Wireless World calculator project

[Electronpusher0]

I hope this is the correct section for this post.

I have a recollection of Wireless World embarking on a year long project to build a desk calculator using TTL logic chips only to have Sinclair launch their pocket calculator at the time of the second or third installment. My memory says that WW decided to continue the project though already out of date.

I have checked on the American Radio History web site but can find nothing about this project.

Is my memory completely false?

Peter

[Craig Sawyers]

The only things I've trawled up were August to December 1967, a digital computer built using discrete germanium transistors, and in 1972 a short series of articles using a TI chip TMS1802 for a desk calculator. http://www.vintagecalculators.com/ht...ess_world.html

Maybe it was in Elektor or PW?

Craig

[WME_bill]

Wireless World calculator.
Originally published in WW September and October 1972; as a joint project with Advance Instruments. It was based upon the Texas TMS18082 I.C. introduced in 1971, so they were quick off the mark.
It became the Advance 162 series of calculators. I used them at one time and quite liked them. The top of the range was programmable, but less versatile than the HP models, which were aimed at the scientific / academic market.
They were aimed more at the office and desk top market, for which the large keyboard was very convenient. The Sinclair was a pocket calculator for you to carry around, s different market again.
I have the di-azo circuit diagrams of most of the models produced. The reliability was not as high as was really needed for the commercial market, and Advance withdrew them after a couple of years.
wme_bill

[Slothie]

I can remember seeing a ttl calculator while going through some magazines in the American radio history site.. I think it was Practical Wireless.

[dave_n_t]

Practical Electronics did a 'Digi-Cal' series, which I'm pretty sure was TTL based. And the author may have been a R. Coles (less sure about this). I'm also fairly certain that there was a letter to the mag complaining that the TTL calculator was superceded by the single-chip versions now available, but the editor responded that the educational value of the Digical series was worth it, even if you didn't make the thing.


hth


dave

[Slothie]

That was the one i was thinking of. Cant remember the dates but it was probably the mid 70s because that was the period I was looking at.

[Kala_12]

Just found it, It's practical electronics july 1972 page 267
Stuart.

[Electronpusher0]

Thank you everybody!
I found the WW project using the Texas chip but the one I remembered was definately TTL based.
Seems my memory was only faulty in the name of the magazine, Practical electronics, not WW.

The dates fit my recollection, if PE started a year long project in July 72 and sinclair launched their pocket calculator in Sept 72 that would be about installment 3.

Originally Posted by dave_n_t

I'm also fairly certain that there was a letter to the mag complaining that the TTL calculator was superceded by the single-chip versions now available, but the editor responded that the educational value of the Digical series was worth it, even if you didn't make the thing.
dave

That was as I recalled it.

I attach a quote from the editorial of October 1972

Peter

[Bill]

Just to throw a spanner in the works:
An apprentice friend at work built a calculator, no later than 1970. It was in a die cast box, thumbwheel switches to set the figures, Nixie tube display and definitely used TTL.
The devices used , eg 7441 Nixie driver , were all obtained from our production lines in the factory. There was tacit approval from management for him to build it to demonstrate the technology. It certainly worked though was a bit cumbersome to set each calculation. I am also fairly sure it was based on a WW design.
I am clear about the year because I hadn't been there long, spending a few weeks in each department, both he and I were at the time in the "Evaluation" section.

[Slothie]

I have a TTL/Nixie tube calculator made by Olympia which is on my list of things to restore to working condition. It was working 30 or so years ago but has spent most of the intervening time in a hot loft so will need all its capacitors replacing before i risk plugging it in!

[Philips210]

Hi.

The PE 'Digi-Cal calculator project ran from July 1972 to May 1973. What an interesting and educational project, and how we take LSI ICs for granted these days!
With the huge number of TTL ICs it must have cost a small fortune to build (and run) . I can't imagine too many were built. Deserves a place in the science museum I think.

Could be a great retro project to undertake now assuming that all components can be obtained. The power supply could easily be updated so that would be a small hurdle out of the way.

Regards,
Symon

[TonyDuell]

I find the Digi-Cal a curious design.

With mechanical calculators, it was difficult to transfer a number from one register (set of counting wheels) to another. So it made sense to have dedicated multiplication counter, accumulator, etc registers.

With electronic calculators it's relatively easy to move numbers between registers. The idea of dedicated multilication counters,etc, makes a little less sense.

