The controls, top to bottom as they appear in the panel. Dropdowns are selection-only unless noted; numeric boxes accept typed values.

The Layer Type sets the coil's winding architecture: how many separate conductors are wound, how many layers deep, and in what wire form. This choice determines the turns that sit in the gap — and therefore BL — as well as the coil's DCR and inductance, so it is one of the most consequential inputs on the panel. Which types appear in the dropdown is itself configurable — see Hiding Layer Types in the Data Manager section.

The families fall into a few groups. SVC (single voice coil) winds one conductor in 1–8 layers. DVC / TVC / QVC wind two, three, or four independent coils on the same former for multi-channel or series/parallel wiring. I/O (inside/outside) winds on both faces of the former wall — the ISV construction, covered in Inside/Outside (ISV) Winding. Multifilar (bifilar / trifilar / quadfilar) winds several conductors together as a set. Edge and Flat use rectangular wire for a high fill factor. Stepped windings vary the layer count along the height to shape the BL(x) curve.

The I/O layer types wind wire on both faces of the former wall, so both stacks sit in the magnetic gap and both are swept by the gap flux. The industry name is ISV — Interleaved Sandwich Voicecoil. It is not a dual voice coil: the inner and outer stacks are normally one continuous electrical winding.

Three things make the harder wind worth it:

1. Sensitivity. Active conductor in the gap roughly doubles for the same coil height — peak BL roughly doubles for the same coil mass.
2. Thermal. The inner stack faces the pole across air alone, with no former wall in the heat path, and the wire-to-steel proximity area roughly doubles.
3. Inductance flattening. The two stacks straddle the former, so inductance variation with excursion is more even.

What it does not buy: BL(x) linearity. The added symmetry is radial, not axial, so the BL curve shape is unchanged. The costs are a harder wind and more complex lead-out routing through the collar.

Why series only. A continuous wire runs up the inside face and back down the outside, with the winding sense reversing at the former edge so both stacks push the same way. Both lead-outs exit the same end of the former, one from each side, through the collar — two terminals per input, no internal joint. Splitting the stacks into independent inputs would need four lead-outs, two of them off the inner face with nowhere clean to exit; paralleling them is out because the unequal circumferences give unequal DCR and an uncontrolled current split between the stacks.

Geometry convention. For every I/O type, VCID is the inside diameter of the inner winding, not the former. Input 1 is wound against the former on both faces and carries full turns; each later input loses one wire diameter of winding height per face.

Why edge-wound stops at one layer per face. Edge-wound flat wire cannot be stacked radially: each flat winding acts as a knife working its way into the layer beneath, and winding-machine tension forces it in. That is why 2 Edge I/O is the edge-wound ceiling. The only route to more conductor on an edge-wound coil is multifilar — and multifilar adds turns per unit of axial height, not conductor across the radius, so it does not deliver the layer-count BL multiplier that round-wire I/O does.

The Layer Types catalog names every winding architecture SpeaD can build. This topic shows what they actually look like wound — turn by turn, at the wire size the solve arrived at.

Open one from the Winding Layout button on the Voice Coil tab toolbar. The window draws a half section through the coil: the former on the left, the winding built outward from it, every turn drawn at its true wire diameter. Because a coil is mostly empty space in the middle, the drawing places a standard break between the axis and the stack — every dimension shown is a true value, but the dead air is compressed so the winding stays legible at any wire pitch.

Reading the drawing:

• L1, L2 … label each layer, ordered outward from the former.
• The arrow beside a layer label is the winding-direction annotation — layers alternate. It marks direction only; SpeaD does not model winding sense.
• IN and OUT mark the lead ends. Odd layer counts finish at the top, even counts at the bottom.
• A final part-turn is drawn translucent. The layer table carries the exact count.
• E is the winding height, and the diameters below the coil are its true ID and OD.

The panel beside the drawing cross-checks the layout against the solve: turns and outside diameter are shown from both, and agreement is reported in green. If the two differ the line turns red and reads Layout and solve differ - re-run Calculate and reopen. Press Enter in any coil field to re-solve, then reopen the window; if it persists, report it. The window’s toolbar also has Copy to Clipboard and Print. The wire diameters it quotes, and the ones in the drawings below, are over the insulation; the Wire Size box on the panel shows the bare conductor.

The figures below were all drawn at one fixed reference coil — VCID 40.30 mm, Wind Height 25.00 mm and DCR held at 3.98 Ω — with only the layer type changed between them. That is not the start design a new tab opens with (4 SVC on a 38.55 mm former), so the numbers on your own coil will differ.

1 SVCsingle layer, single voice coil

The simplest case, and a useful one to read first because everything else is a variation on it. One layer wound directly onto the former, turns stacked along the winding height, the lead entering at the bottom and leaving at the top. With a single layer there is no nesting to see and no stagger between layers — the coil is exactly one wire thick, so its outside diameter is the former ID plus two former walls plus two wire diameters.

2 SVCtwo layers, single voice coil

One winding, wound up the former and back down again. Two things appear here that the single layer had no way to show. The turns nest: the second layer sits in the valleys of the first rather than directly on top of it, and the drawing staggers them accordingly. And the second layer carries one turn fewer than the first — 57.4 against 58.4 — because the half-wire setback at each end costs a turn per layer. That is the calculation’s own convention, drawn symmetrically at both ends.

With an even layer count the wire finishes at the end it started, so IN and OUT are both at the bottom.

