Sunday, April 23, 2017

AT&T 2300/2350 Video Information Terminal - Part 1

As we said before, I am in an active search for CAPTAIN or NAPLPS terminals using the Yamaha V9938. Why I am in this search? Because this little excerpt from  "V9938 MSX-VIDEO Technical Data Book":
"CAPTAIN terminals and NAPLPS terminals using the V9938 have already been developed. We hope that the V9938 will be a standard video processing device on a worldwide basis."
The "V9938 MSX-VIDEO Technical Data Book" is from August of 1985 and the V9938's copyright is from 1984. That give a narrow time window for these supposed terminals be released. But the text says "developed" not "released" which opens the possibility that these terminals were released in later dates. Which doesn't helps a lot in my search.

But then I found something in Kazuhiko Nishi's resumé, talking about the V9938:

Kazuhiko Nishi's resumé entry about V9938

The Japanese text:
MSX2に使用するためにスプライトとグラフィックスが表示可能なプロセッサーをYAMAHAと共同で開発した。このLSI V9938は、500万本以上生産され米AT&T社情報端末にも採用された。
When translated by Google to English:
Video sprite processor development
In collaboration with YAMAHA, we developed a processor capable of displaying sprites and graphics for use on MSX 2. This LSI V9938 was produced in more than 5 million units and was also adopted for AT & T Inc. information terminals.
Gives a very good clue to follow. AT&T was one of the creators of NAPLPS protocol and also sold NAPLPS terminals, maybe now we can find the mythical terminals with V9938 inside. So, let's leave the Japan terminals for a while and go to check the devices from USA.


The first Videotex systems came from Europe, in the seventies, the best graphics they can do is to use some semi-graphical characters to assemble pictures in the screen. The semi-graphical characters are OK for that time, as we can see in ZX-81 and TRS-80 Model I (and remember that the Viewdata is older than both, first prototype is from 1974).

A research group from Communications Research Centre in Canada followed other path. Given their background building CAD programs and plotting devices, they doesn't wanted to tie the graphical capabilities of their videotex system to any resolution available at that time. They created a language to describe the graphics as dots, lines, circles, arcs and polygons. This videotex system is called Telidon and the first public terminal is from 1978.

Why all this story? Well, the Telidon was the basis for the NAPLPS. We already saw that the videotex was the technological buzzword of the late seventies/early eighties, but the AT&T had not a system to offer. To mind this gap they choose to join the Telidon project and added some new commands and features.

That extended videotex system is the NAPLPS, acronym for North American Presentation Level Protocol Syntax, announced by CRC, Bell Canada and AT&T in 1983. With the tecnology itself consolided, AT&T then joined Knight-Ridder Newspapers and released the Viewtron, initially at Southern Florida.

Some NAPLPS screens from an InfoWorld's article about Viewtron launch
(InfoWorld - 1984/04/09)

In this joint-venture, the Knight-Ridder was responsible to provide the contents (news, articles, etc) and AT&T to provide the infra-structure; including the NAPLPS decoders. The decoder provided was the AT&T Sceptre:

The Sceptre Terminal and its wireless keyboard

You can see the services provided by Viewtron and the AT&T Sceptre at wild in this video:

Although the Viewtron services looks very complete (online shopping in 1980's!!!) and the Sceptre was an impressive machine, in 1986 the Viewtron was finished. Not enough consumers that were willing to pay for the terminal and the service fees. Being fair with Viewtron, the same happened with most Videotex services.

Back to this blog subject, MSX archeology, the AT&T Sceptre can't be the NAPLPS terminal with V9938. The time window doesn't match, it was released in 1983. And in Wikipedia's Sceptre entry it says the video controller is the MC6845.

Luckly to us, the Sceptre wasn't the only one NAPLPS terminal released by AT&T. In my search I found at least other three for personal use: AT&T 1300, AT&T 2300 and AT&T 2350. And, as you can deduce from this post's title, I bought one AT&T 2300/2350 Video Information Terminal. As you can see in the picture below, It looks promising...

Finally, I found a NAPLPS terminal that uses the Yamaha V9938!!

The AT&T 2300/2350 was released in 1987, so it can be one of those NAPLPS terminals "developed" before Aug-1985, we know that sometimes the time between project and release can be very long. But I don't believe it's the case.

Handwritten serial number

Probably older models of NAPLPS terminals at AT&T were the ones cited in the "V9938 MSX-VIDEO Technical Data Book", so more search is needed. But, at least, this is a proof that V9938 was used by some dedicated NAPLPS terminals.

Next post we'll talk a bit more about this equipment.

