Design a DC machine by moving sliders.
Generator or motor is the same machine read in two directions. Set the rating, the speed and how hard you are willing to work the iron and the copper, and the armature, the air gap, the slots, the poles and the field winding are designed live — with every step showing the formula it came from.
This is an interactive design studio and needs JavaScript to run. Here is what it does.
The two numbers that decide how big it is
A DC machine’s size is not chosen, it is computed — and it comes from exactly two decisions. The output equation Pa = Co·D²·L·n packs them into one coefficient: Co = π²·Bav·ac·10⁻³. Bav is how hard the iron is worked, ac is how much copper is crowded onto the armature surface, and together they fix D²L for a given power and speed.
Everything after that is consequence. Push Bav up and the machine shrinks but the magnetising current and the iron loss climb; push ac up and it shrinks but the armature runs hotter and armature reaction distorts the field harder. The studio draws the section from whatever those two give you, so the trade is something you watch rather than something you read about.
Why the air gap is the expensive millimetre
The gap is not set by clearance. It is set so the mmf it absorbs is a fixed fraction — about 0.55 — of the armature mmf per pole, because that is what stops armature reaction from dragging the field out of shape under load. Widen the gap and the field winding has to grow to push flux across it; narrow it and the machine becomes more sensitive to load.
That makes the gap the one dimension where magnetic, mechanical and economic limits all meet, and it is why this studio prints it to two decimal places while rounding the yoke to the nearest millimetre.
Every assumption is shown, not hidden
A DC design cannot start without assuming things that only exist at the end — the efficiency, the internal voltage drop, the field current fraction. The honest way to handle that is to assume, carry through, and then check. This studio lists every default it had to take with the page it came from, and tells you when the computed efficiency disagrees with the one you assumed, which means the design has not converged yet.
Where the book prints only a graph — Carter’s coefficient, the B–H curves — nothing is invented. Those stay inputs, and the figures that depend on them are reported as missing rather than guessed.
Common questions
How are the main dimensions of a DC machine chosen?
From the output equation, Pa = Co·D²·L·n, where the output coefficient Co = π²·Bav·ac·10⁻³ (Sawhney Eqn 9.2, p.493). The two loadings in it are the only real choices: Bav, the average flux density over the pole pitch (0.4–0.8 T, p.494), and ac, the ampere-conductors per metre of armature periphery (15,000–50,000 A/m, p.496). Everything about how big the machine turns out follows from those two and the speed.
How many poles should a DC machine have?
Sawhney fixes it in §9.12.1 from three limits at once: the frequency f = p·n/2 should stay around 25–50 Hz, the current per brush arm should not exceed about 400 A, and the armature mmf per pole should stay inside the band of Table 9.3 or the field distorts under load. More poles shortens the flux path and lightens the yoke, at the cost of a shorter pole pitch and more brush arms. The studio shows every pole count it weighed and why each passed or failed.
What sets the air gap in a DC machine?
The gap is chosen so that the mmf it absorbs is a definite fraction — usually 0.5 to 0.7, commonly 0.55 — of the armature mmf per pole (p.507), which is what keeps the field from being distorted by armature reaction. That gives lg = (0.5 to 0.7)·ATa / (800,000·Bg·Kg). Kg is the gap contraction factor: Sawhney prints Carter’s coefficient only as a graph (Fig 4.9) and assumes Kg = 1.15 when it is not supplied, and this studio can instead compute it in closed form from the slot geometry.
Why does the efficiency appear as an input when it is also an answer?
Because the armature current cannot be found without it, and the losses cannot be found without the current. So a figure is assumed to start, the design is carried through, and the real efficiency falls out at the end. If the two disagree the design has not converged, and the studio says so rather than printing the answer as though it had — put the computed figure back into the assumption and run it again.
Built and operated by hjLabs.in.