4 SVCfour layers, single voice coil

The same winding continued for two more passes. The one-turn-per-layer loss now reads as a clear progression — 46.9, 45.9, 44.9, 43.9 — and the direction arrows alternate up, down, up, down. Note the mean diameter column climbing with each layer: outer layers are longer per turn, which is why turns and wire length are not proportional to each other.

6 SVCsix layers, single voice coil

At six layers the build is becoming a substantial fraction of the coil diameter: 40.30 mm inside, 47.14 mm outside — the winding is now more than 3 mm thick on each side. This is where the gap has to be designed around the coil rather than the other way about, and where the outer layers sit far enough from the pole that they see appreciably less field than the inner ones.

8 SVCeight layers, single voice coil

The deepest single-coil build in the catalog, and the clearest illustration of the trade. Eight layers reach 50.11 mm outside diameter on a 40.30 mm former — nearly 5 mm of winding per side — and the stagger between layers is plainly visible down the stack.

Read this series together and one thing stands out that no single drawing shows: the wire gets thicker as layers are added — Ø0.343 at one layer, 0.428 at two, 0.533, 0.604, and 0.658 at eight. Nothing was changed between these designs except the layer type. Because the solve is holding DCR at 3.98 Ω, more layers mean more turns and more wire length, so the only way to hold the resistance is a larger conductor. Layer count is not a free choice: it sets the wire size for you.

4 DVCfour layers, two independent voice coils

Two separate windings on one former, and the first thing the drawing tells you is how they are arranged: not interleaved, but stacked. Winding 1 takes the inner two layers, Winding 2 the outer two. Each winding is coloured, and its leads are labelled and coloured to match — IN1/OUT1 against IN2/OUT2.

That arrangement has a consequence worth understanding before you wire one up. The two coils do not occupy the same place in the gap. Winding 2 sits further out, at a larger mean diameter, so it sees a slightly weaker field than Winding 1. The layer table also shows it carries fewer turns — 112.2 against 116.2 — because every layer outward loses a turn to the setback.

Those two differences pull against each other, and the result is better than either suggests. Fewer turns at a larger diameter comes out to almost exactly the same length of wire: working the layer table through, the two windings differ by about 0.15% in conductor length, so their DC resistances are near enough identical. A designer wiring these in series or parallel can treat them as a matched pair, which is the whole point of a dual voice coil.

8 DVCeight layers, two voice coils of four layers each

The same two-winding arrangement taken deeper: four layers each. Winding 1 runs from the former out to layer 4, Winding 2 from layer 5 to the outside. With this much radial build the gap between the two windings’ working diameters widens — Winding 1 averages around 42 mm, Winding 2 around 46 mm — so the field difference between them is more pronounced than on the four-layer version.

The length matching still holds, but less tightly: about 0.6% apart here rather than 0.15%. Deep builds trade some of that natural balance for their extra turns.

6 TVCsix layers, three voice coils of two layers each

Three independent windings, two layers apiece, stacked outward from the former in the same way — Winding 1 innermost, Winding 3 outermost, each with its own colour and its own numbered lead pair.

This is the clearest case of the balancing act described above, because there are three windings to compare rather than two. Turn counts fall steadily outward — 116.8, 112.8, 108.8 — while mean diameters climb, and the resulting conductor lengths land within 0.1% of one another across all three. Three coils, three different turn counts, three different diameters, and effectively one resistance.

8 QVCeight layers, four voice coils of two layers each

Four independent windings on one former, two layers apiece — the deepest division in the catalog. Four colours, four numbered lead pairs, and the leader lines fanning out at the bottom of the drawing are the reason the layout view exists: on a real coil these four pairs emerge within a few millimetres of one another, and knowing which is which is a production problem before it is a design one.

The arithmetic is worth pausing on. Turn counts fall in an exact progression — 117, 113, 109, 105 — four fewer for each winding outward, two layers each losing one turn to the setback. Mean diameters climb correspondingly. Conductor lengths across all four windings spread by 0.46%.

Read the multi-coil group together and the pattern is consistent: two windings, three, or four, the same trade of turns against diameter keeps them within a fraction of a percent of one another on resistance. That is what makes multi-coil drivers wirable in series, parallel, or any combination without matching parts by hand.

The four Inside/Outside (ISV) types read differently from everything above, and it helps to know what to look for before opening one. The wire runs on both faces of the former, so the former wall is buried inside the winding rather than sitting underneath it, and the layer numbers no longer run in physical order across the page. Two conventions follow. The diameter at the bottom left is the inside diameter of the inner winding, not the former — the former diameter is an output, and it moves outward as inputs are added. And the Stack column in the layer table becomes meaningful: In for the inner face, Out for the outer.

2 SVC I/Oone winding, one layer on each face

The base ISV build, and worth reading against the plain 2 SVC above, because from the outside the two coils are almost the same object: 116.6 turns against 115.9, Ø0.429 wire against Ø0.428, 42.22 mm outside diameter against 42.13. What changed is where the conductor sits.

Two figures in the layer table carry the whole construction. Both layers show 58.3 turns — equal, where the stacked 2 SVC lost a turn on its second layer. On an I/O type the one-wire setback is charged per input, not per layer, and here there is only one input. And the two mean diameters, 40.73 and 41.79, are 1.06 mm apart: half a wire out to the inner layer’s face, 0.10 mm of former wall, half a wire in to the outer layer’s centre — 0.53 mm of radius, which doubles in the diameter column. Nothing nests across the former; there is a wall in the way.

