Number Representation Mistakes: Fix Signed Values, Complements and Overflow

The same eight bits can represent 234, −106, −21 or −22. Learn which convention controls the answer, how to preserve values when widening, and when carry-out does not mean signed overflow.

KnowledgeGate Team

Exam prep & CS education

Updated 4 Nov 20265 min read
Number Representation Mistakes: Fix Signed Values, Complements and Overflow

You can convert every binary digit correctly and still get the wrong answer by choosing the wrong signed convention. The same bits can mean different values, so calculation must follow interpretation. These worked examples extend the Digital Logic notes with a small set of checks to make before calculating: identify the representation, fix the width and choose the correct arithmetic rule.

1. Reading the bits before choosing the representation

Comparison table showing how 11101010 has four different decimal values under unsigned and signed conventions.

A leading 1 does not automatically mean a negative number. That shortcut fails whenever the pattern is unsigned, and even signed conventions disagree about its value.

Take the eight-bit binary pattern 11101010. All values in the final column below are decimal.

Convention

Working

Value

Unsigned

128 + 64 + 32 + 8 + 2

234

Sign-magnitude

Sign bit is 1; magnitude is 64 + 32 + 8 + 2

−106

One's complement

Invert to 00010101; magnitude is 16 + 4 + 1

−21

Two's complement

−128 + 64 + 32 + 8 + 2

−22

The bit pattern has not changed. Only its interpretation has changed. In sign-magnitude, the first bit supplies the sign; in two's complement, it has a negative positional weight.

Corrective habit: write “base 2, width 8, convention …” above your working. Without a stated convention, you cannot infer a signed value from the string alone.

2. Giving fractional bits the wrong weights

Integer place-value habits can mislead you after the radix point. Binary fractional columns have weights 2⁻¹, 2⁻², 2⁻³, meaning one-half, one-quarter and one-eighth.

For 101.101₂, work from the weights:

101.101₂ = 4 + 1 + 1/2 + 0/4 + 1/8 = 5.625₁₀.

Check the fractional part by repeated multiplication:

  • 0.625 × 2 = 1.25: record 1, retain 0.25.

  • 0.25 × 2 = 0.5: record 0, retain 0.5.

  • 0.5 × 2 = 1.0: record 1, stop at remainder zero.

Read those recorded bits top to bottom, giving .101₂. Reversing them is the mistake: integer division remainders are read bottom to top, but fractional multiplication bits are not.

Label the columns first. For the broader conversion method, revisit number representation and complements.

3. Taking a complement without fixing the width

Two-lane diagram contrasts 8-bit and 4-bit two's-complement negation and shows sign extension preserving −5.

“Invert and add one” needs a width. Inverting four bits and inverting eight bits are different operations.

To represent −5 in eight-bit two's complement:

  1. Write +5 as 00000101.

  2. Invert all eight bits: 11111010.

  3. Add 1: 11111011.

At four bits, the same procedure takes 0101 to 1010, then 1011. But padding that result with zeros gives 00001011, which means +11 in eight-bit two's complement. You have changed the value.

Instead, sign-extend the four-bit two's-complement pattern 1011: copy its leading 1 into the new positions to obtain 11111011. Its value remains −5 because 251 − 256 = −5.

Unsigned binary 1011 is +11 and correctly widens to 00001011. Establish the width before complementing; establish the convention before widening.

4. Assuming every signed range is symmetric

Thinking “one sign bit, seven value bits” suggests equal positive and negative endpoints. That works for sign-magnitude, but it misdescribes two's complement.

For eight-bit integers:

Representation

Decimal range

Zero encodings

Unsigned

0 through 255

One

Sign-magnitude

−127 through +127

Two

One's complement

−127 through +127

Two

Two's complement

−128 through +127

One

The n-bit two's-complement range is −2^(n−1) through 2^(n−1)−1. Thus, eight-bit 10000000₂ represents −128, while 01111111₂ represents +127.

Try negating the minimum: invert 10000000 to 01111111, then add 1. You get 10000000 again. The intended answer, +128, is outside the range, so this is overflow, not a valid positive result. Check the range before accepting a negation.

5. Using carry-out as the signed-overflow test

Unsigned and signed addition use the same bit addition. Their overflow tests answer different questions, which makes carry-out an easy trap.

For eight-bit two's-complement addition:

01111111 + 00000001 = 10000000

The operands are +127 and +1. Their mathematical sum is +128, but the stored result represents −128. There is signed overflow even though there is no carry-out from the eighth bit.

Now compare:

11111111 + 00000001 = 1 00000000

Under two's complement, this is −1 + 1 = 0. Discard the ninth bit, leaving 00000000; no signed overflow occurs. Under unsigned interpretation, the same addition is 255 + 1 = 256, which exceeds the eight-bit maximum and does overflow.

For signed addition, two operands with the same sign producing a stored result with the opposite sign signal overflow. Mixed-sign addition cannot overflow: the sum stays between the operand values.

Write the operand signs before adding, then compare the result's sign. Apply that distinction in the two’s complement practice questions.

6. Carrying the one's-complement rule into two's-complement subtraction

Both methods complement the subtrahend, but their carry rules differ. Mixing them changes an otherwise correct answer.

For four-bit two's-complement subtraction, calculate 5 − 3 by encoding −3 as 1101:

0101 + 1101 = 1 0010

Discard the carry-out. The retained result is 0010₂ = 2₁₀. Adding the carry back would incorrectly give 0011₂ = 3₁₀.

For four-bit one's-complement subtraction, encode −3 by inverting 0011 to 1100:

0101 + 1100 = 1 0001

Here, add the end-around carry to the retained bits: 0001 + 0001 = 0010, again giving 2.

Write the representation name beside the subtraction before selecting a carry rule. Then check against the decimal calculation, 5 − 3 = 2.

7. Finish with a representation check, not another guess

Before accepting an answer, check four things:

  • Base: are the integer and fractional weights correct?

  • Width and convention: how many bits, and which interpretation?

  • Range: can the mathematical answer fit?

  • Arithmetic rule: should you discard carry, add it around or test signed overflow?

For example, 10000000 means 128 as eight-bit unsigned binary and −128 as eight-bit two's complement.

Redo one example from each mistaken rule with the convention written down, then check the decimal result. For structured study beyond these corrections, the Digital Electronics course is an optional next step.