The unreleased Commodore HHC-4's secret identity

Once upon a time (and that time was Winter CES 1983), Commodore announced what was to be their one and only handheld computer, the Commodore HHC-4. It was never released and never seen again, at least not in that form. But it turns out that not only did the HHC-4 actually exist, it also wasn't manufactured by Commodore — it was a Toshiba.

Like Superman had Clark Kent, the Commodore HHC-4 had a secret identity too: the Toshiba Pasopia Mini IHC-8000, the very first portable computer Toshiba ever made. And like Clark Kent was Superman with glasses, compare the real device to the Commodore marketing photo and you can see that it's the very same machine modulo a plastic palette swap. Of course there's more to the story than that.

Recall, as we've discussed in other articles, that 1980s handheld and pocket computers occupied something of a continuum and the terms were imprecise, though as a rule handheld computers tended to be larger and emphasize processing power over size and battery life, while pocket computers tended to be smaller and emphasize size and low power usage over capability. Thus you had computers that were "definitely pocket computers," like the Casio PB-100/Tandy PC-4 and Sharp PC-1250/Tandy PC-3, and "definitely handheld computers" like the Kyotronic 85 series (TRS-80 Model 100, NEC PC-8201A, etc.), Convergent WorkSlate and Texas Instruments CC-40, but also handheld-class computers sold as pocket computers like the Sharp PC-1500/Tandy PC-2 and Casio PB-1000C, and even a few handheld-sized computers with capabilities more typical of pocket computers like the VTech Laser 50. Small systems like these were hot in their day, not least from novelty and size appeal, and nearly every computer company was dabbling in them. The vast majority originated in Japan, primarily from Casio and Sharp, but also from other lesser-known manufacturers like Canon. Because of their unique constraints on size and battery usage, some notable exceptions notwithstanding (at least one of which we'll have a look at), pocket computers in particular often used unusual architectures seen nowhere else.

Toshiba is a contraction of Tōkyō Shibaura Denki kabushiki-kaisha (Tokyo Shibaura Electric Co., Ltd.), formed in 1939 as a merger between Shibaura Seisakusho (Shibaura Engineering Works) and Tōkyō Denki (Tokyo Electric). Both were licensees of American conglomerate General Electric who owned substantial portions of each of them at the time. Although it carried the well-known nickname for years, the unified enterprise did not formally rename itself to Toshiba Corporation until 1979. As Tokyo Shibaura Electric, Toshiba produced its first computer in 1954, the TAC, an EDSAC clone with 7,000 vacuum tubes and 3,000 diodes delivered as a prototype to the University of Tokyo. It eventually produced minis like the TOSBAC-3400, based on Kyoto University's 1961 KT-Pilot, and produced its first microprocessor, the 12-bit TLCS-12 ("Toshiba LSI Computer System"), in 1973 for Ford automotive engine control units.

The upgraded TLCS-12A became the basis of Toshiba's first microcomputer, a single-board evaluation system, and then reworked around the Intel 8080A (using the compatible Toshiba TMP9080AC at 2.048MHz) for the 1978 TLCS80A • EX-80, or EX-80 for short. At that time it was the cheapest system that could output to a TV using built-in hardware, displaying 10x26 text (not a typo) or 80x26 graphics from its standard 1K of RAM. In 1979 Toshiba developed the T-400 as a demonstration personal computer, using the upgraded EX-80BS ("BASIC System") and Toshiba's second-source version of the Intel 8085. It had a 32x24 text display, eight colours and up to 256x192 graphics, supporting up to 36K of RAM. Toshiba intended to offer it to overseas retailers as an OEM, but it got little import interest, and the company took it back to the shop to refine the design.

Thus was the nucleus of the Toshiba Pasopia line, though it quickly became more of a generic brand than a distinct technical segment and many machines ultimately sold under the label weren't closely related. The original 1981 Pasopia, sold in the United States as the Toshiba T100, was a 4MHz Z80 system offering two flavours of BASIC, 64K of RAM, three-voice sound using the Texas Instruments SN76489 DCSG and up to 640x200 graphics. A notable later option, along with the more typical disk drives, RS-232 and parallel expansions, was an optional LCD mounted on its keyboard, but we'll come back to this at the end. On the other hand, the 1982 Pasopia 16 was a completely different PC-compatible system, running Intel's HMOS 8088-2 at 6MHz (as compared to the slower original NMOS 8088) and supporting up to 256K of RAM and MS-DOS 2.3 and CP/M-86. It had custom graphics hardware with up to 16 colours and 640x500 resolution. The P16 was also sold in the United States as the T300 and in Europe as the PAP.