The rest closes on the same arithmetic: 40.73 less one wire gives the 40.30 mm inner ID, 41.79 plus one wire gives the 42.22 mm OD. IN and OUT both leave the bottom, one off each face, which is the lead-out convention described in the ISV topic.

4 DVC I/Otwo windings, two layers each, one per face

Read the layer labels along the top of the drawing: L3 L1 L2 L4, not 1 2 3 4. The numbering follows the windings, the drawing follows the radius, and the two do not agree. Physically the stack runs L3 (Winding 2, inner face), L1 (Winding 1, inner face), the former wall, L2 (Winding 1, outer face), L4 (Winding 2, outer face). Winding 1 lies against the former on both sides and Winding 2 wraps around the outside of it on both sides — a sandwich, symmetric about the former.

That symmetry has a consequence the mean-diameter column shows plainly. Winding 2 gives up 0.77 mm of mean diameter on the inner face and gains 0.78 mm on the outer, so its two mean diameters sum to 40.73 + 43.34 = 84.07 — the same figure as Winding 1’s 41.50 + 42.56 = 84.06. The diameter term cancels.

On the stacked 4 DVC further up this page, the climb in diameter was exactly what compensated for the turns lost outward, and the two windings finished within 0.15% on conductor length. Here there is nothing left to do the compensating. The turn counts are 58.2 and 57.2, and the conductor lengths come out 1.7% apart — the turn difference passed straight through. That is a DCR mismatch between the two windings, and it grows with every input added.

6 TVC I/Othree windings, two layers each

The same construction with one more shell: L5 L3 L1 — former — L2 L4 L6, Winding 1 innermost against the former, Winding 3 outermost on both faces. Turns fall by exactly one per input outward — 58.0, 57.0, 56.0 — and the mean-diameter sums repeat the pattern: 85.62, 85.63, 85.63, identical to two decimal places across all three windings.

With the diameters cancelling, the spread follows the turn counts alone: 3.6% in conductor length between Winding 1 and Winding 3.

8 QVC I/Ofour windings, two layers each

The deepest I/O build in the catalog: four shells about the former, L7 L5 L3 L1 — former — L2 L4 L6 L8. Four lead pairs, all leaving the same end of the former, half of them off the inner face, all routed through the collar. That routing is the practical limit on this construction, not the winding.

Turns run 57.9, 56.9, 55.9, 54.9; mean-diameter sums run 87.19, 87.20, 87.19, 87.19. Conductor length between the innermost and outermost windings is 5.5% apart — which is one of the reasons the I/O types are wired in series rather than paralleled.

Read the four I/O drawings together and one number refuses to move: the wire size. Ø0.429, 0.430, 0.431, 0.432 across two, four, six and eight layers — where the stacked SVC series ran from Ø0.343 to Ø0.658 over the same range. The turn counts explain it. Every I/O input is exactly two layers, one per face, so each input arrives as very nearly a copy of the reference two-layer coil at around 116 turns, and holding 3.98 Ω asks for the same conductor every time. Adding inputs on an I/O type adds coils, not turns per coil.

What it does add is diameter: 42.22, 43.77, 45.33, 46.89 mm — a steady 1.56 mm per input, all of it outward, because the inner winding ID is the value you entered and does not move. The former travels outward with it, sitting at roughly 41.3, 42.0, 42.8 and 43.6 mm across the four builds.

The multifilar types work in the third possible direction. The layer-stacked multi-coil types divide the coil radially, the I/O types divide it about the former, and multifilar divides it axially: two, three or four conductors wound side by side up the same layer, sharing the winding height between them. The layer table announces it in two places — the Wind column reads a range (1-2, 1-3, 1-4) because every conductor is present in every layer, and the Stack column is blank because there is no inner or outer face to distinguish.

One number in these tables is worth checking by hand, because it explains the whole family. The turns shown for layer 1 are the winding height divided by the wire diameter, exactly: 25.00 ÷ 0.343 = 72.9 on the 2 Bifilar, 25.00 ÷ 0.301 = 83.0 on the 2 Trifilar, 25.00 ÷ 0.275 = 90.9 on the 2 Quadfilar. The Turns column is counting wire cross-sections, not turns of one conductor, and the conductors are sharing them. Each filar gets its 1/N share of every layer, which is why the filars in a multifilar set are matched exactly rather than approximately.

2 Bifilartwo conductors wound together, two layers

Two wires laid side by side up the former and back down again. In the drawing they alternate turn by turn all the way up the stack — orange, blue, orange, blue — rather than occupying separate regions of the coil, and a single IN/OUT pair is marked because the bundle enters and leaves together.

Against the plain 2 SVC above: 144.8 turns instead of 115.9, on Ø0.343 wire instead of Ø0.428, and a slightly smaller outside diameter — 41.80 mm against 42.13. More turns on thinner wire in a tighter envelope, but the turns are split between two conductors, so each one carries 72.4 — well under the single coil’s 115.9. The one-wire setback still applies per layer: 72.9 then 71.9.

2 Trifilarthree conductors wound together, two layers

Three wires in the bundle, so each layer divides three ways and the wire has to come down again to fit: Ø0.301. Total turns rise to 165.1, but per conductor that is 55.0. Outside diameter 41.64 mm.

