Saitek risc 2500 v1.02 v1.03 v1.04 differences ?

[TurboKing]

Hello all

About RISC 2500,
Do you know what the differences are between versions 1.02 , 1.03 and 1.04 ?
Any strength improvement from 1.02 to 1.04 ?
I 've got the 1.03 so just wondering if it's worth getting a new R2500 but with the lattest version
thank you for your help

[Larry]

Hello all

About RISC 2500,
Do you know what the differences are between versions 1.02 , 1.03 and 1.04 ?
Any strength improvement from 1.02 to 1.04 ?
I 've got the 1.03 so just wondering if it's worth getting a new R2500 but with the lattest version
thank you for your help

Well we discussed this briefly a short time ago. The consensus is that
the different versions are the result of minor changes in the source code,
which will not result in a strength change.
However, I found something out this week. There is a RISC2500 on ebay
at the moment:
http://cgi.ebay.com.au/ws/eBayISAPI.dll ... :MEBIDX:IT
I asked the seller for the information on the screen when you press the
right/left arrow keys. This is what it says:
128K ROM 128k RAM V1.04 21 April 92
You see the date? Now I've got a V1.03 that is dated later than this V1.04
by about 6 months. This tells us that the higher numbered versions were
not necessarily produced later. The scrappy pieces of historical information
on chess computers left for us decades later is frustrating.
I happen to have two of these things, a V 1.03 and a V 1.04. However
I can't set up positions for comparison because one is 128kb and the other
is 2mb. I'm not willing to start pulling RAM chips out.
all the best
Larry
PS As an afterthought, and this is not to do with your question, but I
noticed in another thread about a year back, possibly on another forum,
a poster said he had a RISC2500 from late in the production run. You
may be aware that the later Mephisto Montreux was a virtual continuation
of the RISC2500, and had the CPU mounted on the other side of the
mainboard. His RISC2500 had it's CPU also mounted on the other side,
ie, the 'Montreux' side. What can we deduce from this? Well perhaps the
Montreux is nothing more than a RISC2500 with yet different attitude
settings, ie, a V1.05 ?
all the best
Larry

[Larry]

Hello all

About RISC 2500,
Do you know what the differences are between versions 1.02 , 1.03 and 1.04 ?
Any strength improvement from 1.02 to 1.04 ?
I 've got the 1.03 so just wondering if it's worth getting a new R2500 but with the lattest version
thank you for your help

Well we discussed this briefly a short time ago. The consensus is that
the different versions are the result of minor changes in the source code,
which will not result in a strength change.
However, I found something out this week. There is a RISC2500 on ebay
at the moment:
http://cgi.ebay.com.au/ws/eBayISAPI.dll ... :MEBIDX:IT
I asked the seller for the information on the screen when you press the
right/left arrow keys. This is what it says:
128K ROM 128k RAM V1.04 21 April 92
You see the date? Now I've got a V1.03 that is dated later than this V1.04
by about 6 months. This tells us that the higher numbered versions were
not necessarily produced later. The scrappy pieces of historical information
on chess computers left for us decades later is frustrating.
I happen to have two of these things, a V 1.03 and a V 1.04. However
I can't set up positions for comparison because one is 128kb and the other
is 2mb. I'm not willing to start pulling RAM chips out.
all the best
Larry
PS As an afterthought, and this is not to do with your question, but I
noticed in another thread about a year back, possibly on another forum,
a poster said he had a RISC2500 from late in the production run. You
may be aware that the later Mephisto Montreux was a virtual continuation
of the RISC2500, and had the CPU mounted on the other side of the
mainboard. His RISC2500 had it's CPU also mounted on the other side,
ie, the 'Montreux' side. What can we deduce from this? Well perhaps the
Montreux is nothing more than a RISC2500 with yet different attitude
settings, ie, a V1.05 ?
all the best
Larry

I'd like to make a little addition to this thread if I may. Call me eccentric,
but I now own no fewer than three risc2500's. I was'nt looking to get
another one, but a few weeks back I casually put the minimum bid on one
and went to bed. The next morning, lo and behold, I had won the thing.
Still reading, huh?... ok, well I will keep writing! Out of curiosity I took the
back off this one, and was surprised to discover that it's CPU is also on
the other side of the mainboard, ie, the supposed 'montreux' side. The
surprising thing is that this machine is dated 21 Oct 1992, ie, early in the
production run and well before the Montreux was ever even in the
planning stage. This tends to tell us that Saitek simply had two suppliers of
mainboards, it was not a case of one mainboard for risc2500 and one for
the montreux.
Both machines were made in China (is'nt everything?), but would'nt you
have thought the managements of early chess computers would have kept
a log of developments for posterity?
all the best...
Larry

[Kostea]

Larry,

The date "21 Oct 1992" that you are referring to as the manufacturing date is actually the date when a particular software version was released.

Saitek RISC 2500
version 1.03 -> 14 Oct 1992
version 1.04 -> 21 Oct 1992

Mephisto Montreux
version 1.00 -> 10 Dec 1994

These are the only versions I have seen, there are probably others for which I don't have any data.