Friday, April 21, 2017

V9938 and Videotex

Search for hidden MSX devices is like follow a breadcrumb trail composed of old magazines, small pictures in auction sites, ancient web pages, rumours in forums, etc. And the trail sometimes deceives us.

Even the search for MSX Video Titlers, which were easier to look for and their existance was known by the MSX community, gave some bad surprises.

To make the search process a bit worst,  the definition of a MSX based system is not the same between different people, so while someone believes that a machine with a V9938 and a Z80 is a MSX based appliance, the definition in use by this blog states that the candidate machine need run an unmodified (or slightly modified) MSX BIOS. Then we hit the definition of "slightly modified", and so on.

One of the bread crumbs that brought many failures is this small excerpt from "V9938 MSX-VIDEO Technical Data Book":
"CAPTAIN terminals and NAPLPS terminals using the V9938 have already been developed. We hope that the V9938 will be a standard video processing device on a worldwide basis."

You can know more about CAPTAIN here, and about two CAPTAIN terminals in these other posts. If you are lazy and need those informations in condensed form:

  1. Videotext was some kind of "Internet";
  2. CAPTAIN was the Videotext system in Japan, created by NTT;
  3. There is at least one CAPTAIN terminal that is a MSX computer and;
  4. At least another one that isn't.

Coming back to V9938 technical document, NTT is the only company which was thanked nominally by ASCII:

 "Finally, we would like to express our deep gratitude to the people at
NTT as well as the other related manufacturers for their valuable
opinions which contributed to the development of the V9938."

What interest NTT would have in a Video Processor development?

We know that a MSX1 can be a videotex terminal, as they actually were used by Telesp, one of the state owned Brazilian telecommunications company (which now is part of a Spanish group, Telefonica). The Telesp's videotex, called "Videotexto", was compatible with the videotex system used by France and many other countries, the Antiope.

Antiope is a system defined in the seventies, so it's graphical capabilities are from this era: 40 characters per row and 24 rows. Each character can have a background and a foreground color from a eight colors palette (but most terminals are B&W). Using a set of semi-graphical characters, you can have a screen resolution of 80 x 72 "pixels". Some later implementations can redefine the characters to emulate a higher resolution.

While the TMS9128, another son of the seventies, and his cousins are compatible with the Antiope specifications for Videotex, they aren't capable of all the requirements for advanced high resolution graphics like the ones needed by NAPLPS and NTT CAPTAIN.

In the article "Video Display Processors Draw the Line on Expensive Graphics Systems" in the 1984's "Mini/Micro Northeast Conference", Ron Peterson shows the main limitations of VDP from Texas:

"The TMS9128 has 3 architectural constraints which limits its compliance with NAPLPS:

1) A resolution of 256 x 192 pixels instead of the minimum NAPLPS SRM resolution of 256 x 200 pixels.
2) The TMS9128 does not have a full color bit map. Therefore, problems occur when there are more than two colors in a 1 x 8 pixel block on the display
3) The TMS9128 has only 15 colors; the NAPLPS SRM requires the display to be able to handle 16 colors from a palette of 512."

NTT CAPTAIN System had similar minimal requirements for their terminals, as we can see in this table from "Videotex Terminal Based On Personal Computer":

Pattern Terminal 204V x 248H Both characters and figures are transmitted as "pattern". (photographic transmission only)
2 Hybrid Terminal 204V x 248H Characters,marks,and mosaics are displayed at high speed using code transmission. Figures are displayed as "pattern".
3 High Resolution Hybrid Terminal 204V x 496H
408V x 496H
High resolution version of the Rank 2 terminal
4 Command Terminal 204V x 248H NAPLPS like geometric figure display is possible in addition to the Rank 2 terminal function

The V9938 overcomes all three "architectural constraints" of Texas VDPs:

  1. With 212 pixels of vertical resolution, the V9938 handles the CAPTAIN requirements (204 pixels) and still have enough space on screen to have a text row with status or navigation information. As NAPLPS SRM needs a lower vertical resolution than the CAPTAIN, it's also supported by the V9938.
  2. The V9938 have bitmap modes. Many. With different resolutions and amount of colors.
  3. By last, talking about the amount of colors, one of the V9938's bitmap modes supports 256 simultaneous colors and others have 16 colors from a 512 palette, exactly the same characteristics specified by NAPLPS (CAPTAIN asks for 16 simultaneous colors, but AFAIK it's from a 4096 colors palette).

The "Thanks to NTT" and those technical specifications are strong evidence that the V9938 was *created* to be used in videotex terminals. As a bonus, it can be used in Teletext terminals;  the teletext systems usually have the same technical specifications for screen resolutions and colors of their videotex equivalents (e.g.: NABTS is the Teletext version of NAPLPS).