These two computers ended up forming two completely separate lineages that nevertheless were also badged as Pasopias. Due to poor sales outside their native country, their descendants were sold only in Japan. On the classic Pasopia side, Toshiba released the Pasopia 5 in 1983, a cost-reduced version of the O.G. Pasopia with the same features, along with the Pasopia 7 (and later the closely related Pasopia 700 in 1985), an upgraded and partially-compatible successor featuring a second TI SN76489 sound chip, three times the video RAM and more video modes. For its part, the Pasopia 16 separately evolved into the 1984 Pasopia 1600, with an 8MHz HMOS 8086-2 and up to 384K of video RAM, and the Pasopia 1600 TS100 and TS300, which had an 8MHz 80286 and up to 704K of RAM, with either dual 5.25" floppies (TS100) or one floppy and a 20MB hard disk (TS300). To confuse things further, Toshiba also sold a line of MSX machines called the Pasopia IQ, releasing the HX-10, HX-20 and HX-30 series in 1983, 1984 and 1985 respectively. Some of these machines were sold in Europe, with only the Japanese versions having Pasopia IQ branding, and the last few models supported the MSX2 standard. Toshiba finally retired the Pasopia brand in 1986 with the non-Pasopia MSX2 FS-TM1, sold in Italy.

Sandwiched in between those two major lines, however, was a much smaller one.

Toshiba developed the Pasopia Mini to get into the growing Japanese handheld and pocket line in 1982, prior to the release of the P5 and P7. This ad clearly fixes the Mini as prior to the 1983 refresh since the "PA7010/7012" (T100) and "PA7020" (P16) are prominently shown but not the later models. Unfortunately I don't have a higher-resolution version of this ad, so I can't translate it in its entirety, but Toshiba advertised it with an "8-bit CMOS" CPU, from 4K to 16K of RAM, and an odd 20K of ROM for BASIC. Typical of many contemporary pockets and handhelds, the LCD display was a single line of 24 characters and was not dot-addressable. Architecturally it had virtually nothing in common with the Pasopia/T100 or the P16 and Toshiba instead proferred it as a portable complement to their larger systems despite its complete incompatibility.

Launched as the IHC-8000 for 54,800¥ in either black or silver, the Pasopia Mini was sold alongside an optional IHP-500 docking station with a wallwart power supply and kinda-sorta-hard case for an additional 45,000¥. The docking station (officially the "mini-printer and peripheral interface") supplied a small built-in thermal printer and 1/8" cassette interface jacks, but also pin headers for connecting a full-size Centronics-style printer, RS-232 serial, and a video monitor with unknown resolution or capabilities. However, other than the tape leads, the other necessary connectors didn't come with the IHP-500 and it's not clear if they were ever actually sold. Regardless, the fact that it could be connected to such devices was prominently played up in the ad; the large red text reads "テレビとつなぐ。パソピアとつなぐ。プリンターとつなぐ。" (It connects with a TV. It connects with Pasopia. It connects with a printer.)

Meanwhile, Commodore was having a dickens of a time cracking the Japanese market through their Japanese subsidiary, headed by founder Jack Tramiel's eldest son Sam Tramiel and VP and de facto chairman Taro "Tony" Tokai. The VIC-20 was actually introduced first in Japan as the 1980 VIC-1001, a heavily localized version with a katakana character set, because Jack felt the American market was too fickle after sinking sales of the PET line. It did reasonably well due to its relatively low price point and few competitors — indeed, its release stalled other Japanese electronic companies by nearly a full year, even Toshiba — but that wasn't the case by 1982 when Commodore first started demonstrating the MAX Machine. The MAX Machine, shown here in prototype form courtesy of a West German TV piece at CES that year, was the forerunner of the Commodore 64 and used the same SID sound chip and a slightly earlier VIC-II video chip, but offered just 4K of memory and drastically fewer expansion options. On top of that, it didn't even have proper keys, just a universally maligned blister keyboard (this prototype used a membrane keyboard). Although designer Yashi Terakura intended the MAX as a game machine, the Japanese market didn't see it that way because of its price and form factor, and miseries like the 510 bytes free with its MiniBASIC cartridge made things worse.

