From: TW on
> The HP49G was the first model to drop the Saturn CPU (a custom design

Just a quick edit here - it still ran a 4mhz saturn. It wasn't until
the 49g+ when it became emulated.

> from HP that had been in use for decades) and use an ARM  to emulate it..
> The interface was revamped by incorporating the MetaKernel. This
> wizardry was achieved by employing Cyrille de Brebisson and Jean-Yves

There were a few more of those 'french wizards' besides those two. By
working 20 hour days, living at the HP facility, and the fact that one
could simply say "I need a choose box engine" and that took the place
of a 300 page specification document, they were able to get it done.
Nothing short of a miracle in many ways. . .

TW
From: Bruce Horrocks on
On 18/03/2010 03:29, TW wrote:
>> The HP49G was the first model to drop the Saturn CPU (a custom design
>
> Just a quick edit here - it still ran a 4mhz saturn. It wasn't until
> the 49g+ when it became emulated.
>
Gah!, I knew something significant would slip through.

>> from HP that had been in use for decades) and use an ARM to emulate it.
>> The interface was revamped by incorporating the MetaKernel. This
>> wizardry was achieved by employing Cyrille de Brebisson and Jean-Yves
>
> There were a few more of those 'french wizards' besides those two. By
> working 20 hour days, living at the HP facility, and the fact that one
> could simply say "I need a choose box engine" and that took the place
> of a 300 page specification document, they were able to get it done.
> Nothing short of a miracle in many ways. . .

There's always space in Datafile for a warts 'n' all expos�...

--
Bruce Horrocks
Surrey
England
(bruce at scorecrow dot com)
From: Han on
Fascinating history... and what about this Kinpo OS that creeps up now
and then. Are they at all part of the problem? Or were there hands
tied by the folks from HP who wanted to "dumb down" the calculator. I
seriously believe that the ALG mode is a complete waste of ROM,
especially when in the quick start guide they mention just how much
better RPN is.

