
Using the HP35
Being the first of a new kind of calculator, the HP35 doesn't appear as polished and easy to use as today's calculators.
Here are a few points you may want to try out. Make sure RPN35 is set to Vintage Mode.
 Entering negative numbers
As the page from the manual shown at right explains, you may enter negative numbers by pressing CHS before or after you enter the first digit.
For example, if the display shows Pi = 3.141... and you press CHS, you see 3.141...
Now enter 2. The display shows 2, while the original, positive 3.141... has been pushed on the stack.
Press "+" and you get 1.141..., the sum of Pi and 2.
While convenient, this feature was also confusing, so HP dropped it in later models, never implementing it again.
 x^{y}
The HP35 was the only model to use x^{y} instead of y^{x}.
It made sense because there was no 10^{x} function. To get the antilog of the display value, you simply entered 10 and pressed x^{y}.
 Leading Zeros
When entering a number, leading zeros were not suppressed. For example, typing
00000.000, then EEX would be shown as 00000.0001 00. Pressing x<>y twice would then reveal the formatted number, 1. 04.
 No stack lift after STO
The RCL key or a keyboard entry, including EEX and Pi, did not raise the stack.
Numeric precision
The HP35's numeric precision was limited to 14digit BCD numbers. Errors exceeding the last significant digit were rare.
On the other hand, about 10,000 calculators produced in 1972 exhibited a ROM bug which caused a calculation like 2.02 LN e^{x} to return 2.
A few noteworthy "bad" results:
Calculation 
Correct 
HP35 
Error 
sin(5000)*10 
6.427876097 
6.427876027 
70E9 
ln(0.9995)*10000 
5.001250417 
5.001250000 
417E9 
1.000001^1E6 
+2.718280469 
+2.718281828 
1359E9 
tan(89.99)/1000 
+5.729577893 
+5.729569869 
8024E9 
Note: RPN35 SD does not emulate these errors.
Technology
As can be seen in the picture at right, the HP35 is basically built on five chips  a register and arithmetic chip, a timing chip, and three ROM (readonly memory) chips.
Each ROM chip holds 2560 bits (that's bits, not bytes), arranged in 256 words of 10 bits each. The entire program, including all mathematical functions, fits in 768 words! A total of 30,000 transistors were involved 
about 100,000 times less than the A8X chip inside the iPhone 6 Plus.
Data is shifted around serially, bitbybit. This is one reason for the rather low speed of operation. Another reason is the clock rate of 200 kHz. The design goal was to execute
transcendental functions in less than a second. This was achieved quite nicely:
Operation 
Time (in ms) 
Add, Subtract 
60 
Multiply, Divide 
100 
Square Root 
110 
Logarithmic and Exponential 
200 
x^{y} 
400 
Trigonometric 
500 
Register space was so scarce that trigonometric operations had to use the stack
register T. After completion of the operation, the content of stack register Z was
copied to T.
Miscellaneous
 Battery life was about three to five hours, mainly due to the LED display's power consumption. To extend operating time without turning the calculator off (thereby losing stack and memory contents), users would enter a single decimal point without digits. On the iPhone, this trick is probably of doubtful value.
 Just for fun, enter the following numbers, then turn the calculator upside down:
710.77345
57738.57734 EEX 40
 Evaluate the following expression:
(e.g.: 9 ENTER x 19 ENTER x 22 / + √x √x )
Compare the result with Pi.
 Popular Electronics, May 1972
Enjoy one of the first articles on the HP35.

Entering negative numbers
Inside the HP35