Commodore Japan became aware of the handheld Pasopia Mini and approached Toshiba, presumably through Tokai's industry contacts. At that time Commodore had no portable systems for sale at all through any of their national subsidiaries, and while they were secretly developing the luggable SX-100, SX-64 and DX-64 versions of the Commodore 64, management concluded that a handheld system — even a thoroughly incompatible one — would complement the larger luggables nicely. Commodore owned MOS Technology (developers of the 6502 CPU) and typically preferred to develop their own hardware in-house, but Tramiel's Commodore was certainly not above outside licensing if he thought he could get a good deal. Because Toshiba practically had a shipping product already, Commodore determined they could get it swiftly to market with just minimal changes and quickly negotiated an exclusive agreement to bring it to the United States.

And thus did the Commodore HHC-4 debut using Toshiba's design, presented at Winter CES 1983 alongside the SX-64 family. This portrait is a scan I found of a Commodore marketing circular reportedly distributed at the show and is the most detailed picture of the HHC-4 available (I cleaned it up a little bit by hand in Krita since there were substantial compression artifacts in the original). This is in fact the very picture I showed you way back at the beginning to compare against the Pasopia Mini. Commodore's marketing advertised the machine as also having 4K of RAM (expandable to 16K) and 20K of ROM, just like Toshiba's.

Credit where credit is due: Scott Jones was the first person a number of years ago to suggest to me (for the Secret Weapons of Commodore site) that the Pasopia Mini was a possible match, though it took me a very long time to locate a unit I could get overseas at a reasonable price — turns out they're not very common in Japan either. As pocket/handheld computers go, on a brief glance it could easily be mistaken for the Sharp PC-1500/Tandy PC-2 and is of similar size. This one wasn't in good shape and the LCD has some iffy pixels but it basically works. It came with the IHC-500, IHC-501 A/C adapter and the kinda-sorta-hard case, though the case and the dock are in even poorer physical condition. Other than the Commodore branding and the orange RETURN and SHIFT keys (instead of dark cyan like on the Toshiba original), the keys, their layout and even the BASIC keyword labels above them are identical to the HHC-4.

The LCD is also exactly the same, modulo the iffy elements. On both the marketing shot and on this picture you can see the same character glyphs, same 24x1 layout (in characters, not a continuous dot-addressible matrix), same LCD indicators for the unit of angle for trigonometric functions (DEGrees, RADians or GRADians), COMP and CAL indicators (to be discussed presently), a SHIFT indicator (extinguished in this shot) and an unexplained LCD dot on the far right.

On the CES show floor the HHC-4, presumably named as such for its base RAM, was demonstrated with a black Commodore-badged version of the IHC-500 docking station. This colour image is from the cover of The TORPET February 1983 which Bo Zimmerman sent to me many years prior. The LCD isn't visible in this or the next shot but the rest of the unit matches.

This photograph is scanned from my copy of COMPUTE! March 1983. Although it's a lower quality black and white image, it's easier to see the labeling on the dock itself.

And here's the original IHC-500 for comparison (again, ignore the discolouration on this thrashed specimen). With some squinting and possibly psychoactive substances, on the black HHC-4 dock you can make out the power LED, cassette and printer toggle switches, and the printer line feed button, demonstrating it's also the same device. The IHC-500 says it's a "mini-printer & peripheral interface"; the HHC-4's dock in the COMPUTE! photo says it's a "peripheral interface & mini-printer." Tastes great, less filling.

I could only find one first-person observation of the HHC-4 in actual use, and it was not complimentary: The Transactor in its January 1983 issue, while incidentally noting the HHC-4's Toshiba provenance, complained "the processor is slower than snail doo" and "[s]creen scrolling is abismal" (sic). Perhaps we should crack the Pasopia Mini open to see what makes it tick, yes?

The underside of the unit shows the battery compartment door and the RAM expansion door (the 12K RAM card is inserted through a slot when the door is removed), along with slots to connect with the dock. The back cover just comes off with screws.