On Mar 17, 9:19 pm, Bruce Horrocks <07....(a)scorecrow.com> wrote:
> On 17/03/2010 02:48, Han wrote:
>
> > Can anyone shed some light on the history of the HP 49 ->  HP 49G+ ->
> > HP 50G series in terms of hardware changes as well as operating
> > systems?
>
> If only I could write as well as Verity Stob.[1] Oh well, here goes...
>
> In the beginning there was the HP28C. It 'only' had a 4 line display and
> 2KB of RAM but it introduced RPL, nearly as we know and love today. It
> was a marvel: it could do symbolic algebra which was unique for the
> time. It could symbolically integrate and differentiate. And it was
> really fast: it would really quickly tell you that it had run out of
> memory. Still it was quite a big hit, setting the standard for the
> models to come.
>
> Next came its big brother the HP-28S. (Funny how, in the calculator
> world, big brothers are younger than little brothers.) It had 32Kbytes
> of RAM. And in those days, 32Kbytes was so enormously big that it needed
> to be *managed*. So out went the single line of variables in the 28C and
> in came 'directories'. Ta da!  Yes, you could organise variables into
> directories and subdirectories. Directories were really useful - when
> you lost the entire memory of the calculator trying to change batteries,
> you could now exclaim 'Gosh darn, I've just lost five entire directory's
> worth of work!' Which sounds as though you've been so much more productive.
>
> The 28S was very popular. People loved its clamshell design with two
> keyboards, meaning that the keys could be clearly labelled without
> clutter. They didn't love the battery door though, which broke quite
> often, even if you weren't a clumsy oaf with the finesse of an elephant
> wearing boxing gloves. So along came the HP48SX.
>
> SX stood for Scientific Expandable, but it was all too easy to think
> 'Expendable' instead. Except that it wasn't really expendable since
> things still cost 'too much' in those days. However, it solved the
> 'losing data when changing batteries' problem by having a serial port
> which allowed it to be backed-up to a PC. Hallelujah! It had 32KBytes of
> RAM but you could add memory cards of either 32 or 128Kbytes. One of the
> cards could be merged with main memory to give unheard of amounts of
> working RAM. (Actually I use the phrase 'unheard of' metaphorically - if
> someone actually had one of these cards then, boy, did you hear about it.)
>
> The other major change was to drop the clamshell case. Apparently
> 'marketing' (those people in the back office who don't get out much
> except when they've run out of coke) did some 'research' and discovered
> that 'students' expect a calculator to look like a rectangular block
> with a big screen at the top (as big as possible, irrespective of the
> actual resolution) and lots and lots of buttons underneath. Preferably
> with about 5 or 6 labels on each. Any calculator designer who produced
> designs that did not follow this model exactly was sent off to the
> Siberian salt mines. (As an aside, I have some wonderful prototype case
> designs carved in solid rock salt.) The HP48SX was as near perfection as
> there has ever been in the RPL line. But perfection can be improved upon
> by applying the 'less is more' philosophy.
>
> The HP48S was the result. It had identical software to the HP48SX but
> less hardware - specifically only 32KBytes of RAM and no expansion
> ports, so you couldn't add memory cards. It retained the serial port but
> so slavish was the adherence to the 'less is more' philosophy that even
> the model name had a letter removed.
>
> The 48S was actually a remarkable product - remarkable because it sold
> in large numbers - much more than expected. But even more remarkably, it
> caused sales of the HP48SX to increase as well. So much so that HP felt
> they could afford to introduce an updated version, the HP48GX.
>
> The GX was even more student oriented (aka in the absence of critical
> thought, we'll try and be as TI like as possible). So it introduced
> drop-down menus for everything. These were slow. Really slow. Really,
> really slow. But they could at least be turned off. So that was good.
> Best thought of as 'shop demo mode' i.e. looks good and really easy to
> use when dad is shown it by the bloke in the store. (Dad being the buyer
> of HP calculators and so has to 'approve' the purchase even though he
> doesn't really have a clue and it is the son who's going to college and
> will actually use it.) Also those menus make it really easy for the
> bloke in the store to look like he knows what he is doing. (At this
> point I feel I should diverge into an honourable mention for Richard
> Nelson who was one of those blokes in a store. In fact he was the bloke
> in his *own* store. And he did know what he was doing, even if he didn't
> look like it.)
>
> The GX wasn't well received. The colour scheme on the keyboard was
> changed from nice clear orange and something (I forget and I don't have
> mine with me at the moment) on the SX to green and purple on the GX.
> These were hard to tell apart under anything less than bright lighting
> and, for those with the wrong type of colour blindness, they were all
> but impossible to tell apart under any lighting.
>
> Naturally the GX needed some more features so list processing was added
> to RPL. An attempt was made to allow more or less any command to work on
> either its traditional single arguments or lists of arguments. This
> mostly worked as expected but there were some surprises. Also some
> explicit list processing commands were added. These turned out to have
> been rather poorly coded and faster, more robust community derived
> versions in userRPL were quickly developed.
>
> The GX came with 128KBytes of main memory and is expandable with two
> memory card slots like the SX. There were some electrical changes (IIRC)
> which enable larger cards to be used. At the expense of compatibility
> with the SX ROM cards. But that wasn't too big a problem since SX ROM
> cards were fantastically expensive and only corporate types could afford
> them. The GX was soon followed by a cheaper, HP48G model released with
> only 32KB of main memory and no expansion card slots. By now (1998),
> 32KB was looking rather miserly and HP quickly came out with the HP48G+
> with 128KB instead. (This actually makes a very nice calculator and well
> worth getting if you find a cheap one on you know where.)
>
> Here endeth the 'old testament' namely the old certainties of one true
> religion - RPN - and a large enter key. What followed can only be
> described as 'new testament' because ever increasing amounts of
> forgiveness were demanded from HP fans.
>
> The HP49G was the first model to drop the Saturn CPU (a custom design
> from HP that had been in use for decades) and use an ARM  to emulate it..
> The interface was revamped by incorporating the MetaKernel. This
> wizardry was achieved by employing Cyrille de Brebisson and Jean-Yves
> Avenard, two French students who had developed and were selling an
> alternate interface for the 48GX. HP then nullified this stroke of
> genius by saddling them with the worst rubber-key keyboard design since
> the Sinclair Spectrum. They were also forced by the 'marketing' people
> to implement an algebraic mode and, rubbing salt into the wound, it was
> the default mode. This was, apparently, so that 'students' would be more
> 'comfortable' with it. All it probably did was contribute to an increase
> in bullying - previously the one HP user in a class of TI sheep could at
> least claim he was being different and thinking for himself; now it just
> looked like his dad had gone to store *on his own* and bought the wrong
> one because it looked vaguely like a TI.
>
> What should have been a great machine was let down by an awful keyboard
> and it is now fondly remembered as 'the frozen Hamster-butt blue one'.
> It did also introduce flash ROM and so the ability to eradicate bugs by
> loading a newer ROM into the machine.
>
> Next was the HP49G+. This was so good that eBay prices for HP41s in
> virtually any condition went through the roof. The 49G+ replaced an
> awful rubber keyboard with an awful plastic keyboard. So not really an
> advance. HP were really, really struggling with keyboards at this time
> as the other models in their range at the time were just as bad or
> worse. No doubt the true story will emerge at some point in the future
> when a Chinese multi-gazillionaire writes his memoirs explaining how he
> got rich by screwing HP.
>
> Finally, (skipping some oddities like the HP48GII) we come to the HP50G
> which is still on sale today, has a quite passable keyboard and which
> you all know and love, so I'll shut up. If they update it with the
> keyboard technology used on the new 30B then we'll almost be back to
> where we were in 1982. ;-)
>
> Here endeth the history of the HP48/49/50 series.
>
> [1]http://www.theregister.co.uk/odds/stob/
> --
> Bruce Horrocks
> Surrey
> England
> (bruce at scorecrow dot com)

From: TW on
> Fascinating history... and what about this Kinpo OS that creeps up now
> and then. Are they at all part of the problem?

Somewhat. The code they made is the low level OS that runs the
saturnator (the arm saturn emulator written by Hydrix). A lot of it
was not very good. For example, the kinpo C fseek command is a POS.

> especially when in the quick start guide they mention just how much
> better RPN is.

That's because I wrote it. :-)

TW