And yet the Digi-Cal is very much of the first type of design. Interestingly, at about the same time as the articles were published, Hewlett-Packard were selling desktop calculators built from TTL (OK + firmware ROMs and Intel 1103 DRAMs) which were very much of the 'general purpose processor + ROM firmware to make it a calculator' design. I am thinking of the HP9810 and HP9820 machines. The processor (including the memory interface circuitry) is about 100 TTL chips (including 7 256*4 PROMs).

If I was going to build a calculator from TTL I would certainly think more of that sort of design.

[julie_m]

Originally Posted by TonyDuell

With electronic calculators it's relatively easy to move numbers between registers. The idea of dedicated multilication counters,etc, makes a little less sense.

When something really game-changing comes along, it often is not obvious just how tightly the game is bound up with the limitations of the old ways.

Look at the Iron Bridge, for example. It's held together with dovetails and mortice and tenon joints that were cast into the various sections. That's not the way you would make a metal bridge today (although it's probably exactly how you would make a plastic bridge..... though you'd expect better moulding tolerances today. Darby's original parts look similar from a distance, but corresponding ones could not simply be swapped: the give-or-take is tens or even hundreds of millimetres) but it was all anyone knew about in the 1770s.

And then consider early attempts to use transistors as though they were valves, and the breakthrough that came from realising they could be better used in ways valves simply couldn't be used.

It's sometimes hard to separate an abstract process from the physical tools used for carrying out that process. So maybe it was necessary to start by building electronic versions of mechanical calculators first, before realising that the mechanical method actually was hampered by restrictions that the electronic method did not require you to emulate. The crucial insight was that numbers can be stored in a way that does not force a particular rôle in a calculation; but up until then, augends, addends, minuends, subtrahends, multiplicands, multipliers, dividends and divisors had all been stored in specific ways.

[TonyDuell]

Sure, but the Digi-Cal was hardly the first electronic calculator. There had been computers around for about 20 years by the time that project was published and I would have thought that the differences between what was easy/sensible mechancally .vs. electronically would be well known and understood by then.

Incidentally, I've always found the distiction between 'augend' and 'addend' to be curious. Subtraction and division are not commutative, so it makes sense to have different names for the 2 inputs. Multiplication is commutative (at least for real numbers and complex numbers) but for actually doing it it can make sense to, say, take the number with the smaller number of digits as the multiplier. But I can't think of an adder (mechanical or electronic) or indeed a way of adding by hand which is not symmetical on the 2 inputs.

[Red to black]

Originally Posted by TonyDuell

Incidentally, I've always found the distiction between 'augend' and 'addend' to be curious. Subtraction and division are not commutative, so it makes sense to have different names for the 2 inputs. Multiplication is commutative (at least for real numbers and complex numbers) but for actually doing it it can make sense to, say, take the number with the smaller number of digits as the multiplier. But I can't think of an adder (mechanical or electronic) or indeed a way of adding by hand which is not symmetical on the 2 inputs.

Was this not because subtraction and division done electronically in binary at the simplest level still used an adder albeit with say the "two's compliment" so you would still need to define which input was which ?

[woodchips]

Thing with multiplication is that you have to store both number until it is complete, many cycles later. That is why you have separate registers.

If not found, look at John Wolffs web site johnwolff.id.au and the Madas and Marchant stripdowns.

[julie_m]

To perform the twos complement operation, as any fule kno, you flip the bits and add one.

In 8 bits, "flipping the bits" is equivalent to subtracting from 255; so flipping the bits and adding one is equivalent to subtracting from 256.

On the 6502 processor, the only difference between the ADC and SBC instructions is that SBC inverts each bit of the subtrahend/addend with a bunch of EOR gates as they are fed into the "B" input of the adder. So that takes care of flipping the bits. We still have to add one; but we can let the processor take care of that for us, by setting the carry flag with an SEC instruction before we start. Then we end up with A + (255 - B) + 1 = 256 + A - B. The excess of 256 means that if B is smaller than A, then the carry will be set, adding one also to the next more significant byte; if B is larger than A, the carry not being set borrows one.

(If you ever end up with the subtrahend already in the accumulator and need to evaluate something minus A, then you can use EOR #&FF, SEC and ADC..... it's only two extra bytes and two extra cycles.)

[Red to black]

Whoa! Julie
Way above my understanding , I basically only understood Tony's post due to a very 'rudimentary' basic digital course I did 30 odd "ahem" years ago at College that used the then obligatory MARC (?) training board