A 1974 part still worth understanding — and an interactive truth table, because a table you can drive teaches more than a table you can only read.
The 74xx47 takes a four-bit binary-coded-decimal input and lights the seven segments of a numeric display. Its outputs are active low and open collector: they sink current rather than source it, so a segment lights when its output is pulled down to ground. That makes the part a natural match for common-anode displays — and, as it happens, for filament tubes.
Codes 0000–1001 give the decimal digits 0–9. The remaining six codes,
1010–1111, are not errors and not blanks: they produce six defined
non-digit patterns, shown in the table below.
Three control pins override the decoding entirely — and they do so in a strict order of precedence that most published truth tables leave implicit.
| Dec | D | C | B | A | a | b | c | d | e | f | g |
|---|
Click the switches, click any row of the table, or use the keyboard:
0–9, arrow keys, and L / R / B
for the three control pins. ● = segment lit, i.e. that output held low. Note the classic font of
this part: the 6 has no top segment and the 9 no bottom segment —
a signature that distinguishes it at a glance from later decoders, which draw both digits with tails.
This is where a plain table misleads. The three control pins do not sit on equal footing:
1. BI wins over everything. Hold pin 4 low and every output goes off, whatever the BCD code — and whether or not lamp test is asserted.
2. LT, provided pin 4 is not held low. Pull LT low and all seven segments light, ignoring the BCD inputs. It is how you prove that no segment has failed.
3. RBI, only on zero. With RBI low and the code 0000, that zero is
suppressed instead of displayed.
And pin 4 is not an input at all, or not only. It is a wire-AND node serving as blanking
input and ripple-blanking output: when the ripple-blanking condition fires, the chip itself pulls
that pin low. Chain it to the RBI of the next stage and leading zeros vanish along the display —
089 becomes 89 — with no logic beyond a wire. The simulator above models the pin
this way, and reports which of the two roles it is playing at any moment.
The part is usually paired with LEDs, but it drives an incandescent display just as well. A Numitron is a seven-segment tube whose segments are tungsten filaments in a glass envelope: no preferred electrode, so the common electrode goes to the positive rail and the decoder sinks each filament to ground through a series resistor.
Filaments impose one constraint that LEDs do not. Tungsten of this size has a thermal time constant of roughly a tenth of a second — you can watch a digit fade rather than switch. Any refresh fast enough for the eye, above about 50 Hz, is far shorter than that: the filaments never cool between frames, and a multiplexed display would show every segment of every digit at once. Multiplexing, the standard economy of LED displays, is therefore not available here. Each tube needs its own decoder, permanently driven. The constraint is thermal, not electronic, and no amount of cleverness in the driver removes it.
Data from Texas Instruments, SDLS111, BCD-to-Seven-Segment Decoders/Drivers (March 1974, revised March 1988), function table T1, verified row by row against the reference tables used in my own workshop notes. The page is written from scratch: no code, markup or artwork is taken from any other source. Filament time-constant figures measured on the bench, August 2026, on an RCA DTF104B.