Return to Floodgap Retrobits' Tandy! Pocket! Mania!

TPM: Tandy Radio Shack Pocket Computer PC-4 and Tandy Pocket Scientific Computer PC-5, PC-6, PC-7

As we explained on the main page, the PC-4/5/6/7 represented the major alternative architectural fork, viz., the Pocket Computers based on Casio designs rather than the Sharp units that were the basis of the PC-1/3/8 and PC-2.

In broad strokes, all of the units are very similar and use the same essential BASIC dialect, making the entire line upwardly compatible (except for the very first PC-4 which has some peculiarities to be discussed below). Although the Casio and Sharp units weren't compatible across the two families, some of the general concepts in the Sharp interpreter were carried over into Casio's BASIC, particularly split RUN and PROgram (in this case, WRiTe) modes, and the standard 26 fixed variable allocation. However, a critical advance of the Casio units over the Sharp units was their use of segmented programming spaces, allowing up to ten program slots ("P0" through "P9") in what could almost be considered to be a rudimentary filesystem. You could fill up P0 with one big program or have multiple other ones in the other spaces, all pulling from the same pool of memory. Program slots could call other slots, a primitive sort of library system, and program slots could be written out to tape either individually or all ten at once. Although the Sharp units tried to mitigate their single stored program limitation with keyboard shortcuts, theirs was clearly the less elegant approach. The segmented programming space feature was common to all of the Casio units.

Unfortunately, the Casio units were obviously selected because of their pricepoint: while this wasn't a big liability in 1983 with the PC-4, it was definitely a problem in 1985-6 with the remainder, and as a result this page spans the gamut from one of the best, two mediocre, and the positively worst Tandy Pocket Computer in the bunch. What a way to clobber the line!

Remember: in the computer option descriptions, M = memory expansion (with maximum available), T = cassette tape interface, P = printer interface, S = serial interface. Lists of non-Tandy peripherals and options are not guaranteed to be exhaustive. MSRPs best guess from available contemporary catalogues.

[The PC-4 family (both the original and the B model). Click the picture for an enlarged view.]

Tandy Radio Shack Pocket Computer PC-4

CPU: HD61913A01 (CMOS 4-bit) @ 200kHz
LCD: 12 character, non-dot-addressable
Base memory (free/total bytes): 544/1024
Size: 165x71x9.8mm, 116g (0.25lb)
Options: +1K M, T, P

[The PC-4 (26-3650). Click the picture for an enlarged view.] 26-3650/26-3650A
Introduced 1983, "1985"
Originated as Casio PB-100
OS: Custom (10K ROM)

[The PC-4 (26-3650B). Click the picture for an enlarged view.] 26-3650B
Introduced 1985
Originated as Casio PB-100F, modified
OS: Custom (10K? ROM)

Tandy Radio Shack Pocket Computer PC-4

Do you get the sneaky suspicion that I'm the teensiest, weensiest bit biased towards the PC-4? Do you have the faintest inkling whatsoever that I have a soft spot for this deceptively diminutive charmer?

['That's right!'] No, I won't make any bones about it -- the PC-4 is my favourite Pocket Computer, not least because it was the first one I ever owned, as a Christmas gift from my folks. It went off to camp, went to school and went to the rollerskate rink, and managed to survive a near-constant thrashing on the same pair of batteries for several years until it met an untimely end at the hands of a sizeable Coca-Cola spill. It taught me programming, played a lot of homemade games, kept (a small collection of) notes, ground through calculus problems and even computed atomic weights for my high school chemistry course until its tragic demise. A few years later, I acquired another one from a TRS-80 reseller and put it right back to work where the old one had left off, this time crunching the math for concentrations in the biochem lab and running a few homebrew games. That second unit is still in my collection.

Obviously, my point is not to say that the PC-4 is the only useful or functional Pocket Computer, because clearly you could do stuff like this with the others, and in many cases better (certainly faster given that ghastly clock speed; the clock is actually 400kHz, but divided down by two for the CPU). Where the PC-4 stands out is its size, especially compared to a monster like the PC-2; its tiny power consumption and reasonable durability; and most of all, the fact that it was dirt cheap (a 1983 MSRP of just $69.95). Even on the used market, it still exists in enough numbers to keep prices reasonable for collectors and users. Plus, the segmented programming space combined with its minimal battery draw ensure that it'll be a long time before you'll need to dump stuff out to tape, and you won't have to take the tape interface with you to have a portable program library. For this reason, the PC-4 is far and away the unit I recommend to people with a casual interest in Pocket Computers because it has a really attractive form factor and it carries everything in a single package to get you started off quick. (For the next step up, consider a 4K PC-3.)

