Convert Decimal to BCD

Binary-Coded Decimal stores each decimal digit as 4 bits (nibble). 57 → 0101 0111, not 111001 (which would be 57 in pure binary). BCD is less compact but avoids binary-to-decimal conversion at display time – ideal for 7-segment LED clocks, calculators, and financial systems where every displayed digit must match the stored value exactly.

Packed stores two digits per byte – the high nibble is the first digit, the low nibble is the second: 57 = 0x57. Unpacked (also called “zoned”) uses one digit per byte with the high nibble zeroed or used for a zone code: 57 = 0x05 0x07. Packed is half the size but needs alignment; unpacked is easier to manipulate byte-by-byte.

COMP-3 is IBM’s packed decimal format used in COBOL since the 1960s. Digits are packed two per byte, and the LAST nibble encodes the sign: C (hex 12) for positive, D (13) for negative, F (15) for unsigned. So -42 → 04 2D. This way the sign travels with the number without needing a separate flag, which saves space in fixed-width mainframe record formats.

Packed BCD packs two digits per byte, so an odd-digit number like 123 doesn’t fit cleanly. Auto-pad prepends a zero nibble: 0 1 2 3 → 0x01 0x23. Without auto-pad the tool warns you – some hardware formats require a specific alignment convention and this toggle lets you see the raw un-aligned output to understand the problem.

No practical limit. The logic uses BigInt for precision, so a 100-digit decimal encodes correctly – each digit is independent in BCD. The standard parseInt would cap at 2^53 ≈ 16 decimal digits, which is why the naive version of this tool used to break silently at that boundary.

Because 4 bits can encode 16 values (0-15), but BCD only uses 10 of them (0-9). The 6 “wasted” codepoints (1010-1111) are illegal in BCD. A 100-digit number takes 400 bits in BCD vs ~332 bits in pure binary – about 17% overhead. You trade that space for trivial digit-by-digit display.

Not directly – standard BCD is integer-only. Fixed-point BCD conventions exist (implicit decimal point at a known position), and “packed decimal” in COBOL PIC clauses encodes the scale separately. If you need fractions, store the scale as metadata and BCD-encode the scaled integer value.

Real-time clocks (RTCs – DS1307, PCF8563 chips store hours/minutes/seconds as BCD), financial systems where decimal-exactness is required (payment processing, accounting), embedded 7-segment display drivers, and legacy COBOL systems that haven’t migrated off COMP-3. Modern CPUs have largely deprecated BCD instructions (x86-64 dropped them entirely).

This tool goes decimal → BCD: you give a number, it packs it. The BCD-to-hex sibling takes a hex string (like 0x4527) and interprets each nibble as a BCD digit. Round-trip: encode 4527 here in packed mode → get 0x45 0x27, then feed that to the BCD-to-hex decoder and you get back 4527.

Yes. All arithmetic happens in your browser via JavaScript and BigInt. No upload, no tracking, no account. Load the page once and the tool keeps working offline indefinitely.