2 Quadfilarfour conductors wound together, two layers

Four wires, Ø0.275, 180.9 turns total and 45.2 per conductor. Read the three two-layer multifilar drawings in order and the trend is clear: as filars are added, each conductor gets both shorter and thinner — 72.4 turns at Ø0.343, 55.0 at Ø0.301, 45.2 at Ø0.275 — which is what holding one resistance target on each conductor requires. Meanwhile the coil itself barely moves, and moves the wrong way: 41.80, 41.64, 41.54 mm. More conductors in the same two layers means a slightly smaller coil, not a bigger one.

4 Bifilartwo conductors wound together, four layers

This one is worth setting directly beside the 4 DVC drawing further up the page, because the two come out with the same numbers: 228.3 turns and 43.66 mm outside diameter on both. Same copper, same envelope, two independent conductors either way. Everything that differs is in how those conductors are distributed.

The 4 DVC gives Winding 1 the inner two layers and Winding 2 the outer two, and the two finish 0.15% apart on conductor length — close, but not equal, and they sit at different mean diameters so they do not see quite the same field. The 4 Bifilar puts both conductors in all four layers, half of each: 29.3, 28.8, 28.3, 27.8 turns apiece, 114.2 in total for each wire. The two are identical by construction — same turns, same mean diameters, same length, no residual mismatch at all — and each one is very nearly the reference two-layer coil repeated (Ø0.427 over 114.2 turns, against Ø0.428 over 115.9).

That is the reason to choose bifilar over DVC when the two windings must match: interleaving removes the mismatch instead of minimising it. What it costs is the DVC’s separation — the two conductors are wound together all the way, so they cannot be handled or terminated independently along the stack.

4 Trifilarthree conductors wound together, four layers

Three matched conductors, 261.1 turns total, 87.0 each, Ø0.374 wire, OD 43.27 mm. The comparison to draw is with the 6 TVC above, which also gives three independent windings: that one reaches 338.2 turns but needs 45.17 mm of outside diameter to do it, and its three windings land within 0.1% of one another rather than dead equal.

So the choice between them is not really about matching — both are good — but about turns against envelope. Stacking the three windings radially buys turns per winding and costs diameter; interleaving them axially holds the diameter down and costs turns.

4 Quadfilarfour conductors wound together, four layers

Four matched conductors, 286.5 turns total, 71.6 each, Ø0.342 wire, OD 43.03 mm. Layer turns run 73.1, 72.1, 71.1, 70.1 — and 25.00 ÷ 0.342 is 73.1, so the same arithmetic holds at four filars as at two.

Set this against the 2 Bifilar at the head of the group and something useful falls out. The wire is the same — Ø0.342 here, Ø0.343 there — and so is the turn count per conductor: 71.6 against 72.4. Two conductors in two layers and four conductors in four layers produce the same conductor, four times over instead of twice. Doubling the filar count halves each conductor’s share of a layer; doubling the layer count hands it straight back. What actually changed is the coil: 41.80 mm outside diameter becomes 43.03 mm, which is the price of the extra two conductors.

Against the stacked 8 QVC further up, which also delivers four windings: that one reaches 444.1 turns at 46.68 mm outside diameter, roughly 111 turns per winding. So the same trade as trifilar against 6 TVC, one step further along — radial stacking buys turns per winding and costs 3.6 mm of diameter; axial interleaving holds the build down and gives each conductor less.

6 Bifilartwo conductors wound together, six layers

294.7 turns total, 147.4 per conductor, Ø0.484 wire, OD 45.83 mm. Six layers of a two-wire bundle: each conductor now carries more than the reference coil does, and the wire has grown to suit.

6 Trifilarthree conductors wound together, six layers

337.4 turns total, 112.5 per conductor, Ø0.426 wire, OD 45.18 mm. Compare the 6 TVC higher up the page: 338.2 turns at 45.17 mm. Three windings, six layers, the same copper in the same envelope — interleaved here, stacked there.

6 Quadfilarfour conductors wound together, six layers

370.5 turns total, 92.6 per conductor, Ø0.389 wire, OD 44.78 mm. There is no six-layer four-winding stacked type to set against this one; the catalog has no way to divide six layers four ways radially, and interleaving does it without difficulty.

8 Bifilartwo conductors wound together, eight layers

349.8 turns total, 174.9 per conductor, Ø0.529 wire, OD 48.23 mm — against the 8 DVC’s 350.5 turns at 48.21 mm. The same pairing again, at eight layers.

8 Trifilarthree conductors wound together, eight layers

401.6 turns total, 133.9 per conductor, Ø0.466 wire, OD 47.30 mm. Eight layers do not divide evenly by three, and there is no stacked three-winding eight-layer type to compare it with.

8 Quadfilarfour conductors wound together, eight layers

The deepest multifilar build: 443.0 turns total, 110.8 per conductor, Ø0.425 wire, OD 46.70 mm. The 8 QVC reaches 444.1 turns at 46.68 mm. Four windings either way, and the two constructions land on the same coil.

All twelve multifilar types are drawn above, and read as a set they collapse into one rule worth carrying away. Divide the layer count by the filar count and you have the single-coil winding each conductor is getting. The 2 Bifilar is two copies of the 1 SVC — 72.4 turns each on Ø0.343 wire, against the 1 SVC’s 72.9 on Ø0.343. The 4 Bifilar and the 8 Quadfilar are both copies of the 2 SVC. The 8 Bifilar is two copies of the 4 SVC: 174.9 turns on Ø0.529, against 181.8 on Ø0.533. Each conductor falls a few percent short of its single-coil twin, because a conductor spread over more layers pays the one-wire setback more times and sits at larger mean diameters, but the correspondence is close enough to design with.