The PC-4 was originally based on Casio's ultra-low-end PB-100 system, chosen because it was small and inexpensive. Although a 12-character LCD was unavoidably cramped for a display, and having just over half a kilobyte of memory made for tight programming, this was certainly acceptable for 1983 based on its physical size and its low cost. (It wouldn't be later, but we're getting ahead of ourselves.) This initial form of the PC-4 (26-3650) was released in 1983 and used the BASIC of the PB-100, a fairly parsimonious if mildly limited interpreter. Naturally possessing both the dual-mode and 26-fixed-variable conventions of the Sharp models' BASICs, PC-4 BASIC had some additional constraints to allow it to make the most of the little free memory available. Instead of the dimensionable string length of the PC-3, the PC-4 limited users to eight byte strings and used a large 30-byte accumulator simply called $ for larger manipulations. Dynamic variable allocation and additional array space could be accomplished manually using the DEFM command, although the string length was immutable. The PC-4 was otherwise a very simple BASIC with most elemental BASIC functions and control statements, plus additional trigonometric functions; notably there was no REMark statement, but that would have been terribly wasteful anyway. Tandy thoughtfully included separate Programming and User Manuals free in the box, one targetted at the beginner programmer and one for those already familiar with BASIC, respectively (the boy wonder mascot I use for decoration is from the Programming Manual), the only Tandy Pocket Computer so equipped.

The internals of the PC-4/PB-100 and its successor units are intriguing. The CPU is a Hitachi microcontroller, the HD61913A01, and is used throughout the series; the internal design and presumably the instruction set appear to be related to the 8-bit Hitachi HD61700 series. I say "appears" because the BASIC and operating system ROM is entirely internal to the microcontroller and is never exposed on the bus. RAM is accessed over a 4-bit data bus in a serial fashion. The microcontroller also controls the keyboard, the LCD and external peripherals. The PC-4/PB-100 uses a custom character set, not ASCII.

[A sample PC-4 (26-3650B) program from the Programming Manual showing typical style. Click to download an enlargement.] 544 bytes of programming space made for programs with a minimum of waste, as the scheduler (written for the later 3650B) at right demonstrates. This sample program also elegantly shows how to use the separate program spaces as subroutines, though of course you could easily GOSUB within your own space to make your application more encapsulated. The optional 1K memory expansion connected to a flat contact ridge accessed with the back cover off, and was practically essential as the near-tripling of memory greatly enhanced the number and complexity of programs storable in the unit (not to mention a lot more dates in your scheduler book). With 1568 bytes available after its installation, suddenly the PC-4 became a much more enjoyable little system. The almost identical FX-700P's memory map and hardware description undoubtedly also apply to the PC-4; the major difference is that the top of memory is of course lower depending on how much is installed. The FX-700P has 2K of RAM, therefore 4096 nybbles, the same as an expanded PC-4. Although a 12-bit address is sent over the internal bus to the RAM chips, theoretically permitting 8192 nybbles of memory, it doesn't seem like the most significant bit is used.

For Casio, the PB-100 was the basis of a great many successor units, mostly featuring upgraded BASIC and additional software features, but Tandy only chose one of its direct descendants to market. The 1985 PC-4 (26-3650B), the member of the PC-4 series in widest circulation, did not mitigate the small memory but did improve the BASIC considerably using the version extant in the PB-100F. This new BASIC variant could dynamically allocate variable space even within a program, added READ and DATA statements, added ON-GOTO/GOSUB, offered password protection, added STR$ (VAL already existed) and various degree/minute/second conversion functions, changed the name of various commands slightly, and just for yuks, allowed REM. It was fully upwards compatible with its older sibling, which was itself redesignated as the 26-3650A; this label variant is extremely difficult to locate as few of them were produced. The only place the "A" appeared, in fact, was on the box and the rear of the unit; it appears that the 3650As were simply old stock PC-4s that were rebadged and sold until gone, as they lasted only a few months.

However, what was most important to consumers was that the price on the new 3650B PC-4s was reduced down to $59.95, appropriate as little else about the technology had changed. These units were also introduced without the Radio Shack label, just like the 4K PC-3, and were simply called the Tandy Pocket Computer PC-4. (Compare the label on the two units above to see the difference.)

It is noteworthy that the 26-3650/26-3650A's BASIC dialect was effectively a dead end: the PC-5/6/7 are all based on the 26-3650B's BASIC, and all of the subsequent Casio-Tandy models required conversion of the unique keywords of the original PC-4 (such as VAC, etc.).