The electronic design hails from that awkward period where surface-mount technology was available but many components were still through-hole, yielding a curious hybrid of the two (a particularly interesting landmark is the little TDK DC/DC board pushed on top). The main "CPU" logic board is on the top half with the LCD and keyboard, with the "BUFfer" interface board on the bottom. Power wires from the battery compartment and to the piezo speaker are visible. The RAM card, if it were present, would slot into the compartment under the piezo.

The buffer-interface board on the bottom handles connections to and from the RAM card (if installed) and the docking station. A handful of CMOS logic chips control signaling back to the CPU board through a ribbon connector. The 4049s are standard hex inverters but the 40H138 is the Toshiba equivalent of a 74HC138 decoder-demultiplexer and the 40H367 is equivalent to the 74HC367 logic hex buffer/line driver.

The top is where the action is and is dominated by two Toshiba T-chips. We've dealt with T-chips before, which are Toshiba-specific multi-function ICs. The two T-chips in the Pasopia Mini are labeled T6808 and T6809A (not to be confused with the Motorola 6809). Between them are two TC5517 2K SRAMs (to equal 4K) and a Toshiba 40H245, equivalent to the 74HC245 octal bidirectional bus buffer.

A couple things to remember about T-chips: their numbers are roughly sequential and higher numbers indicate more recent designs than lower ones. Consecutively numbered chips are usually variations of the same chip with things like different on-die mask ROM. For example, the 1985 VTech Laser 50 has two T-chips marked T7812 and T7813 that service separate halves of the LCD. Two-CPU (note: not a multiprocessing system) arrangements are very common in earlier pocket and handheld computers because of the then-limited I/O and ROM capacity on any one die. The Sharp PC-1211/Tandy PC-1, Casio FX-780P/Tandy PC-5 and Casio FX-790P/Tandy PC-6 are other well-known examples (conversely, the dreadfully cut-down Casio FX-5200P/Tandy PC-7 is an example of what has to happen to eliminate one of those CPUs). Most such systems alternate between the two processors using a single pool of common RAM. In the case of the Laser 50, the chip numbers compared to T-chips found in other related VTech computers and various MSX machines make it likely they were based on the Z80 and differ only in their onboard ROM.

However, these two chips substantially predate those better-known ones. They are consecutively numbered with the same 92 pin count, so they're probably the same chip modulo mask ROM, which would make this probably another split-CPU system. 20K is an unusual amount of ROM — 8K or 16K would be more typical, and we'll discuss its unusually weak BASIC and character set a bit later — but without any proof to the contrary let's assume that the ROM complement is also evenly split. As far as I/O lines, the T6809A appears to service the majority of the LCD (probably the characters) and the entirety of the BUFfer board, while the T6808 has a few LCD lines (probably the indicators) and what looks like the keyboard. We do not see an obvious crystal or oscillator here, and it may well be internal. Most similar systems ran at around a few hundred kilohertz for power-saving reasons.

The actual CPU architecture is much harder to guess at, and all we know for certain is that it's 8-bit and CMOS. Of the then-common 8-bit architectures, in 1983 two miniaturized CMOS Z80s would have been surprising and Toshiba was never known to be a licensee of the MOS 6502 (despite having the stock symbol 6502 — no joke!). It should be parenthetically noted that at the time Toshiba did have licenses to produce MCS-48 (i.e., Intel 8048) and 8085 chips also, and the CMOS parts in their TLCS-84 line of 8048-compatible microcontrollers could run on as little as 10mA, but it's not clear they had enough I/O to do the things these chips are doing. Toshiba had their own in-house small architectures as well like every other semiconductor maker, so it seems reasonable to conclude the T6808 and T6809 are very likely custom low-power designs, something also typical of most contemporary pocket-handhelds.

In fact, one of those Toshiba-specific architectures appears in the IHC-500 dock. The dock plugs into the BUFfer board with this side connector; the slots on the underside of the main unit ensure a proper mate. The cassette interface and printer can be turned on and off individually to reduce power usage.

The dock has multiple connectors, including a power jack and remote, mic and earphone cassette jacks, and three pin headers. The smallest of the pin header connectors is the 10-pin RS-232 serial and it shouldn't be too hard to figure out at least the basic transmit and receive pins, but the 16-pin Centronics-compatible parallel connector will be a little harder and the largest connector for the video monitor is completely unknown.