So the filar count and the layer count are not two knobs pulling against each other. The filar count is how many matched conductors come out of the coil; the ratio of layers to filars is what each of them is worth.

The other thing the set settles is the choice between interleaving and radial stacking. Wherever the catalog offers both — and it does at four combinations — they arrive at the same coil:

• 4 layers, 2 windings: 4 Bifilar 228.3 turns / 43.66 mm, 4 DVC 228.3 / 43.66.
• 6 layers, 3 windings: 6 Trifilar 337.4 / 45.18, 6 TVC 338.2 / 45.17.
• 8 layers, 2 windings: 8 Bifilar 349.8 / 48.23, 8 DVC 350.5 / 48.21.
• 8 layers, 4 windings: 8 Quadfilar 443.0 / 46.70, 8 QVC 444.1 / 46.68.

Same turns, same outside diameter, within a few tenths of a percent in every case. What differs is not the amount of copper but its arrangement, and that decides three things: the multifilar windings are matched exactly rather than to a fraction of a percent, because every conductor sees every layer; the stacked windings sit at distinctly different mean diameters and so see distinctly different field, where the interleaved ones share the same average position in the gap; and the stacked windings can be handled as separate regions of the coil, where a multifilar bundle is wound and terminated as one. Choose on those, not on the numbers — the numbers are the same.

Everything to this point has been round wire. The edge-wound and flat-wire types use rectangular ribbon, and three things change on screen. The drawing fills with rectangles rather than circles. The header line above the layer table reads Ribbon 0.517 x 0.259 mm (incl. insulation) instead of a single diameter — radial dimension first, axial second. And the Oval Ratio field on the Voice Coil panel comes alive: it is disabled for every round type and enabled here; a value below 1 is refused (field red, solve blocked), and left empty it is read as 1.

The ratio means the same thing in both families — wide face to narrow face, n:1 — and what changes is which way the ribbon is turned. Edge types stand it on its narrow face, so the wide dimension runs radially and the narrow one axially. Flat types lay it down, and the two swap. With Oval Ratio 2 the 1 Edge solves to 0.517 × 0.259 and the 2 Flat SVC to 0.259 × 0.519: the same ribbon, rotated a quarter turn.

The ribbon is not a separate thing the solver picks. It is the round wire the solve arrived at, reshaped at constant cross-sectional area to the ratio you asked for — which is why the manual can say the rotation pairs are the same coil, and why both dimensions are quoted on the same over-insulation basis as a round wire diameter. Set Oval Ratio to 1 and a flat or edge type returns the round wire it started from.

Two pieces of arithmetic run through all six drawings below. The turns in a layer are the winding height divided by the axial dimension — 25.00 ÷ 0.259 = 96.5 on the 1 Edge, 25.00 ÷ 0.519 = 48.2 per layer on the 2 Flat SVC. And the mean diameter climbs by exactly twice the radial dimension from one layer to the next, with nothing subtracted: ribbon does not nest, because there are no valleys between turns for the next layer to drop into. Round layers step by less than a full wire; these step by the full ribbon.

1 Edgeribbon on its narrow face, one layer

Ribbon 0.517 × 0.259, one layer, 96.6 turns, OD 41.53 mm. The whole coil is a single wrap: the narrow face lies against the former and the build outward is one ribbon width (0.517 mm a side), 40.50 mm to 41.53 mm.

That is the point of edge winding and also its limit. A single layer of ribbon puts more conductor in the gap than a single layer of round wire could, because the rectangle leaves no gaps between turns and no gap between the turn and the former — but there is nowhere sensible for a second layer to go, which is why the catalog offers deeper edge-wound builds only in the I/O form.

1 Edge BFtwo ribbons wound together on edge, one layer

Bifilar edge winding: two ribbons alternating up the single layer, 0.415 × 0.207 each, 120.6 turns total and 60.3 apiece, OD 41.33 mm. The Wind column reads 1-2 exactly as it does on the round multifilar types — both ribbons are in the one layer, sharing the winding height between them, so a narrower ribbon fits more turns in.

2 Edge I/Oribbon on edge, one layer on each face of the former

The way to get two layers of edge-wound ribbon: put one on each face rather than stacking them. Ribbon 0.648 × 0.324, 77.2 turns per face, 154.3 in total, OD 43.09 mm. The Stack column reads In and Out, and the drawing shows the former wall running down the middle of the winding, exactly as on the round I/O types.

The step from the inner mean diameter to the outer is 1.49 mm — half a ribbon out to the inner layer’s face, 0.10 mm of former, half a ribbon back — and both layers carry the same 77.2 turns, since the setback on an I/O type is charged per input and there is only one.

2 Edge BF I/Otwo ribbons on edge, one layer on each face

Both ideas at once: bifilar ribbon, one layer per face. 0.519 × 0.259, 96.4 turns per face, 192.7 in total, OD 42.58 mm. This is the deepest edge-wound build the catalog offers, and the ribbon it solves to is the same one the 1 Edge uses — four times the conductor in the same section, arranged two ways at once.

2 Flat SVCribbon laid flat, two layers

Ribbon 0.259 × 0.519 — the 1 Edge ribbon turned on its side — wound two layers deep. 48.2 turns in each layer, 96.4 in total, OD 41.54 mm.

