Microstrip Mitered Bend Calculator — Optimum Mitre (Chamfer)
Optimum mitre for a 90° microstrip bend from the Douville and James relation: the percentage of the corner diagonal to cut away, the cut depth along the diagonal and the chamfer length along each outer edge, ready to draw in a PCB layout.
Douville–James (1978) optimum-mitre relation, fitted to measured bendsAPI slug microstrip-mitered-bend
Formula
Douville & James, IEEE Trans. Microwave Theory Tech., MTT-26(3), pp. 175–181 (1978)
| Symbol | Quantity | Your value |
|---|---|---|
| W | Strip width (mm) | 3 mm |
| h | Substrate height (mm) | 1.6 mm |
| M | Optimum mitre, as a fraction of the diagonal (%) | 57.2 % |
| d | Diagonal of the unmitred corner, outer to inner (mm) | 4.243 mm |
| x | Depth cut away along the diagonal, from the outer corner (mm) | 2.426 mm |
| a | Chamfer length along each outer edge, from the old corner (mm) | 3.43 mm |
Assumptions
6 items
- 01The optimum mitre is Douville and James’s fit M = 52 + 65·exp(−1.35·W/h) %, the cut that measured least reflection on their bends; the calculator evaluates it directly.
- 02The bend is 90° between two strips of the same width W. The cut is perpendicular to the corner diagonal d = W√2, a depth x = M·d from the outer corner, leaving d − x to the inner corner, and each outer edge loses x√2.
- 03Neither εr nor frequency enters the result: εr is only checked against the εr ≤ 25 the fit was measured to, and a W/h below 0.25 is flagged as outside its measured range.
- 04Below W/h of about 0.225 the relation asks for more than 100 %, a cut through the inner corner, and the calculator refuses the input rather than return it.
- 05Only the geometry of the cut is given: the bend’s residual reflection, its electrical length and the line’s impedance are not computed.
- 06Synthesis solves the width (0.001–50 mm) or the substrate height (0.001–10 mm) numerically for a target mitre or remaining diagonal, or the width for a target chamfer. The material picker sets εr, which changes only the range check.
Worked example
The narrowest strip that keeps 0.1 mm of copper across its mitred corner
On 0.254 mm (10 mil) RO4003C a narrow strip’s optimum mitre cuts deep into the corner. Suppose the fabricator’s minimum copper feature is 0.1 mm: solve the narrowest width whose optimum mitre still leaves 0.1 mm between the cut and the inner corner, then lay out the chamfer.
| Step | Expression | Result |
|---|---|---|
| Width for d − x = 0.1 mm (solved) | W√2 × (1 − M(W/h)) = 0.1 mm; h = 0.254 mm | 0.2383 mm |
| Width-to-height ratio | 0.238255 / 0.254 | 0.938 |
| Optimum mitre | 52 + 65 × exp(−1.35 × 0.938) | 70.3 % |
| Corner diagonal | 0.238255 × √2 | 336.9 µm |
| Cut depth from the outer corner | 70.3/100 × 0.3369 | 236.9 µm |
| Chamfer along each outer edge | 0.2369 × √2 | 335.1 µm |
| Copper left across the corner, d − x | 0.3369 − 0.2369 | 100 µm |
How it works
A right-angle corner in a microstrip line adds metal at the outside of the bend. The extra area is excess capacitance to ground, which reflects part of the signal, more so as the frequency rises. Cutting off the outer corner, a mitre, removes that capacitance; cut too much and the bend turns inductive instead. The optimum lies between, and Pozar's Microwave Engineering (4th ed., section 4.6) describes mitring as the usual compensation when there is no room for a swept bend.
The Douville and James optimum
Douville and James measured symmetric microstrip bends over a range of widths and substrates and fitted the mitre that gives the lowest reflection (IEEE Transactions on Microwave Theory and Techniques, 1978):
Drawing it
The cut removes a right isosceles triangle from the outside of the corner. Along each outer edge it starts back from where the old corner was, so in a layout tool draw the plain corner and chamfer it with a 45° line whose ends sit that distance from the old corner. What remains between the cut and the inner corner, , is the narrowest copper the signal passes.
Validity
The relation is an empirical fit; it does not depend on or on frequency. Douville and James give it for and , from measurements spanning from 0.25 to 2.75 and from 2.5 to 25. The calculator flags inputs outside either bound. Below of about 0.225 the extrapolated mitre would exceed 100%, cutting past the inner corner and severing the line, and the calculator refuses it. The relation is for 90° bends between lines of equal width.
Practical tips
- Find the line width for your target impedance with the microstrip calculator first, then mitre with that width.
- Above W/h of 3 the optimum barely moves (53.13% at 3, tending to 52%), so wide lines can share one mitre.
- Where the board has room, a swept bend with a radius of at least three line widths avoids the corner discontinuity altogether.
- The mitre corrects the reflection, not the bend's electrical length; include the bend in the line length of a phase-critical path.
Common mistakes
- Measuring the mitre from the inner corner. M is the fraction of the corner diagonal cut away from the outer corner.
- Using the same 45° chamfer for every line. The optimum depends on W/h: a narrow line needs nearly the whole corner cut, a wide line little more than half.
- Mitring a bend between lines of different widths with this relation. It was fitted to symmetric bends of a single width.
- Ignoring the copper left at the inner corner on narrow lines. At high mitre percentages it can fall below the fabricator's minimum feature.
Frequently asked questions
5 questions
What is a mitered bend?
A right-angle microstrip corner with its outer corner cut off at 45°. Removing that copper cancels the excess capacitance of the square corner, which would otherwise reflect part of the signal at high frequencies.
What is the optimum mitre for a microstrip bend?
Douville and James found M = 52 + 65 × exp(−1.35 W/h) percent of the corner diagonal, cut from the outer corner. For a line as wide as its substrate is thick that is 68.85%; it rises to 98% for narrow lines and falls toward 52% for wide ones.
Does the optimum mitre depend on the dielectric constant?
No. The relation depends only on the ratio of strip width to substrate height. It was measured on substrates with relative permittivity from 2.5 to 25, and outside that range the calculator warns that it is extrapolating.
Is a mitred bend better than a curved bend?
A curved, swept bend with a radius of at least three line widths has almost no discontinuity, but it takes more board area. A mitred corner fits where a tight right angle is needed and, at the optimum, keeps the reflection low.
How do I draw the mitre in a PCB layout tool?
Draw the square corner, then chamfer the outside with a 45° line whose ends sit the chamfer length, x√2, back from the old corner along both outer edges. The calculator gives that length directly.
References
Experimental Study of Symmetric Microstrip Bends and Their Compensation · Microwave Engineering, 4th ed.
- Experimental Study of Symmetric Microstrip Bends and Their Compensation — R. J. P. Douville and D. S. James, IEEE Transactions on Microwave Theory and Techniques, vol. MTT-26, no. 3, pp. 175–181, March 1978 — the optimum-mitre relation
- Microwave Engineering, 4th ed. — David M. Pozar (2012), §4.6, pp. 209–210 — compensated microstrip discontinuities: mitred bends and why they work
The mitre is Douville and James's relation evaluated directly: 68.85% at W/h = 1, 98.38% at 0.25, falling to 52% for wide strips.
Shop Components
As an Amazon Associate we earn from qualifying purchases.