With the back off, we can see it has a rechargeable Ni-Cd battery pack. In most pocket/handheld peripherals the device is run by the battery, not by the A/C adapter input. In this case the unit appeared to run fine with the wall adapter, but the battery pack would certainly be dead and it looked like a standard four-cell 4.8v pack, so I replaced it with an off-the-shelf one. The new pack seems to run and charge fine. Although it looks like it should be able to power the computer as well, it will not turn on with the computer connected without batteries, even with the A/C adapter. The IHC-501 adapter uses a 5.5mm barrel jack and is officially 9V 500mA centre-positive, but (presumably unregulated) it measures 15.3V on the multimeter. There are no battery charging or capacity indicators, though this wouldn't have been unusual for the time.

The dock's logic board is a more conventional through-hole arrangement, most likely for cost reasons. Other than the logic chips, the two main ICs are also Toshiba-specific, a TCP4620B and TCP4630B. Both chips are single-chip ROM, RAM, I/O and CPU microcontrollers based on Toshiba's 4-bit TLCS-46A architecture, both with 34 GPIO lines and a clock speed of 100kHz (divided by four from a base 400kHz clock); the sole difference between them is their memory capacity (96 nybbles of RAM and 2Kbytes of ROM in the 4620; 160 nybbles and 3Kbytes in the 4630). Both chips are "B" variants which can be placed into a hold state to save power. Compared to Toshiba's other current 4-bit microcontrollers, the TLCS-46A series was their most power-thrifty and consumed just 0.4mA on the standard 400kHz clock compared to as much as 70mA for the TLCS-43, but TLCS-43 chips could also run two and a half times faster. The TLCS-46A is a Harvard architecture that defines a 4-bit accumulator, a 4-bit status register (zero, I/O, carry and hold), two 4-bit index registers (H and L) which can be combined into an 8-bit address, a 4-bit "zero page" or general purpose register (Y), multiple GPIO registers ranging from 4 to 8 bits in width, and a 12-bit program counter, though in neither chip here is there sufficient ROM to max it out.

The interface connector to the Mini is at the top left and the ribbon cable to the printer mechanism is at the bottom left. The TCP4630B seems to primarily service the thermal printer, which makes sense given its larger ROM, leaving the TCP4620B to handle the serial and parallel ports. It seems most likely that the docking station participates as a member of the system bus, allowing the T6808/9 in the Mini to talk to the docking station CPUs directly. On the other hand, nothing on this board is remotely powerful enough to drive a composite video signal and a number of the bus lines go directly to the video connector. This suggests the serial and parallel connectors were most likely passive and with a little of work we could probably fashion cables for them, but the video connector was almost certainly an active device and pulled data directly from the Mini's RAM over the bus for display. Indeed, the Mini seems to sense the absence of the display separately of the printer and tape interface (more when we get to BASIC). Regardless of how it worked, however, I can find no evidence it was ever sold.

That said, it's also not clear the video adapter could have done a whole lot other than text, and even then of only a very limited sort because the IHC-8000's character set is quite constrained: only numbers, uppercase, basic punctuation and these few special characters. ASCII values under 32 and above 95 print nothing on the LCD. While the dot-addressable graphics of the Sharp PC-1500/Tandy PC-2 are clearly exceptional for the genre, the Pasopia Mini has no graphics characters or even mathematical characters, other than for scientific notation. Other pockets like the Sharp PC-1250/Tandy PC-3 had similar limitations, but they also didn't purport to hook up to a TV set.

There are two distinct modes on the IHC-8000. Again, modal operation wasn't unusual for pockets which usually had separate program and run modes, but here the modes are CAL for calculator mode and COMP for computer (i.e., BASIC) mode.

In CALculator mode, a triangle prompt is displayed at which you can enter floating point expressions for immediate evaluation. These are in fact full BASIC expressions, with all standard numeric operators, operator precedence and parentheses supported, and you can even define and reference floating point variables. Variables are limited to one character (i.e., A-Z) and are not defined by default: if you reference a variable not currently defined, you get an error (ERROR 3) instead of assuming a default value of zero. Other than numeric functions no other BASIC keywords are supported, not even LET (e.g., you can assign A=5, and the answer is 5 and you can use A again in other expressions, but LET A=5 is a syntax error [ERROR 1]), and anything that generates a string as a result is not allowed.