In addition to the 1K expansion pack (unfortunately, no other option packs were made for the rear option slot, RAM, ROM or otherwise), Tandy also offered a rebadged 20 CPL/55 LPM thermal printer for $79.95, along with a separate tape interface for $39.95. The printer used a rechargeable supply, which could not power the computer or the rest of the system, but the cassette tape interface used standard alkaline batteries. Designed to "stack" one on the other, the net effect was something of an unstable joke and made the entire package swell to over double in size. In fact, the printer was so dependent on the cassette interface for structural support that the printer package actually included a set of dummy prop struts for users who didn't have the tape connector (plus a huge thumbscrew securing the printer to the tape interface as an insurance policy). Helpfully, both interfaces include "rests" to store the port covers just like other Pocket Computer peripherals. The "family picture" above shows a PC-4 saddled with both units in their typical configuration. Tapes written by the 26-3650/26-3650A are fully compatible with the 26-3650B although the converse, naturally, is not guaranteed.

A very nice option, but unfortunately very hard to find, is the Casio FP-40 thermal printer. This unit can print up to 80 CPL and best of all is powered by AA batteries. However, it requires an SB-42 link cable to connect it which is even harder to find than the FP-40 itself.

Due to its respectable sales, the PC-4 was a particularly long-lived series and was not retired until 1987.

Here are the Tandy options/catalogue numbers and their Casio equivalents for both PC-4s:

We have the user manual for the 26-3650A (rebadged original PC-4), the PC-4 cassette adapter manual, and the service manual for the "O.G." 26-3650 (which should apply generally to all PC-4s including the 26-3650B) on our Gopher server. (If your browser doesn't speak Gopher, you can get a Gopher client or add-on that does.)

Other PC-4 sites (note that many sites are completely unaware of the 1985 version):

[The PC-5 and PC-6, with PC-6 RAM expansion option. Click the picture for a larger view.]

Tandy Pocket Scientific Computer PC-5, PC-6

CPU: HD61747 series (CMOS 8-bit)
LCD: 24 character, non-dot-addressable
Size: (folded) 142x71x18mm, (open) 142x142x9mm, 165g (0.36lb)
OS: FX series

[The PC-5 (26-3670). Click the picture for a larger view.] PC-5 (26-3670)
Introduced 1985
Originated as Casio FX-780P
CPU: HD61747B10/B11 @ ???kHz
Base memory (free/total bytes): 3552/4096
Options: T, P (via link cable)

[The PC-6 (26-3672). Click the picture for a larger view.] PC-6 (26-3672)
Introduced 1986
Originated as Casio FX-790P
CPU: HD61747B20/B25 @ ???kHz
Base memory (free/total bytes): 7520/8192
Options: +8K M; T, P (via link cable)

Tandy Pocket Scientific Computer PC-5, PC-6

(A bit of housekeeping: because the PC-5 and PC-6 are so similar this text is intended to be mostly applicable to both units except where specified.)

On paper, the PC-5/6 ought to be the best Pocket Computers Casio ever made (originating as the Casio FX-780P and FX-790P respectively) and Tandy ever released. ['Yeah!'] After all, here's a 4K (8K if you're a PC-6) system with advanced mathematics and statistical analysis features (anyone for HYPCOS? linear regression?), a data bank, a full assembler, compatibility with PC-4 peripherals, upwards compatibility with the revised PC-4 BASIC, and the most extensive keyboard of any Tandy Pocket Computer system -- who'd say no to that? Give me two!

Then you open the box and reality sets in. Yes, it's a big keyboard, but most of the business end is in a membrane arrangement on the top which is not only mushy to type on but terribly clumsy because of the clamshell design (and flexes in a fairly alarming fashion). Yes, you can use PC-4 peripherals, but you have to get a link cable and the arrangement is even sillier than what the PC-4 had to put up with. Yes, it has an assembler. but it's a ridiculously underpowered one for a completely contrived architecture having nothing to do with the CPU itself. Worse, it had extremely primitive I/O and no direct access to the hardware, couldn't coexist with other data in the Data Bank, and couldn't be mixed with BASIC programs. In plain terms, it's this sort of "commitment to compromise" that makes the PC-5/6 fall far short of its marketing.

This isn't to say that the PC-5/6 are all bad. In fact, they do have some very nice features, particularly the built-in advanced trigonometry and fast statistical analysis packages (thus very deserving of the Pocket Scientific Computer moniker, the first Pocket Computers to be designated as such), and their impressive Data Bank feature which allows part of the program area to become a very simple data store. BASIC extensions allow inserting, modifying and deleting records, plus simple search and retrieval. The computer even has a primitive editor for entering them and then accessing them later. Like the PC-1, the twin CPU approach was to handle all the on-board software, the huge keyboard and the double-wide LCD. One CPU just couldn't do it all, so one handles part of the LCD, part of the keyboard and BASIC, and the other handles the other parts of the LCD and keyboard, plus the Data Bank, Assembler and I/O port. (Remember this for the PC-7.)