For example, a RISC 2500 could have been manufactured in 1995 but it will show "21 Oct 1992" because of the software version 1.04 which was used in the EPROM.
It is very time consuming for a company to change and write in the EPROM memory the dates when every single chess computer was made. It was just not worth it. They did not think that collectors would have cared about these details 15 years later.

Larry, you definitely have a Mephisto Montreux CPU installed in a Saitek RISC 2500 case and running RISC 2500 software.

Originally, RISC 2500 used an ARM2/3 CPU made by VLSI TECHNOLOGY INC. The reason I'm saying ARM 2/3 is because I'm not sure if it was an ARM2 or an ARM3. I could not find its datasheet on the Internet. I live in USA but was born in Europe, and that's where I left all my catalogs and valuable electronic tools I collected for decades. That ARM 2/3 CPU was in a PLCC84 plastic package, so the printed circuit board was designed for that CPU's pinout. Later on by the time Mephisto Montreux came on the market, the technology advanced, and Saitek used an ARM6 CPU made by MITEL/ZARLINK. The problem was that the MITEL/ZARLINK chip came in a different plastic package and pinout, and a different number of pins than the original VLSI CPU.

Saitek reused the RISC2500 board in the Montreux but now they were in a dilemma. What to do: 1) redesign the printed circuit board to change CPU pinout from ARM 2/3 to ARM6, or 2) solder the new ARM6 CPU on a pin converter/adapter small board to bring the pinout in the right sequence? They used the second option, and that's why the CPU ended up on the bottom side of the circuit board in a Montreux. The whole CPU assembly was just too tall to have room to be installed on the "correct" upper side of the board, while they had enough clearance under the PCB.

By the way, that mounting of CPU on the wrong side of PCB makes a Montreux much harder to repair. There is a lot of stuff to desolder in order to replace the CPU. On the other hand, you would only need a cheap PLCC chip extractor for popping out the CPU of a RISC2500 and no need to use desoldering tools at all.

[Larry]

Hi Kostea, You know I had a suspicion that the eprom date and the date of
manufacture were not necessarily the same, especially when I got another
machine with the exact date on it. I have never seen a risc2500 V.02 but
if I do see one I will bid on it. Same for a montreux.
Regarding the different mainboards, there is also a different voltage for
the montreux to the risc2500 mainboard. I came unstuck in the last few days
over this. When I took delivery of my new risc2500 I used the voltmeter to
compare the voltages of the adaptors. There was a 0.3 voltage output
difference between them, which I assumed was within the manufacturer's
tolerance, and that they were interchangeable. They are not!...even though
they are both labelled HGN5050. I put the lower voltage adaptor on my latest
risc2500 with the montreux mainboard and it kept freezing with an error message. I
referred the matter to the seller, who actually frequents this forum, and he
assured me the machine was fine when he sold it. He suggested a power supply
problem, which turned out to be the case. On putting his adaptor on his machine, the problem vanished. Just 0.3 volt made the difference...
I guess Saitek never considered that collectors later on might quite well
own more than one risc2500 or more than one montreux. I have now labelled
my adaptors and my machines with the correct voltage, since the
adaptor model number can't be trusted.
Thanks Kostea, and thanks to Christian (the seller)
Larry

[Larry]

Kostea, I'm wondering if I can have a bit more of your wisdom. You seem
to have a level of technical knowledge regarding dedicated chess computers
beyond anyone else we have had on this forum.
Of course my question is in keeping with this thread and pertains to the
differences between risc2500 and montreux. Note that, at least according
to chesscomputer.info, the montreux has 256kb of ROM while the risc2500
only has 128kb of ROM. This may account for the vastly bigger opening
book of the montreux as compared with the risc2500, still according to
chesscomputers.info.
Since we have settled that
my risc2500 has a montreux mainboard and processor in it, does it follow
that it also has the extra ROM memory onboard, which is included in all
montreux's?... and therefore a montreux opening book library? Or is it the
case that the ROM is included in the EPROM, which means the opening book
is etched into the EPROM?
I hope this has'nt caused confusion.
have a nice day, as I am...
Larry
PS Kostea, while I've got you, do you happen to know what would cause a
pencil thin horizontal line to go blank on a GAL display? And if there is a cure?
I'm currently having this problem. It's happening on the upper row of text.
thanks again...

[Larry]

PS Kostea, while I've got you, do you happen to know what would cause a
pencil thin horizontal line to go blank on a GAL display? And if there is a cure?
I'm currently having this problem. It's happening on the upper row of text.
thanks again...

Sorry, that should read LCD display, not GAL display.

[Kostea]

Larry,

First of all let me clarify what an ARM CPU is. It actually is a type of CPU architecture invented by a company named ARM. They produced ARM "cores", and they sold and licensed the cores to other companies which assembled the ARM core along with other special circuitry, packaged it, and then sold the final product as CPUs on the market. The ARM company did not produce CPUs itself, and never sold CPUs on the market.