Expressions can include ASC(), SIN(), COS(), TAN(), ATN(), SQR(), EXP(), LOG() (base 10), LN() (natural log), FRE() (more about that in a moment), SGN() and ABS(), and possibly other functions I haven't found yet. You can also use the pseudovariables PI and RND (a pseudovariable for a pseudorandom number), neither of which very reasonably can be assigned to. Bitwise logical operators are also supported, but only AND and OR (no NAND, NOR or NOT, though there is a unary minus, and no XOR or EOR). Operations in CALculator mode may change variables, but they won't alter any program in memory.

The E operator (a small capital E, ᴇ) lets you enter numbers in scientific notation like 1ᴇ6 for a million or 1ᴇ-6, though precision seems limited to 11 significant digits plus two digits of exponent, e.g., entering 9.999999999999999×10⁹⁹ got truncated into 9.999999999×10⁹⁹. Multiplying that by 10 got what appears to be an illegal quantity error (ERROR 2), and dividing 9.999999999×10^-99 by 10 yielded zero.

Once the result is on-screen, you can immediately type and append to it to use it in the next calculation, or use the cursor keys to edit it. It's a strange and arguably superfluous mode, but I suppose it saves you having to enter PRINT commands all the time.

Even by the standards of pocket computer BASICs, Pasopia Mini BASIC (in COMPuter mode) is pretty limited too. I should note that this machine came with no manual, so all of this I figured out by trial and error. On startup there are 3068 bytes available with the default 4K (as provided by PRINT FRE(0)). Like in CALculator mode, BASIC variables are limited to one character, though in COMPuter mode string variables are also supported (i.e., A-Z and A$-Z$), and again not defined by default: if you reference a variable not currently defined, string or otherwise, you get an error (i.e., floating point variables don't default to 0 and strings don't default to the empty string). As you define variables, they also pull from the memory pool, and PRINT FRE(1), will show memory minus BASIC text and defined variables as opposed to PRINT FRE(0), which only shows memory minus BASIC text. (You can also evaluate FRE(0) and FRE(1) from CALculator mode.) It is possible to DIMension a numeric or string array from direct mode with up to two subscripts (three is an error), and it is possible to re-DIMension them at will, neither of which is not allowed in many other BASICs.

The major BASIC keywords have shortcuts accessible by pressing SHIFT, then pressing the key. (Remember, no lowercase.) Those keywords are *CSAVE and *CLOAD (cassette save and load), *VERIFY (cassette verify), *LIST, *EDIT (requires a line number argument), MODE (all I got was syntax errors trying to use it, though), BEEP, WAIT and PAUSE (more below), *AUTO (automatically numbers lines with the increment specified), LPRINT and *LLIST (print and list to the printer), GOTO, GOSUB, RETURN, INPUT, PRINT and USING (more below), LET, IF, THEN, FOR, TO, STEP, NEXT, END and *RUN. Unlike most pockets with separate run and program modes, Pasopia Mini COMPuter mode has a direct mode like most home computers, and the keywords I marked with an asterisk can only be used in direct mode. If you start editing a line (say, with EDIT 10), you can use the up and down keys to go through other lines in your program text; changes are only committed when you press RETURN.

There are specific considerations given for the single line display. Both EDIT and LIST scroll long lines, LIST doing so automatically. PRINT prints the given string (and supports USING for formatted printing, such as PRINT USING "##.##";PI,), but it will immediately clear the screen and return, making it impossible for you to read what was printed — unless you stick a comma at the end, in which case it waits for a key. You can even do this multiple times (e.g., PRINT 5,2,). Alternatively, this is where the PAUSE command comes in, which is just like PRINT, and even supports USING, but waits for about one second after. However, putting a semicolon at the end of a PRINT doesn't seem to allow you to append with the next PRINT statement, and if you PRINT or PAUSE something too long for the LCD, it just truncates it instead of scrolling. There is a TAB(), but not SPC().

WAIT waits for roughly one second (it's not like WAIT on other platforms which watches a memory address), or if given a number, for about that many tenths of a second. (You can PRINT and then WAIT instead of using PAUSE, too.) BEEP beeps for one second, but BEEP with any non-zero value turns on the beeper (oy!) and BEEP 0 turns it off. The piezo is perfectly capable of sounding at different frequencies but the driver hardware doesn't seem to support that.