Another nice feature is that the PC-5/6 are the only Pocket Computers with a secondary backup battery, meaning you can replace the main batteries without losing everything; RAM is maintained by the auxiliary battery if the main cells die or are removed.

However, these nice-to-haves are drowned out by some very concerning practical issues. Besides the fact that my Atari 400 has better tactile feedback than this device, it doesn't make sense to put commonly used keys like the alphabet on the membrane while single-stroke, less operated functions like the trig operations were on the real keys on the bottom. Better to toss the membrane entirely, of course, but touch typing is more important when you're putting in long strings of text than just getting the sine of 30 degrees. In addition, build quality for the case can only be charitably called mixed: my NOS PC-6 unit was in real danger of splitting down the hinges after just a few times opening and closing it, and the front latch practically begs to be cracked. (In fact, most of the PC-5 and PC-6 systems you'll find on eBay indeed suffer from broken hinges and latches, and there is no easy way to repair them. Protect your system's hinges by putting finger pressure over the hinge covers as you open the clamshell, or tape them over for reinforcement.)

The assembler is also a serious disappointment. In fairness, Casio didn't come up with the architecture: it's actually a Japanese-standard 16-bit virtual machine, the 1969 COMP-X, programmed in a simple assembler format called CAP-X. COMP-X was devised as a means of ensuring engineers taking the standard Information Technology Engineer Examination (ja-JP: 情報処理技術者試験) didn't have an unfair advantage being asked about a computer language they were particularly proficient in. Ergo, I guess it made some sense to come up with a totally alien architecture and programming language to show examinees had basic computing knowledge while giving none of them a special edge. However, the COMP-X virtual machine was primitive even by 1970s standards, and CAP-X originally supported only 12 basic operations (halt-and-jump, jump if not zero, jump on condition bit, jump to subroutine, left/right shift, load 8-bit immediate and zero, add, subtract, load from memory, store to memory, logical AND and logical XOR). Because COMP-X originally had no memory-mapped I/O or ports, READ and WRITE instructions were later added to CAP-X for user communication. But all these did was dump and accept register contents at a prompt and had no means of customization, making them little better than a manually operated debugger.

There were few systems that could run CAP-X code directly, even in Japan (ironically most of the others were from Sharp), and for its home market that may well have made the FX-780P/790P appealing as a portable system for hobbyists and students who had to learn it. But as a programming language CAP-X is dire and badly limited, and compared to BASIC on the PC-5/6 is impractical for more than simple tasks: because of its limited range of operations, programming in it is wordy and cumbersome, even more than you would expect from a low-level language. Its artificial nature does make the system immune to a rogue program trashing memory or locking up the computer, and it is several times faster than BASIC on those algorithms it can reasonably do, but even this advantage is blunted because you can't call CAP-X code from BASIC to get the best of both worlds. The assembler also takes over the Data Bank by using it as its text editor, preventing you from storing anything else there but source code. With all those caveats COMP-X/CAP-X thus served little pedagogic purpose for the American domestic market, and almost no practical one. Tandy didn't even mention where the language originated from in the manual -- why would they when no one would be taking that test here anyway? When COMP-X was replaced on the ITE exam by COMET in 1986 and CAP-X by CASL, Casio quickly abandoned the older virtual machine in future devices, none of which were sold by Tandy.

Finally, the PC-5 makes this worse with less RAM and actually having a memory expansion slot but covered up inside with a plastic sticker. (That said, it's not clear the system would use one even if connected.) What really doesn't make sense is the two units were introduced nearly simultaneously; the PC-5 lasted less than a year before it was replaced by the PC-6, but both units retailed for the same price of $119.95! The units can be easily distinguished by their labels (of course); the front RESET button, which is only on the PC-6; and the MEMO key, which is "MEMO (SEARCH)" on the PC-6.

Despite their various and multiple sins, it's startling how popular the PC-6 is today and I can only attribute this to the form factor, which is admittedly fairly sexy -- and probably why Tandy jumped on them as "mini laptops." Indeed, more than any other model, the PC-6 tends to command the highest prices for a Tandy Pocket Computer on auction and used computer sites. The PC-5 is considerably more uncommon because of its short lifespan, but being less powerful is not particularly prized. By contrast, the PC-6 had the longest lifespan of any Pocket Computer, still appearing in Radio Shack catalogues as late as 1991 long after the rest of the line had faded away. As evidence of this popularity, the official $119.95 MSRP never changed!

