Why a linear spring suits a door at all
A torsion spring is about as simple as a machine element gets: wind it a turn, it gives back a fixed number of inch-pounds, wind it another turn and it gives back the same again. The torque it can hand over therefore falls in a straight line as it unwinds, which is exactly the wrong shape for lifting something of constant weight. A bucket in a well does not get lighter on the way up.
A sectional door does. Every panel that crosses the curve and picks up the horizontal track stops hanging on the cables and starts resting on its rollers, so the weight the cables carry falls roughly in proportion to how far the door has travelled. Two straight lines, both falling. That coincidence is the whole reason the arrangement works, and it is why a vertical lift door — which never turns over and never sheds any hanging weight — needs a drum whose radius shrinks as it winds, to bend the demand curve down by hand.
What the default door shows
A 155 lb door on 4 in drums asks the shaft for 310 lb-in with all of it hanging, and pays out 84 in of cable in 6.68 turns. A spring rate of 40 lb-in per turn wound to 7.5 turns gives 300 lb-in at the floor, which lands 5 lb short of the door: the door is 5 lb heavy closed. At the top the spring still has 0.82 turns on it, worth 33 lb-in, and the door has nothing left hanging, so the residual reads 16 lb the other way and the door wants to stay up.
The reason those two ends do not match is visible in the slopes. The demand falls 46.4 lb-in per turn on average; the spring gives back 40. Two lines with different slopes cross once and diverge either side of the crossing, which is why a door can feel right at waist height and wrong at both ends of its travel.
The drum is a lever you chose without noticing
Nothing in the torque figure depends on how many cables there are. Two cables each carry half the door at the same radius, so the two contributions add back to the whole door weight times the radius — swap to four cables and the shaft is asked for exactly the same thing, with a quarter of the load in each cable instead of half.
Drum diameter is different. It multiplies the torque directly: the same 155 lb door on 5 in drums asks for 388 lb-in rather than 310, a quarter more, while the wrap drops from 6.68 turns to 5.35. Bigger drums, more torque, fewer turns. That trade is fixed by geometry and is the same trade an opener sees from the other side of the shaft.
Where the wrap stops being a division sum
On a flat grooved drum every wrap sits at the same radius, so the turns are just the lift divided by one circumference and nothing is hiding. Put any taper on it and that division is wrong, because the radius is moving while the cable is winding. Cable paid out over n turns is pi times the quantity D times n plus t times n squared over two, so the turns come out of a quadratic rather than a division.
The size of the error is not small. Take the same 84 in of lift on a drum starting at 4 in and growing half an inch per turn: the honest answer is 5.08 turns, where the division sum says 6.68. Thirty per cent out, in the direction that matters, because it is also the figure that says whether the grooves run out. Set the taper field to zero and the page does the division; give it a taper and it solves the quadratic.
What is deliberately not here
There is no spring selection on this page, no wire size, no cycle life, no recommended wind count and no verdict on whether anything is balanced. Those are the supplier chart and the door documentation, and the rate and wind fields exist so you can put that chart in and see what the arithmetic makes of it, not so the page can hand you a part number.
Friction is missing too, and it is not a small omission. Roller and track drag always opposes the direction of travel, so it adds to the effort going up and subtracts going down. A door with a measured 5 lb imbalance can still take 20 lb to start moving. The gap between the two directions is the friction, and it is the one number here that only a scale on your own door can supply.
Questions people ask
How much torque does a garage door need at the shaft?
Weight times the winding radius, and that is the whole calculation. A 155 lb door on 4 in drums is 155 times 2, or 310 lb-in with all of the door hanging. The same door on 6 in drums is 465 lb-in. What the springs on a particular shaft actually deliver is a separate question with a separate authority — the chart the spring supplier publishes for those springs — and this page does not answer it.
Why does the number of cables not change the torque?
Because splitting the load does not move it. Two cables each carry half the door at the same radius, so their two moments add back up to the full weight times the radius. Four cables would give each one a quarter of the door and the shaft would still see 310 lb-in. Cable count changes the tension in a single cable, which is a cable question, not a shaft question.
Why is the door heavy at the floor and light at the top?
Two straight lines with different slopes. The torque a wound spring gives back falls by a fixed amount every turn it unwinds, while the weight hanging on the cables falls as sections roll onto the horizontal track. Unless those two slopes match, the lines cross once and separate either side of the crossing. At the defaults the demand falls 46.4 lb-in per turn and the spring gives back 40, which puts the door 5 lb heavy closed and 16 lb light open.
Can I work out drum turns by dividing the lift by the circumference?
Only on a flat drum, where every wrap sits at the same radius. On any tapered or cone drum the radius moves while the cable winds, so the cable paid out grows as a quadratic in the turns rather than linearly. On a 4 in drum growing half an inch a turn, 84 in of lift takes 5.08 turns and the division sum would tell you 6.68 — thirty per cent out.
Does this tell me whether my springs are right?
No, and it is built not to. It computes what the shaft is asked for and what a rate and wind count you type in would give back, and prints the difference. It states no spring rating, no wire size and no cycle life, and it issues no verdict on sizing or balance. A wound torsion spring stores enough energy to kill; that part goes to somebody who does it for a living.