The machine also responds to other keywords and functions that don't have shortcuts. MID$() is supported for substrings and CHR$() for characters, what few there are, and of course all the numeric functions valid in CALculator mode are valid in COMPuter mode. NEW and CLEAR are self-explanatory, and it also supports READ, DATA and RESTORE (and RESTORE may take an optional line number). There is also a LOCATE that takes two parameters, presumably the screen coordinate, despite doing nothing on the LCD. This is interesting because CLS generates an ERROR 8 even though you should be able to clear the LCD screen too. That error is the same error generated by things like LLIST and CSAVE if not docked (ergo a "device not present" complaint), but CLS will generate that error even if it is docked, which again strongly indicates the video adapter had additional hardware the computer can detect. I imagine LOCATE only does something meaningful with the video adapter and it's surprising it doesn't error out in the same way.

On the other hand, it doesn't seem to support INKEY$, KEY$, GET, DEF FN, SOUND, PEEK, POKE, CALL, OPEN or CLOSE (all yield syntax errors regardless of any arguments I could think of). USR(), STR$() and VAL() don't generate syntax errors, but all of them yield ERROR 3 regardless of argument. If there are graphics commands like DRAW or PLOT, I couldn't find any. Since the ROMs are built-in to the T-chips, barring a means to display memory, decapping them is the only thing that could tell us the keyword set definitively — or how a whole 20K of ROM is needed for this limited of an implementation.

The Transactor's reviewer is right, by the way: the BASIC is quite slow, again suggesting the clock speed for the T-chip CPUs is probably on the order of a few hundred kHz tops. Its syntax and features wouldn't have been a strong fit with the larger Pasopia line, and it would really have been a bad fit with Commodore's. The odd thing is why Commodore chose the Pasopia Mini for the HHC-4, when another Japanese manufacturer was showing off their own HHC — even named the HHC — with the same or better features, and with a much more familiar architecture.

That computer is the Panasonic HHC from Matsushita, sold also in different cosmetic livery under their Quasar brand. It was originally developed in 1981 by Friends Amis, Inc., formed by Atari ex-president Ron Gordon, based on an earlier design sold by the company used as a simple language translator. That earlier unit used a Mostek 3870 microcontroller with installable ROM "capsules" to add languages. Matsushita asked Gordon's company to develop something similar with comparable expandability but as a programmable general purpose computer instead, and launched it in 1982. This is one of my Panasonic HHCs with its working printer-cassette interface, the most common accessory.

The full Panasonic HHC system could do everything the Commodore HHC-4 was slated to, yet the full system actually existed and you could buy it. This is the complete briefcase system you could have bought: in the left column the printer-cassette interface, RS-232 serial interface and RAM expansion (4K or 8K); in the right column an acoustic coupler, composite/RF video adapter and the computer itself; and in the middle, linking them all together, is a 7-slot expansion I/O adapter (the top is another slot). The video adapter can generate 32x16 characters in text mode using 5x7 character glyphs, or two semi-graphics modes that supported 128x64 with 8 colours or 128x128 with 4 colours. All of it here fit into in a Panasonic-badged attaché case that you could buy too. Built-in software includes a clock and alarm, calculator and basic text editor, driven by a simple menu system on its 26-character 1-line screen. It also had a built-in interpreter to run stored SNAP programs, a simple text-processing language.

A rear door opens to yield several available sockets for "capsules."

These capsules (this is my bag o' capsules) were ROMs with a capacitor and a little plastic handle, and could be fitted and removed without the need for a chip puller. They could also be made out of simple frames without weird cartridges or card edges, and the frames were keyed so they could only be inserted the right way up. Besides prototype EPROMs and more typical ROMs like BASIC or the SNAP development ROM, the capsule feature made the Panasonic HHC very popular in the insurance industry: you can see several of these insurance ROMs for "AUTO CA" (presumably California car insurance), "U-LIFE" (presumably some sort of life insurance) and "NYL TARGET" (New York Life, also life insurance). Everything was in ROM, so it started up instantly ready to go, and updating with new rates and policy requirements was as simple as sending out new ROM capsules. As a result it isn't hard to find HHCs with their printers for sale in good working order as around 70,000 units were sold; probably thousands of these portable insurance workhorses churned out tons of quotes over their service lives. The other expansion options are rather rarer and I don't have examples of my own, but I've seen them personally in other collections and I know they exist.