An example easier to understand. The ARM company produced "brains". Then another company called BODYWORKS (fictitious name) purchased the brain from ARM, added a head, neck, trunk, some limbs, and packaged them in the shape of a human body and sold it on the market. (that's how man was created, don't believe in either religions or Darwinian evolution)
Then ARM designed faster and more inteligent versions of that brain, and so on.

Saitek RISC 2500 was released in 1992. It was designed to use an ARM2 or ARM3 core because that was the version of ARM core which was available at the time. The ARM6 core came out on the market in 1992, but it was too late to be incorporated in the great majority of RISC 2500 machines.

Made up example: Saitek commisioned TASC Holland to design a chess computer. They spent most part of 1990 and 1991 to design it. Saitek buys bulk 10,000 ARM2/3 CPUs and the rest of components and spends the first half of 1992 to manufacture the final product as RISC2500 and sold it on the market at the end of 1992. They noticed that in the meantime a new ARM6 CPU was released but they decided to use up all remaining supplies of ARM2/3. Then they buy 1,000 ARM6 CPUs to build the Mephisto Montreux in 1995. At the same time they still make the RISC2500 but ran out of the original ARM2/3 CPUs and use some of the newly purchased ARM6 chips.

In conclusion, most RISC2500 have ARM2/3 type of CPUs and some from the last batch have an ARM6 CPU, while Montreux only had ARM6 CPUs.

Larry, the last RISC2500 you got has the most recent ARM6 CPU, while the earlier ones you bought have the ARM2/3 CPUs. Now what makes a RISC2500 a RISC2500, and what makes a Montreux a Montreux is not the CPU but the software from the EPROM chip. One way to tell what software you have is to just turn on the chess computer and watch the LCD display. Another way is to read the inscriptions on the EPROM chip. TMS27C010 is a 128kb EPROM, and TMS27C020 is a 256kb. RISC 2500 software is 128kb, Montreux is 256kb.

As a curiosity, the software for RISC2500 is 128kb and used the full 128kb EPROM, while the software for Montreux is 150kb and the remaining 106kb of the 256kb EPROM is blank/empty. So the Montreux software is only 22kb bigger than RISC2500 which most likely accounts for the larger opening book.

[Cubeman]

Thanks Kostea, you have a good way of explaining these things and it is nice to have you on this forum.

[Kostea]

Larry,

If you ever took apart a LCD display, you would have seen that the display has one or two conductive rubber "connectors". They are of a rectangular parallelepiped shape (think of a match from a matchbox) and are almost as long as the length of the LCD display.

Clean those elastic rubber connectors with isopropyl alcohol. Also clean the golden contacts on the printed circuit board, and the areas of the display in contact with the rubber connectors. Sometimes the golden contacts on the PCB are slightly corroded, you can only see that faint oxidation (or dirt) through a jeweler's magnifying glass and not with the naked eyes. I use a small flat precision screwdriver to "scratch"/remove the oxidation, without scoring the contacts. Use the slightest pressure, be gentle. The flat head of screwdriver has to be just a littler larger than the contact pad. If the flat head is smaller than the contact, you can score the contact, if it's much larger that you can touch 2 or more contact pads at the same time, do not use it because at that flat head size the screwdriver is too heavy to be precisely controlled by hand and can easily damage the contacts.

Some people use the finest/smallest grain sandpaper and lightly sand the PCB's contacts. Just make sure you clean the PCB contacts after sanding because you can leave a fine layer of conductive metal sand which can create a shortcircuit between the golden contacts. I myself do not use the sanding method.


Sometimes you just have to tighten a little bit more the tiny screws of some LCD displays, and they compress the rubber connectors to make a better contact again.

From what I wrote above you might think that I perform heart surgery on those contacts. I don't, it's easy stuff, but I like to be as delicate and careful as a man's hand can possible be.

If it still doesn't solve your problem, you just have to use a bigger hammer. Just kidding

[Kostea]

Cubeman,

Thanks. I just wish I had more chess computers to scrutinize and maybe shed more light on their software and hardware, to supplement the info already posted on 'schach-computer.info' website.

[Larry]

Thanks again Kostea,

Saitek RISC 2500 was released in 1992. It was designed to use an ARM2 or ARM3 core because that was the version of ARM core which was available at the time. The ARM6 core came out on the market in 1992, but it was too late to be incorporated in the great majority of RISC 2500 machines.



In conclusion, most RISC2500 have ARM2/3 type of CPUs and some from the last batch have an ARM6 CPU, while Montreux only had ARM6 CPUs.

This being the case, you've got to wonder what the advantage of the
later ARM6 type CPU is over the earlier ARM2/3 type CPU's. Perhaps just
lower power consumption? Or cheaper to produce? They don't seem to perform any better. I have
set up a couple of endgame positions, and at least in those positions the
solving time was identical between a risc2500 V1.04 (ARM6), and a
risc2500 V1.04 (ARM2 or 3). Both machines 2mb.
regards,
Larry
PS Just got your latest post after submitting this one...we are online together...so thanks for this in depth description. Few technically inclined
people will do this.