When they say desk, they mean _desk_
October 5, 2020 7:10 PM Subscribe
@TubeTimeUS has a great twitter tear down explainer [threadreader] of a Friden EC-130. What is an EC-120? Why one of the first electronic desk calculators. Retailing for $2,100 in 1964 (~$20K today), it didn't feature any new fangled integrated circuits. Instead the entire thing ran on discreet components. And the memory, which stores 4 13 digit numbers, isn't implemented with transistors (it would have needed 520 of them) rather memory was implemented with an analogue, mechanical, magnetostrictive delay line via a coiled torsion wire. It was also the first calculator to use RPN.
More on Friden and the EC-120 at the Old Calculator Museum (including the service manual and an engineering report on the magnetostrictive delay line), and Vintage Calculators.
@TubeTimeUS previously.
More on Friden and the EC-120 at the Old Calculator Museum (including the service manual and an engineering report on the magnetostrictive delay line), and Vintage Calculators.
@TubeTimeUS previously.
I love RPN - so crazy that the world hasn't adopted it.
posted by PhineasGage at 7:54 PM on October 5, 2020 [6 favorites]
posted by PhineasGage at 7:54 PM on October 5, 2020 [6 favorites]
DECIMAL POINT
posted by genpfault at 7:56 PM on October 5, 2020 [1 favorite]
posted by genpfault at 7:56 PM on October 5, 2020 [1 favorite]
A friend has one of these and it's delightful. The solenoid keyboard lockout on error is so immediate that there's nothing like it in modern keyboards. It's way prettier than the 1969 Sharp Compet that I have.
posted by scruss at 8:16 PM on October 5, 2020 [1 favorite]
posted by scruss at 8:16 PM on October 5, 2020 [1 favorite]
Instead the entire thing ran on discreet components.
Well, I won’t tell if you don’t.
posted by zamboni at 8:21 PM on October 5, 2020 [14 favorites]
Well, I won’t tell if you don’t.
posted by zamboni at 8:21 PM on October 5, 2020 [14 favorites]
DECIMAL POINT
I laughed at this too until I realized how weird a decimal point is all alone on a keyboard with few keys.
Does anyone know how much an equivalent mechanical calculator cost at the time? Who would buy a $20k calculator? Were people laid off because of this device? Was there a room of people doing manual calculations or was it more organic?
It would be really cool to know the story behind this from a business perspective. It would be interesting to know who bought this and what business problem it solved. It seems obvious now but I'm sure this was not solving a new problem and there were already efficient non-digital solutions in place.
posted by geoff. at 8:48 PM on October 5, 2020 [2 favorites]
I laughed at this too until I realized how weird a decimal point is all alone on a keyboard with few keys.
Does anyone know how much an equivalent mechanical calculator cost at the time? Who would buy a $20k calculator? Were people laid off because of this device? Was there a room of people doing manual calculations or was it more organic?
It would be really cool to know the story behind this from a business perspective. It would be interesting to know who bought this and what business problem it solved. It seems obvious now but I'm sure this was not solving a new problem and there were already efficient non-digital solutions in place.
posted by geoff. at 8:48 PM on October 5, 2020 [2 favorites]
I'm sure this was not solving a new problem and there were already efficient non-digital solutions in place
Mechanical desktop adding machines had attained a high degree of speed and reliability by that time due to the advent of electric motors, but they were pretty damn noisy. The silent operation of this would have made it a flashy desktop flex for the executive who still loved to crunch numbers.
posted by CynicalKnight at 9:25 PM on October 5, 2020 [4 favorites]
Mechanical desktop adding machines had attained a high degree of speed and reliability by that time due to the advent of electric motors, but they were pretty damn noisy. The silent operation of this would have made it a flashy desktop flex for the executive who still loved to crunch numbers.
posted by CynicalKnight at 9:25 PM on October 5, 2020 [4 favorites]
Does anyone know how much an equivalent mechanical calculator cost at the time?
There are a few prices of contemporary mechanical desktop calculators ranging from $100-$300 on the Vintage Calculators site. It should be noted though that the comptometer style mechanical persisted into the 70s when electronic calculators were becoming price competitive because a skilled operator was much faster with them because all digits of a number could be entered at the same time.
Even at $200 a calculator was a significant investment. For perspective a brand new base model Rambler listed for $1953; the new in 64 Mustang was only $2,368 for the straight six coupe. 5 year old cars would sell for less than $500.
It would be really cool to know the story behind this from a business perspective.
The EC-130 was mostly a halo product though they were practically silent which was a big improvement over the mechanical models -- imagine a accounting office with a dozen mechanical calculators going, I'd bet today hearing protection would be mandatory.