Set that against the 1 Edge and the two are the same coil. 96.4 turns against 96.6; 41.54 mm outside diameter against 41.53. One layer of ribbon standing on edge and two layers of the same ribbon lying flat occupy the same section and hold the same conductor, because rotating a rectangle does not change its area and the coil has to reach the same resistance either way. The choice between them is a winding-shop choice, not an electrical one.

Note also that both layers carry 48.2 turns. There is no one-turn setback here, and there should not be: the setback on round wire comes from the second layer dropping into the valleys of the first, and flat ribbon has no valleys.

2 Flat BFtwo ribbons laid flat, two layers

Ribbon 0.208 × 0.416, two conductors sharing both layers, 60.1 turns per layer and 120.2 in total, OD 41.33 mm. The same relationship holds one step along: this is the 1 Edge BF ribbon rotated, and the two builds land on 120.2 turns against 120.6 at an identical 41.33 mm outside diameter.

4 Flat SVCribbon laid flat, four layers

Ribbon 0.325 × 0.651, four layers, 38.4 turns each, 153.7 in total, OD 43.10 mm. The rotation rule holds a step deeper: this is the 2 Edge I/O ribbon turned on its side, and the two builds arrive at 153.7 turns against 154.3 and 43.10 mm against 43.09.

4 Flat DVCtwo windings, two flat layers each

Ribbon 0.260 × 0.520, Winding 1 on layers 1–2 and Winding 2 on layers 3–4, 48.1 turns in every layer, 192.2 in total, OD 42.58 mm.

This is where flat wire parts company with round, and it is worth looking at closely. Every layer carries the same 48.1 turns — correct, because ribbon does not nest and so pays no setback. But that setback was doing something on the round types: the turns the outer winding gave up were what cancelled its larger diameter. Take it away and nothing cancels. Winding 1 sits at mean diameters summing to 82.04 mm, Winding 2 at 84.12, and the conductor lengths finish 2.5% apart.

The round 4 DVC managed the same job to 0.15%. So a flat DVC is not a drop-in equivalent: the two windings differ in resistance by a margin you can measure, and the outer one is the longer.

4 Flat BFtwo ribbons interleaved, four flat layers

The same ribbon, the same four layers, the same 192.2 turns and the same 42.58 mm outside diameter as the 4 Flat DVC above — and the mismatch is gone. Both ribbons run through all four layers, so both see the same diameters and the same length, exactly.

That makes this the sharpest case in the whole gallery for choosing interleaved over stacked. On round wire the two constructions were separated by 0.15%, which few designs would care about. On flat wire the gap is 2.5%, for no difference at all in copper or envelope. If a flat coil needs two matched windings, this is the one to wind.

It also equals the 2 Edge BF I/O by the rotation rule: 192.2 turns against 192.7, 42.58 mm against 42.58.

6 Flat SVCribbon laid flat, six layers

Ribbon 0.372 × 0.744, 33.6 turns in each of six layers, 201.6 in total, OD 44.96 mm. A single winding, so nothing to mismatch; the ribbon has grown wide enough that only 33.6 turns fit the 25 mm height.

6 Flat BFtwo ribbons interleaved, six flat layers

Ribbon 0.297 × 0.594, both conductors in all six layers, 42.1 turns per layer, 252.6 in total, OD 44.06 mm. Two matched conductors at 126.3 turns each, and no mismatch to report for the same reason as the 4 Flat BF.

6 Flat TVCthree windings, two flat layers each

Ribbon 0.261 × 0.522, Winding 1 on layers 1–2, Winding 2 on 3–4, Winding 3 on 5–6, 47.9 turns everywhere, 287.6 in total, OD 43.63 mm.

The mismatch that showed up on the 4 Flat DVC scales with the number of stacked windings and grows again on the 8 Flat DVC and 8 Flat QVC below. The three windings sit at mean-diameter sums of 82.04, 84.13 and 86.22 mm, so Winding 3 runs 5.1% longer than Winding 1 — against the round 6 TVC’s 0.1%. There is no three-conductor interleaved flat type to escape into, so if a flat coil needs three matched windings, the mismatch has to be designed around rather than wound out.

8 Flat SVCribbon laid flat, eight layers

Ribbon 0.409 × 0.819, eight layers, 30.5 turns each, 244.2 in total, OD 47.05 mm. Worth setting beside the round 8 SVC at the head of the gallery, which reaches 275.9 turns but needs 50.11 mm to do it. Ribbon leaves no gaps between turns or against the former, so eight flat layers build three millimetres less than eight round ones — the fill-factor argument, in one comparison.

8 Flat DVCtwo windings, four flat layers each

Ribbon 0.327 × 0.654, Winding 1 on layers 1–4 and Winding 2 on 5–8, 38.2 turns everywhere, 306.0 in total, OD 45.73 mm.

Four layers per winding puts the two centroids further apart than the 4 Flat DVC managed with two, and the mismatch widens accordingly: mean-diameter sums of 167.23 and 177.68 mm leave the windings 6.3% apart on conductor length, against 2.5% on the shallower build. Depth per winding costs as much here as winding count does.

8 Flat BFtwo ribbons interleaved, eight flat layers

Identical ribbon, identical layer table, identical 306.0 turns and 45.73 mm outside diameter to the 8 Flat DVC above — and the two conductors are exactly equal, because both run through all eight layers. The 6.3% gap closes to nothing for no change in copper or envelope at all.

8 Flat QVCfour windings, two flat layers each

Ribbon 0.261 × 0.522, four windings of two layers, 47.9 turns everywhere, 383.4 in total, OD 44.67 mm. The deepest stacked flat build, and the widest spread in the gallery: mean-diameter sums of 82.04, 84.13, 86.22 and 88.30 mm put Winding 4 7.6% longer than Winding 1.