That made me search and buy many CAPTAIN devices and communication terminals. Those buys were the worst spent money in my search for hidden MSXs. I got many CAPTAIN terminals but the only one in this list that have the V9938 video processor is the NTT Captain Multistation, and this one is not a hidden MSX! Some units even have the MSX logo!

Some CAPTAIN and Communications Terminals

The Zeal E/PASO didn't have a V9938, but a V9958, the same video processor used by MSX2+ and MSX TurboR. But, as we show in a previous post, even while the E/PASO shares the Z80-compatible CPU, the S3527 MSX-SYSTEM and the V9958 video processor with MSX computers, it didn't have the MSX architecture and can't boot the MSX-BIOS without hardware and/or software modifications.

And I am still in the search of the CAPTAIN and NAPLPS terminals mentioned in "V9938 MSX-VIDEO Technical Data Book".

Sunday, March 26, 2017

Zeal E/PASO Communication Tool - part 3

This post brings a sad news: the Zeal E/PASO isn't a MSX machine. I still didn't have the dumps of all ROMs. But, as RetroTechie pointed out in MRC's forum, there are some important connections missing between the Hitachi HD64180 (the main processor), the memories and the Yamaha S3527 (the MSX-SYSTEM). Ricardo Oazem, from Tecnobytes, confirmed this diagnostic.

It's the first time in this blog that a appliance with a MSX-SYSTEM isn't compatible with the MSX architecture. Probably space constraints made Zeal to use the S3527. They needed a keyboard controller and a sound generator, the S3527 have both functions in a package very smaller than a i8255 and a YM2149. The slot handling is not used, the main memory and the embedded software seems to be connected directly to the HD64180.

Even in a machine without the slot handling support, it's possible to run a modified MSX-BIOS, doing a direct mapping the ROM and RAM to the correct addresses. This modified MSX-BIOS can even be used to abstract different port addresses or some hardware incompatibilities. There are "MSX adapters" for other computers with a similar hardware which follows this path to play some MSX games (and probably this works to play almost all 8-32KB games that uses the MSX-BIOS to access the hardware). Maybe this can be done with the Zeal E/PASO.

We now have the contents of these ROM and EPROM.

With the help of many friends at MSX Jau 2016 meeting, now I have two of the ROMs of Zeal E/PASO dumped. The Hitachi HN62321 looks like a Kanji ROM (as we guessed in previous post) and the other labelled as "E/P ver1.00" I did not really know what it is. At first my guess was that the data in there had their bits mixed to help the board circuit routing, but there are some text strings inside the EPROM:

**** P'[*]C NIF'[--->]SVC'[--->]%0'[--->]%1'
******V*A*N*P*C*N*E*C*0*2'[ID: ]%0'[WORD: ]%1'
****[): ]mix'[Name? ]%0'[word: ]%1'
******@**'**[*]c ascii'[ID:]%0'[word:]%1'
******@**'**[*]a eye'[CONNECTED]'[CODE:]'[ID:]%0'[WORD]%1'
****[CODE ->]%0'[WORD ->]%1'
******@**'**[*]a mn'[id :]%0'[word:]%1'

Those text strings looks a lot like PPP's "conversation" strings between a client and a server.  Probably this EPROM is used by the modem controller. If you want to read those ROMs yourself and make your own guesses, you can download both files in this tarball.

With those two EPROMs we still didn't have any code for HD64180/Z80. There a lot of other ROM ICs to read, I hope to have those files soon. Then we'll see if the E/PASO software was derived from MSX-BIOS code and maybe have some clues to adapt a MSX-BIOS to run in this machine. Yes, I still didn't give up the idea of using Zeal E/PASO as a MSX.

And now, the previous posts were updated with the "Not MSX" tag.

Saturday, March 25, 2017


Hi guys, this blog, while in a very slow pacing is not dead yet and its death is not in my plans. But I know that a four months silence isn't a good health sign. I still have a lot of appliances to show, ROM images to provide and information to share.

Why this lack of new posts is happening?

Mostly because I reached my technical skills limits. There are some posts that are waiting for ROM dumps. I have a ROM/EPROM reader, but I am very bad working with a solder iron, so I can only dump the ICs that are placed in sockets. There are at least four posts in this situation.

Other posts are waiting for some additional equipment to arrive from overseas. There are two posts waiting for things to come.

And, last but not least, there is the lack of time/priority to keep all projects running. The activities to write a post includes a lot of research, time to disassemble the equipments, identify the ICs, take pictures, search for datasheets, sometimes read the ROM dump, etc. There are an unknown number of posts waiting to exist, the time shortage prevents they to be written.

I believe that, soon, we'll have some news.