Here are the Tandy options/catalogue numbers and their Casio equivalents for the PC-5/6:

We have the service manual for the PC-6 (which generally applies to the PC-5) on our Gopher server. (If your browser doesn't speak Gopher, you can get a Gopher client or add-on that does.)

If you want to play with CAP-X/COMP-X yourself, here is a full simulator of the Assembler simulator, implementing CAP-X/COMP-X with enhancements in Perl written by yours truly. It includes documentation and some sample games and other programs you can type into your real machine.

Other PC-5 and PC-6 related sites:

[The PC-7. Click the picture for a larger view.]

Tandy Pocket Scientific Computer PC-7

Introduced 1986
Originated as Casio FX-5200P, modified
CPU: HD61747B16 (CMOS 8-bit) @ ???kHz
LCD: 12 character, non-dot-addressable
OS: custom
Base memory (free/total bytes): 1568/2048
Size: (folded) 139.5x87x13mm, (open) 139.5x180.5x10.4mm, 146g (0.32lb)
Options: utter bupkis

[The PC-7 (26-3673). Click the picture for an enlarged view.]

Tandy Pocket Scientific Computer PC-7

Everyone loves a train wreck, so they should be just gaga over this piece of crap. ['Hmmm!'] Okay, that might be a bit too vitriolic, but if there's one Pocket Computer that deserves to be insulted by calling it a calculator, it's this one. Originating as the Casio FX-5200P, this was Tandy's answer to the popular PC-4's marketing slot (priced slightly higher than the 1985 26-3650A/B at $69.95) and easily guaranteed itself instant obscurity by faithfully implementing all of the PC-4's worst elements, i.e., its slow processor, cramped screen and tiny memory, just like it drew on the worst elements of the PC-5/6 with its crufty construction, partial membrane keyboard and folding design. Apparently, what Tandy failed to learn was that anything low-end in 1983 doesn't cut the mustard three years later for ten bucks more. This is the same essential mistake they made with the PC-8, only the PC-7 is far, far worse.

In essence, the PC-7 is a heavily downgraded PC-5, using the PC-4 LCD with just one CPU and eliminating the battery backup. The PC-7 is modestly upgraded over the OEM FX-5200P, which only has 1K of RAM and a slot for a 1K OR-1 RAM pack -- even Tandy couldn't stomach another 1K model, so they prevailed upon Casio to incorporate the OR-1 directly onboard, giving the unit 2K standard and leaving a ghost slot like the PC-5. (Casio does not appear to have ever released this variant of the FX-5200P under its own name.) Like the PC-5, it has the same upgraded BASIC with the PC-5/6's advanced math and statistical functions and keys for same. However, because there's only one CPU and thus half the ROM of the PC-5, only BASIC is supported (no Data Bank, no CAP-X assembler).

So much for any upgrades, now for more downgrades: the PC-7 does not have an I/O expansion port, because that support was also on the second CPU that was eliminated. Yes, you heard right. No peripherals, no printer, and most importantly, no tape. I'm sorry, did you say your batteries were getting low? Sucks to be you, my friend. Don't bother to put anything of substance in this device because once the batteries have to be changed, it'll be going to the great PC-1 in the sky. With that sobering thought, maybe it's not such a great loss that the PC-5/6 Data Bank didn't make it to this unit after all.

But wait, there's less! If you didn't like the clamshell PC-5/6 design, the PC-7 will really steam your broccoli because its flipface is of even lower build quality: all but my NOS unit show obvious flex strain on the join and some have even torn open, revealing the ribbon cable which is clearly being stressed even more by the case distortion. Casio didn't learn anything from the PC-5/6 and went right on slapping the alphabet onto the membrane half of the keyboard, although in this case you're not going to be doing much programming on it anyway.

Plain and simple, this doesn't deserve to be called a Pocket Computer. Sure, you can write programs with it, but because you can't save them you're restricted in a practical sense to quick one-offs and simple functions that you can afford to lose, and where I come from that's what calculators are for. Although technically superior and higher quality (relatively speaking), the PC-8 that followed the PC-7 actually exited the market earlier -- the PC-7 itself lasted until 1990! As far as I'm concerned, about all the PC-7 did do was make the PC-8 look not quite so bad by comparison.

I'd dearly love to tell you Tandy's options and catalogue numbers and the Casio equivalents, but there aren't any.

Other PC-7 sites:

Back to the main page | PC-1/3/8 | PC-2
... Cameron Kaiser

Did I mention I like the PC-4?