible, has a built-in SDRAM controller, serial
port, and many other interesting extras (even USB). It would even be
possible to implement monitor or LCD output in something like VGA
resolution at 256 or 65k colors, essentially for free (It's all on the chip
already). Problems:
1) Getting the chips (cheap in VERY large quantity but onobtainable in
small quantity)
2) Needs to be professionally soldered (expensive equipment) because it
uses a micro-BGA case which is tiny but has several hundred pins UNDER it
(the solder is between the case and the PCB)
3) Maximum 16M RAM - peripherals take away from this. This is, after all a
68000 (effectively 24 bit addresses are generated)
4) Comparatively slow - by my estimates about 1/4 68060 at same clock, if
that. Granted, still faster than SGC...
5) Relatively easy to make assuming you can figure out what exact
configuration of the chip you need - and wade through 600 or so pages of
documentation to find how to make it happen.
There would really only be two ways to build a system like this: as a
replacement for the insides of the old black box, or as a sort of large PDA
(buy smaller size TFT LCD and bolt the 'computer' to the back of it, add
PSU and keyboard, and wallah (as they would say in 'Chicken Run'). Common
denominator is: little or no space for expansion - the chip would pretty
much be maxed out. Fortunately, there is a CF/IDE interface on board,
IIRC... Price: as a replacement for the insides of the old box, about GF
(assuming you can get the 68SZ328 for a reasonable price).

Nasta

PS I should probably mention that a number of brand new Auroras will hit
the QL market soon. Price will be low as this is the last batch ever to be
made. Aurora display drivers for SMSQ/E was also a topic of discussion at
the US show... no definite answers, though.

Reply via email to