Note that the ribbon is the same one the 6 Flat TVC solved to, and the first three windings sit at exactly the same mean diameters. Adding a winding to a stacked flat coil does not rearrange what is already there; it wraps another shell around the outside and widens the spread by another step.

Two things carry away from the rectangular-wire families.

The first is that an edge-wound coil of L layers and a flat coil of 2L layers built from the same ribbon are the same coil. Every pair the catalog offers confirms it: 1 Edge against 2 Flat SVC (96.6 turns / 41.53 mm against 96.4 / 41.54), 1 Edge BF against 2 Flat BF (120.6 / 41.33 against 120.2 / 41.33), 2 Edge I/O against 4 Flat SVC (154.3 / 43.09 against 153.7 / 43.10), and 2 Edge BF I/O against 4 Flat BF (192.7 / 42.58 against 192.2 / 42.58). Rotating a rectangle does not change its area, and the coil has to reach the same resistance either way, so the choice between edge and flat is a winding-shop question rather than an electrical one. The pairing stops at four flat layers, because the catalog’s edge-wound types stop at two.

The second is the one to watch. Because ribbon does not nest, the flat and edge types pay no one-turn-per-layer setback — every layer in every drawing above carries identical turns. On a single-winding coil that changes nothing, and the tighter packing is pure gain: the 8 Flat SVC reaches 47.05 mm where the round 8 SVC needs 50.11. But on a stacked multi-winding flat coil the missing setback removes the mechanism that kept the round types matched, and the windings separate by a margin that grows with both the number of windings and the depth of each:

• Two windings, two layers each — 4 Flat DVC, 2.5% apart.
• Three windings, two layers each — 6 Flat TVC, 5.1%.
• Four windings, two layers each — 8 Flat QVC, 7.6%.
• Two windings, four layers each — 8 Flat DVC, 6.3%.

The round counterparts run at 0.1% to 0.6%. Interleaving fixes it completely and costs nothing — the 4 Flat BF and 4 Flat DVC are the same 192.2 turns in the same 42.58 mm, and the 8 Flat BF and 8 Flat DVC the same 306.0 turns in the same 45.73 mm — but the catalog only offers interleaved flat types in bifilar form. Three and four matched flat windings cannot be wound this way, so on those the spread has to be designed around.

The stepped types are the last group, and the only ones that vary the winding along its height rather than across its thickness. The name reads ends / centre / ends as layer counts: a Stepped 4/2/4 carries four layers over the two end sections and two through the middle. The first layers run the full 25 mm; the ones above the centre count stop short, covering only the ends, and the section drawing shows the resulting shoulders.

The layer table records that in a way worth knowing how to read. On a full-height layer the turn count falls by one from the layer below, the usual nesting setback. On a layer that covers only the ends it falls by two — one for each of the two end bands, since the Turns column sums both. Spotting where the drop changes from one to two tells you where the step is without measuring anything: on the 8/4/8 below it happens between layers 4 and 5, which is exactly the 4 of its name.

Stepped 4/2/4four layers at the ends, two through the centre

Wire Ø0.495, 157.4 turns, OD 44.16 mm. Layers 1 and 2 run the full height at 50.5 and 49.5 turns — 25.00 ÷ 0.495 is 50.5, so layer 1 fills the winding height exactly. Layers 3 and 4 sit only over the ends, 29.7 and 27.7 turns, dropping by two apiece. Each end band works out around 7.3 mm tall, leaving roughly 10 mm of two-layer centre between them.

And this is what it is for. The Bl(x) curve leaves the single rounded peak a uniform coil gives and comes back with two crests and a shallow trough across rest — the extra conductor at the ends holding force up where a uniform coil would already be falling away.

Stepped 6/2/6six layers at the ends, two through the centre

Wire Ø0.540, 187.1 turns, OD 46.44 mm. The same two full-height layers — 46.3 and 45.3, and 25.00 ÷ 0.540 is 46.3 — with four end-only layers above them running 26.9, 24.9, 22.9, 20.9, every one of them two turns down on the last. The deepest step in the catalog relative to its centre: six layers at the ends against two in the middle.

Stepped 8/4/8eight layers at the ends, four through the centre

Wire Ø0.597, 229.0 turns, OD 49.22 mm — the largest stepped coil in the gallery. Four full-height layers at 41.8, 40.8, 39.8, 38.8, each one turn down; then four end-only layers at 19.9, 17.9, 15.9, 13.9, each two turns down. The change in that step size between layers 4 and 5 is the shoulder, and the end bands here are shorter than on the 4/2/4 — around 5.9 mm apiece — so the coil is closer to uniform over more of its height.

Read the Bl(x) plots as illustrations of the mechanism rather than as a comparison between profiles. They are not a controlled experiment: each coil solved to a different wire and to a different outside diameter, so the step profile is not the only thing changing between them. Judging how much trough or peak one profile gives against another needs the curves compared at a fixed envelope, not read off separate windows.

Stepped 2/4/2two layers at the ends, four through the centre

The step inverted: the extra layers go over the middle instead of the ends. Wire Ø0.464, 137.5 turns, OD 43.94 mm. Layers 1 and 2 run the full height at 53.8 and 52.8 turns; layers 3 and 4 cover the centre only, at 15.9 and 14.9.