The real impetus though was the writing on the wall that mechanicals were very shortly to be obsolete. In fact by the mid 70s, a mere 10 years later, practically no one was buying new mechanical calculators. It was a massive technological shift in a blindingly short period of time. I think people, like me, who didn't live through the leveraging of the transistor have trouble groking how rapid and fundamental the shift was.
posted by Mitheral at 10:20 PM on October 5, 2020 [2 favorites]
There are a few prices of contemporary mechanical desktop calculators ranging from $100-$300 on the Vintage Calculators site. It should be noted though that the comptometer style mechanical persisted into the 70s when electronic calculators were becoming price competitive because a skilled operator was much faster with them because all digits of a number could be entered at the same time.
Even at $200 a calculator was a significant investment. For perspective a brand new base model Rambler listed for $1953; the new in 64 Mustang was only $2,368 for the straight six coupe. 5 year old cars would sell for less than $500.
It would be really cool to know the story behind this from a business perspective.
The EC-130 was mostly a halo product though they were practically silent which was a big improvement over the mechanical models -- imagine a accounting office with a dozen mechanical calculators going, I'd bet today hearing protection would be mandatory.
The real impetus though was the writing on the wall that mechanicals were very shortly to be obsolete. In fact by the mid 70s, a mere 10 years later, practically no one was buying new mechanical calculators. It was a massive technological shift in a blindingly short period of time. I think people, like me, who didn't live through the leveraging of the transistor have trouble groking how rapid and fundamental the shift was.
posted by Mitheral at 10:20 PM on October 5, 2020 [2 favorites]
Re. advantages, I was wondering if it was faster and ergonomically easier to use (less key pressure??), especially with touch typing and entering large amounts of data.
posted by carter at 4:02 AM on October 6, 2020
posted by carter at 4:02 AM on October 6, 2020
Re. advantages, from on of the linked articles:
posted by carter at 6:22 AM on October 6, 2020 [1 favorite]
While the Sumlock Comptometer machines were not very sophisticated, they clearly demonstrated that an electronic calculator is vastly faster very quiet, and required no adjustment. Electro-mechanical calculators were noisy, slow, and had lots of moving parts that required regular maintenance to continue to operate properly. Relay calculators, while less noisy, and faster than electro-mechanical machines, also had moving parts (relays rely on mechanical movement to open and close switch contacts), and also had myriad switch contacts that wore over time and required periodic maintenance and adjustment. On top of this, relays take up a lot of space, use a lot of power, and are relatively expensive.Ah of course: the mechanical calculators experienced wear and mechanical forces, went out of whack, and required periodical tuning.
posted by carter at 6:22 AM on October 6, 2020 [1 favorite]
Mechanical calculators were noisy, especially when playing the Friden March.
posted by Monochrome at 6:35 AM on October 6, 2020 [6 favorites]
posted by Monochrome at 6:35 AM on October 6, 2020 [6 favorites]
Yes, RPN. Let's everybody go and learn us Starting FORTH (worth it for the pictures alone).
Raises hand and joins the club of the few, the proud, the learned RPN calculators before the more normal calculators which are weird.
posted by zengargoyle at 3:36 PM on October 7, 2020 [1 favorite]
Raises hand and joins the club of the few, the proud, the learned RPN calculators before the more normal calculators which are weird.
posted by zengargoyle at 3:36 PM on October 7, 2020 [1 favorite]
This is really neat!
The magnetorestrictive delay line is fascinating and new to me. I'm surprised that it was mass-produced and worked. (If I were making the thing, I'd have been tempted to spend an extra thousand dollars wrapping high index microstrip around for hundreds of meters.)
I received a working Sharp El-8 as a graduation gift. I opened it up and poked around. . . but, in the 8 years between this calculator and mine, custom ICs, cheap memory, and small segmented displays happened. It's still amazing what you can do with discrete components and a few silicon packages. But, a decade makes quite a difference.
posted by eotvos at 4:19 AM on October 8, 2020
The magnetorestrictive delay line is fascinating and new to me. I'm surprised that it was mass-produced and worked. (If I were making the thing, I'd have been tempted to spend an extra thousand dollars wrapping high index microstrip around for hundreds of meters.)
I received a working Sharp El-8 as a graduation gift. I opened it up and poked around. . . but, in the 8 years between this calculator and mine, custom ICs, cheap memory, and small segmented displays happened. It's still amazing what you can do with discrete components and a few silicon packages. But, a decade makes quite a difference.
posted by eotvos at 4:19 AM on October 8, 2020
Also, I've never understood the use for RPN. Making what you type into the calculator look like what you write on the page seems well worth the extra key-presses to me. (I'm among the generation who first had multi-line graphing calculators in high-school, so I may be missing something.)
posted by eotvos at 4:26 AM on October 8, 2020
posted by eotvos at 4:26 AM on October 8, 2020
→
Yup, that's a good reason, but RPN worked in the days of tiny calculator memories. The HP-35 had the equivalent of 56½ bytes of register storage and 960 bytes of ROM, and managed to be a useful scientific calculator. The Sinclair Scientific managed with less than half that, but it was famously inaccurate. The Friden basically sang into a wire, and remembered what came out the other end. We don't have wires long enough or singers loud enough to remember full algebraic input.