If we crack it open, we can see the rechargeable Ni-Cad pack (fortunately you can run them off wall power even if the battery pack is dead), the built-in ROMs and the base memory allotment, which is 4K in this unit. But what CPU is it running?

Lift up that black flap at the bottom and you get ... a Rockwell R6502! Wouldn't Commodore have wanted to sell a handheld computer with the CPU architecture it actually owned? (I should parenthetically note that while the manual says the 6502 is clocked at a "standard" 1MHz, the crystal next to it is 4194.3kHz. Assuming that it gets divided down by four for the Φ2 clock, that's actually 1.049MHz, or 4.9% faster than advertised.) The battery life on the Panasonic HHC is definitely shorter than the Pasopia's, but you could do a lot more with it, and you could certainly do it faster. The only significant advantage the Mini had over the HHC is the total physical size of the full system, and even that is largely due to offering fewer options.

Why Commodore went with the Toshiba Pasopia Mini for the HHC-4 over the Panasonic HHC will probably always be a mystery. Presumably Tokai's group would have also heard about Matsushita's forthcoming product, so maybe Commodore couldn't come to an arrangement with them that Tramiel considered satisfactory, or perhaps Matsushita just plain wasn't willing to license or be merely an OEM like Toshiba. After all, the Panasonic HHC was certainly more than successful on its own.

However, the reasons for the HHC-4 never being released are a bit clearer. Commodore was going in far too many speculative directions in a time of general recession and, retrospectively, mere months before the video game crash of 1983. At CES 1983, in addition to the three-computer Executive 64 line and the HHC-4, Commodore introduced an add-on music keyboard chassis with up to three extra SID sound chips, a digital drum pad, and an early version of what would become the Magic Voice speech synthesizer (using its own T-chip, the Toshiba T6721A).

Commodore also demonstrated this absolutely ridiculous remix of the VIC-20 using a 2" (that's right: 5cm) Sony Watchman screen in the case, and even announced they would co-produce the 16-bit Zilog Z8000 for use in future machines. Of all of these, only one machine from the Executive 64 family became available (the SX-64), the Magic Voice wasn't introduced formally until the 264s launched, the music synthesizer and drum pad disappeared somewhere around mid-1984, and the "SX-20" went in the trash heap where it belonged. As for the Z8000, Commodore eventually ended up using it in exactly one product, the Z8001-based CBM 900 running Coherent, which it then dropped during the prototype phase to concentrate on the Amiga.

Meanwhile, the HHC-4 was considered unsuitable for market in its present form, apparently even by Commodore itself. Despite management's initial optimism, Transactor reported rumours that its deficiencies became quickly well known to West Chester and that the internal electronics and software were to be completely overhauled ("Don't expect to see this one until it's improved," wrote the columnist). Depending on who you talked to, COMPUTE! said Commodore planned to sell the basic HHC-4 for $199 (about $630 in 2024 dollars), but Torpet said it would start at just $100 [$320] with another $100-$150 [$475] or so for the interface. At either of those prices, the amount of work required likely could never have been recouped.

For its part, even Toshiba seemed to disown the Mini. As part of the 1983 refurbishment, Toshiba offered an unusual upgrade for the original Pasopia/T100: a 320x64 LCD screen which could be installed on the back of the keyboard. Typical of early 1980s computers, the keyboard contained the logic board, so this made the system self-contained. It did not make it a true portable — at best it was a luggable, as it still had to be plugged into the wall for power — but it sold for just $295 [about $930 in 2024 dollars] or in a package deal with a hard carrying case, battery-backed RAM pack, 300 baud acoustic coupler modem, cables and main T100 unit for $1600 [about $5070]. In this configuration the entire assembly weighed about 25 pounds. With the LCD installed, the Pasopia/T100 looked like this, from Toshiba's Pasopia catalogue that year:

The portable Pasopia was the obvious Mini replacement: as powerful as the T100, because it was a T100. The Mini consequently disappeared from Pasopia catalogues, never to return, and even in its native country is not easy to find today. As for Commodore Japan, the subsidiary finally flamed out after the disastrous introduction of the Japanese Commodore 64, and while the also-unreleased Commodore LCD could arguably be a spiritual descendant, Commodore never produced or licensed another handheld computer again.