Notice that those two fall by one turn, not two. The rule still holds — the drop is one per band, and a centre-loaded type has a single band where an end-loaded type has two. So on these the step cannot be spotted from the size of the drop; it shows instead as the sudden fall in absolute count, 52.8 straight to 15.9, which puts the centre band at about 7.4 mm of the 25.

The curve inverts with the winding. Where the end-loaded types gave two crests and a trough across rest, this returns a single smooth peak — force concentrated at rest and falling away sooner, because the extra conductor is all in the middle where the coil already sits.

Stepped 2/6/2two layers at the ends, six through the centre

Wire Ø0.492, 154.5 turns, OD 45.91 mm. Two full-height layers at 50.8 and 49.8, then four centre-only layers running 14.9, 13.9, 12.9, 11.9 — one turn down each, again the single-band signature. The centre band is about 7.3 mm, much the same as the 2/4/2; what changed is how much winding is piled on it.

Stepped 4/8/4four layers at the ends, eight through the centre

The deepest centre-loaded build, and the last drawing in the gallery. Wire Ø0.568, 206.7 turns, OD 48.79 mm. Four full-height layers at 44.0, 43.0, 42.0, 41.0 — 25.00 ÷ 0.568 is 44.0 — then four centre-only layers at 10.7, 9.7, 8.7, 7.7, every one down by a single turn, the centre-loaded signature again.

What sets it apart from the other two centre-loaded types is how narrow the loaded band is: 10.7 turns of Ø0.568 puts it at roughly 6.1 mm against the 7.3–7.4 mm of the 2/4/2 and 2/6/2. Four layers piled on six millimetres of height, which is the most concentrated winding the catalog will build.

The two halves of the stepped family do opposite things, and the Bl(x) figures show it without needing a number put on it. End-loaded profiles — 4/2/4, 6/2/6, 8/4/8 — move conductor out to where the coil travels, and the curve answers with two crests and a shallow trough across rest: less force at the middle, more of it held out at excursion. Centre-loaded profiles — 2/4/2, 2/6/2, 4/8/4 — do the reverse, piling turns where the coil already sits, and the curve comes back a single peak that is taller at rest and falls away sooner.

That is the whole design choice. End-loading trades peak force for a flatter Bl over travel, which is what a long-throw driver wants. Centre-loading trades travel for force at rest. Everything else about the coil — the wire, the turns, the outside diameter — falls out of the solve as usual.

The Solve for Wire Size (hold one) block has three rows — Wind Height (mm), DCR (Ω), and Wire Size (mm). On multi-coil layer types (DVC, TVC, QVC, Bifilar, Trifilar, Quadfilar, BF) the DCR row reads DCR/coil (Ω): the value is one winding on its own, and Re Series and Re Parallel in the Parameters column give the totals. Wire Size is always the computed output; it is read-only and has no radio. You hold exactly one of the other two with its radio, and SpeaD solves for the rest. The hold choice belongs to each design tab — changing it on one tab does not affect the others — and it is saved with the design. A new tab, or a design saved before this version, starts with Wind Height held.

Hold Wind Height. You fix the coil height, and the solver finds the wire diameter that hits your target DCR at that exact height. The winding height stays as typed. With Step set to Fine or Standard the solved diameter is then snapped to the nearest standard size, and the DCR moves to whatever that wire gives — with the height fixed, DCR is the value that gives way.

Hold DCR. You fix the target DCR, and the solver finds the wire diameter and re-solves winding height together to land on that DCR. Winding height becomes a computed output and is written back, bounded above by the former height. The value you type into Wind Height acts as the starting fill target. The wire is snapped to the Step series first and the height is then solved for the snapped wire, so the DCR is held in every step mode.

The label of the held row is shown in bold blue; the computed field updates in place. In both modes the DCR box then shows the DCR the coil actually achieves, which can differ slightly from the figure you typed because the solved diameter is snapped to a real wire size. The diameter search runs the real coil math directly, then snaps the result to a size per the Step mode.

Wire Size is the bare conductor diameter, not the enameled outside diameter — the IEC series it snaps to are likewise bare-conductor nominals. The enamel thickness per the Build grade is added for calculations, and DCR is computed on the bare conductor.

The Coil Calculated Values panel, directly below the Voice Coil panel on the Initial Design page. Five rows, updating as the solve completes.

Turn count cascades onward to the motor, where it drives BL, so a change to the coil updates the motor result too.

A first coil, from the start design. A new design tab opens with 4 SVC, VCID 38.55, former Aluminum 0.10 at height 40, collar None, Aluminium 1350 wire with PU-155 enamel at Grade 2 (heavy), nesting Standard (0.90), elongation 0% and step Fine (IEC R40). Change any of these to suit your coil.

1. Wind Height is held (the default). Type the winding height you want — the start design uses 28 mm — and the target DCR, 3.68 Ω at the start design.
2. When you press Enter or Tab, leave a field, or pick from a list, the solver runs: it finds the wire diameter that reaches the target DCR at that height, then snaps it to the Fine series. Wire Size resolves — at the start design to 0.500 mm. With the height fixed, the DCR box shows what that wire actually gives.
3. Coil Calculated Values and the Voice Coil tab fill in: OD, mass, Recommended FPID, turn count and wire length; Re Series / Re Parallel appear in the Parameters column. Turn count flows to the Motor tab and sets BL.
4. To design to a resistance instead, click the DCR radio and type the DCR. Now the winding height is solved as well and written back, and the DCR is held.