So the "more efficient" thing is a powerful draw to us engineers. Algebraic input pains me, and I can never quite remember what each calculator wants. For example, to calculate φ — (√5 + 1) / 2, or ~1.618 — it takes the following:
The Casio fx is closest to what you'd write, but while keying √ 5 is more logical, it's not the old calculator way that I learned (5 √, like the Film Card does it). HP sponsored lots of surveys to prove that users made fewer mistakes with RPN, but I've never seen a compelling reason beyond the fewer keystrokes one. Apart from no algebraic user will ever “borrow” your RPN calculator …
posted by scruss at 11:10 AM on October 10, 2020 [3 favorites]
Making what you type into the calculator look like what you write on the page seems well worth the extra key-presses to me
Yup, that's a good reason, but RPN worked in the days of tiny calculator memories. The HP-35 had the equivalent of 56½ bytes of register storage and 960 bytes of ROM, and managed to be a useful scientific calculator. The Sinclair Scientific managed with less than half that, but it was famously inaccurate. The Friden basically sang into a wire, and remembered what came out the other end. We don't have wires long enough or singers loud enough to remember full algebraic input.
So the "more efficient" thing is a powerful draw to us engineers. Algebraic input pains me, and I can never quite remember what each calculator wants. For example, to calculate φ — (√5 + 1) / 2, or ~1.618 — it takes the following:
- HP 48 (RPN): 6 keystrokes
5 √ 1 + 2 ÷ - Casio fx-115MS (algebraic): 9 keystrokes
( √ 5 + 1 ) ÷ 2 = - TI-83 Plus (algebraic): 11 keystrokes (and unmatched brackets because TI)
( 2nd √ 5 ) + 1 ) ÷ 2 Enter - Casio Film Card SL-760c (traditional "slide rule" calculator): 8 keystrokes
5 √ + 1 = ÷ 2 =
÷÷ calculates the square root (slowly!) and there's a fair bit of swapping and adding to memory locations because it's only got a small amount of magnetic core memory.)The Casio fx is closest to what you'd write, but while keying √ 5 is more logical, it's not the old calculator way that I learned (5 √, like the Film Card does it). HP sponsored lots of surveys to prove that users made fewer mistakes with RPN, but I've never seen a compelling reason beyond the fewer keystrokes one. Apart from no algebraic user will ever “borrow” your RPN calculator …
posted by scruss at 11:10 AM on October 10, 2020 [3 favorites]
For me, RPN goes something like this:
(( 3 + 4 ) * 3) is the in-order traversal of a binary tree and it takes the parenthesis (or implementing the PEMDAS to do it right, or other complexity.
The pre-order traversal of the tree gives you a LISP. (* (+ 3 4 ) 7) or similar which is equally as complex.
The post-order (RPN) gives you 3 4 + 3 * which is almost trivial to implement with little complexity or state. If it's a number it goes on the stack. If it's an operator it takes numbers from the stack, does the calculation, and puts the result back on the stack. Easy Peasy.
The thing RPN does is make you re-think the equation into the bits and pieces yourself, the order of the operations is whatever you type in, every number goes on the stack, every operator gets executed immediately. It's that key little thing that inputs to and outputs from operators/functions all come from and go to the stack.
Explore the dark side... try some Forth or Postscript or play with RPN calculators.
posted by zengargoyle at 3:40 PM on October 10, 2020 [2 favorites]
(( 3 + 4 ) * 3) is the in-order traversal of a binary tree and it takes the parenthesis (or implementing the PEMDAS to do it right, or other complexity.
The pre-order traversal of the tree gives you a LISP. (* (+ 3 4 ) 7) or similar which is equally as complex.
The post-order (RPN) gives you 3 4 + 3 * which is almost trivial to implement with little complexity or state. If it's a number it goes on the stack. If it's an operator it takes numbers from the stack, does the calculation, and puts the result back on the stack. Easy Peasy.
The thing RPN does is make you re-think the equation into the bits and pieces yourself, the order of the operations is whatever you type in, every number goes on the stack, every operator gets executed immediately. It's that key little thing that inputs to and outputs from operators/functions all come from and go to the stack.
Explore the dark side... try some Forth or Postscript or play with RPN calculators.
posted by zengargoyle at 3:40 PM on October 10, 2020 [2 favorites]
« Older “Want to go throw some dry ice in the River Styx?” | Penguin of the year 2020: vote early, vote often Newer »
This thread has been archived and is closed to new comments
posted by Dr. Twist at 7:39 PM on October 5, 2020 [7 favorites]