ЕЛЕКТРОНИКА MK 52 & 61

[jklsadf]

The 'k' key is probably because this was originally the indirect key for doing indirect branches and loads/stores. On the elektronika B3-34 and the mk-54 (which are the predecessors to the mk-61 and mk-52), the 'k' key was only used as an indirect key, and not as a 2nd shift key. In russian, the word for indirect starts with a 'k', at least according to google translate.

I'm not sure about why they use an 'f' for the main shift key instead of a phi. Maybe it's because phi has other meanings in math, or maybe it's because they just copied HP or other western manufacturers. "Error" doesn't translate directly to russian either, but it's what's displayed when there's a calculation error.

Looking back even further, the b3-21 (the predecessor of the b3-34 and mk54) had a 'p' for it's primary shift key, although it also had a 2nd 'f' shift key for a few functions. I'm not sure what either stand for.

The fact that these calculators were manufactured into the mid-90s was probably for backwards compatibility reasons, since I can't imagine they would have been cost competitive by then, with the vacuum fluorescent display and large number of ICs. My mk-52 has a date code of '94, although according to wikipedia they were only manufactured until '92 (but wikipedia is probably just wrong, since it also says the mk-61 was only manufactured until '94, but yours has a date code of '95). Apparently a lot of programs were written for these calculators for science, engineering, and business applications, since unlike in the west, personal computers were much more expensive in the soviet union back then. Compatibility goes all the way back to the b3-34 first introduced in 1980 -- even a lot of the undocumented behavior. Some things were kept even from the b3-21, which I don't think make a lot of sense in the later calculators, like x^y instead of y^x. It made sense back on the b3-21, since there was no 10^x function, but I have no idea why they kept it that way on later models. Interestingly, the x^y function preserves y in the register instead of dropping the stack down. I'm not sure why.

By the end of the production run in the '90s, these calculators would have been extremely out of date, using a total of 9 separate ICs for what could have been done in 1 or 2. See https://www.hpmuseum.org/forum/thread-2345.html for more information. The expansion ports at the top were for ROM modules and factory testing.

Most of the ICs use some kind of a ceramic surface-mount/chip-on-board hybrid mounting method, which I've never seen before, probably to reduce cost. The vacuum fluorescent display is one of my favorite parts about these calculators, although they draw a ton of power, and have reliability issues. (Some of the VFDs aren't sealed that well where the pins go through, and the driver board for producing the high voltage needed for the VFD is the biggest source of failures on these calculators due to aging electrolytic capacitors). But with new capacitors and and the driver board properly adjusted, these are one of my favorite displays on a scientific calculator, especially with the slanted top panel on the mk-52 for desktop use. The keyboard (both the physical construction and the debounce algorithms used) are probably the worst part of this calculator. It's impossible to double press a key quickly and have it register twice. Entering a lot of numbers in succession quickly also seems to cause problems.

These calculators were extremely well documented, at least if you can read russian (which I unfortunately can't). Not only did the user manual include schematics, but there was a whole separate book written about these calculators (https://www.hpmuseum.org/forum/thread-23...l#pid41226). Some of the accuracy specifications in the manual are slightly optimistic, although it's nice that they actually included them at all, since most calculator manufacturers (especially today) don't. Calculating 3^201 gives a value of 7.9682684E95 which has a relative error of more than 1.8E-5 compared to the actual value of 7.968419666E95, although the claimed maximum relative error is 1E-6. For most values, though, the calculator does meet its accuracy specifications. This is probably one of the most taxing examples possible. Specifically, it's the calculation listed in the HP 15C's advanced user guide as giving the worst error on the 15C.

These calculators do not carry any guard digits in addition to what's displayed. They also don't round properly, and instead just truncate the answer. Given that, the internal algorithms are actually fairly accurate. (Most calculators do rely on having internal guard digits to give an accurate displayed result, and also rely on handling special cases separately, such as y^x for integer values, to give better accuracy.) I'm not sure if internally they use any guard for calculations (I think it would be difficult not to?)

The manual looks (I personally don't have a copy) like it's printed on extremely cheap/thin/yellowing paper, but that was fairly typical of printed materials in 3rd world countries back in the day, even for things like textbooks. It might be a shock for westerners, but I'm sure most of the target consumers would've been used to similar quality printing...definitely no full-color HP manual quality here. Regardless though, looking at the actual content (although I can't actually understand any of it), the manuals and schematics seem to be fairly detailed and complete, written by actual engineers for technical users. A lot of effort probably went into writing them, although the quality of the paper/printing isn't there.

Overall, the elektronika calculators seem "inspired" from HP calculators, but they're definitely not a direct copy, and the engineers who designed them probably spent a lot of effort on the implementation given the resources that they had. In the west, all of the functionality of the 9 ICs of the mk-52 could've been fit onto a single chip (maybe more for RAM/the EEPROM). Later soviet designs used ICs which were almost a direct copy of the PDP-11 microarchitecture for their BASIC handheld calculators and also personal computers.

The calculators have a ton of weird idiosyncrasies, such as which registers are pre/post-incremented (or if they are modified at all) when doing indirect addressing. The meaning of comparisons is the exact opposite of HP calculators, and skip the next instruction if true. Running a program does not terminate digit entry, which I suppose might be useful if you for some reason wanted to add digits to the end of a number or add an exponent, but so far has just been kind of annoying. The mk-61 and mk-52 can work with hexadecimal numbers and do AND/OR/XOR/NOT, but there's no easy way to enter the values A through F. You can do operations on "Error", which is just a display for a large range of number exponents, and a lot of people apparently did to get weird displays for games and explore undocumented features on the calculators. Calculating x^y preserves y on the stack instead of dropping the stack down. You can continuously enter exponent digits and they'll correctly overwrite the old ones (I'm not sure if this was a feature since there's no backspace key, or to save ROM space).

In general it's like a weird alternate universe where someone who had seen a programmable HP calculator, but never used one extensively, tried to design a calculator.