50 ohm on FR4

Feb 02, 2025 Last reply: 1 year ago 7 Replies

Grok reckons to make a 50 ohm tx line on FR4 I would need a Trace Width of approximately 2.73 mm and a Trace Spacing of at least 2.73 mm. Does that sound about right?



Here's the full response:



For a 50 ohm controlled impedance on a one-sided (also known as single-layer or microstrip) FR4 PCB, you need to consider both the width of the trace (as discussed in the previous response) and the spacing between traces if you're planning parallel tracks. Here's how you can approach this:



Trace Width (W) for 50 Ohms: As calculated earlier, for a 50 ohm microstrip:



Trace Width (W): Approximately 2.73 mm for: Board thickness (h) = 1.6 mm Dielectric constant (er) = 4.2 Copper thickness (t) = 35 µm (1 oz)


Spacing Between Traces: The spacing between traces (S) is critical when you're considering cross-talk or if you're routing multiple traces in parallel. Here's what affects this spacing:



Crosstalk: The closer the traces, the higher the crosstalk. Impedance: Spacing affects the effective dielectric constant and thus the impedance. For a single trace, spacing from the edge of the board or other traces can slightly influence the impedance, but mostly for very high frequencies or very thin traces. General Rule: A common rule of thumb for microstrip lines is to keep the spacing at least equal to or greater than the width of the trace to minimize crosstalk.


For practical purposes:



Spacing (S): A good starting point for spacing could be around



2.73 mm or more, matching the trace width. However, if space is a concern and you're operating at lower frequencies (like under a few GHz), you might reduce this spacing slightly, but you should be cautious about signal integrity. Simulation or Calculation: Use a PCB design tool or impedance calculator that can simulate or calculate the interaction between parallel traces. These tools can give you a more precise spacing for your specific conditions, including frequency, trace length, and signal characteristics. Manufacturing Considerations: Remember that manufacturing tolerances can affect both the width of your traces and the spacing between them. Therefore, slightly increasing the spacing can provide a safety margin.


Summary:



Trace Width: Approximately 2.73 mm for a 50 ohm single trace under the given conditions. Trace Spacing: At least 2.73 mm, or more for less crosstalk, depending on your design constraints and frequency of operation.


Always verify with PCB design tools or through empirical testing if possible, as theoretical models can deviate from real-world performance due to various factors like material inconsistencies or manufacturing variations.


<snipped non-right-sounding advice

ECL data books tended to include an application note that spelled out the constraints more sensibly.

You need to distinguish between microstrip (on the surface of your printed circuit board above a - possibly buried - ground plane) and strip line where the conducting strip of copper is buried within the board.

The thickness and dielectric constant of the insulating layer (or layers) is a critical parameter.

The ECL data books have nice simple formula for 50R and 75R transmission lines. I've got a pair of texts which give longer and more accurate formulae which stay accurate further away from the 50R to 75R range.

"High Frequency Circuit Design and Measurements" by Peter C, L. Yip - ISBN 0-412-34160-3

"Microwave Components, Devices and active Circuits" by Peter F. Combes, Jaques Graffeuil and Jean-Francois Sautereau ISBN 0-471-91277-8

I picked them both up at Heffers bookstore in Cambridge when I needed that kind of information. They aren't anything special.

Anybody who just gives you a trace width in mm isn't doing a proper job.

It doesn't seem to match anything I can get from the site below with 2oz copper and er=4 Does Grok know the difference between stripline and microstrip?

formatting link

Thank you, Edward. Well, to be fair to Grok, it did spell out a lot of qualifiers relating to the precise structure of the FR4 substrate which I did not include in the OP. So that could well have something to do with it. Grok also felt that performing such calcs was somewhat out of its comfort zone and referred me to specific websites for these calculations. However, I just wanted a quick and dirty figure as I'm not attempting metrology-level measurements in this instance and will not be venturing beyond VHF.

The pen-and-paper method given in Pozar's "Microwave Engineering" is to guesstimate whether W/d is >= or <= 2, use the second part of equation

3.197 to find A and B, use the first part to see if the guess is satisfied, use 3.195 to find the effective dielectric constant, and verify with 3.196.
formatting link

Or use the calculator included in KiCad..I get more like 3.1 mm for a width with the parameters given

The pen-and-paper method given in Pozar's "Microwave Engineering" is to guesstimate whether W/d is >= or <= 2, use the second part of equation

3.197 to find A and B, use the first part to see if the guess is satisfied, use 3.195 to find the effective dielectric constant, and verify with 3.196.
formatting link

Or use the calculator included in KiCad..I get more like 3.1 mm for a width with the parameters given

Thanks, Bitrex, that certainly helps. Here's the full reply Grok wrote (although I suspect some of the equations won't render correctly so go by the text primarily)

**************************************************

To design a 50 ohm microstrip transmission line on FR4, you'll need to consider several parameters:

Dielectric Constant (er): FR4 typically has a dielectric constant (er) around 4.4 at 1 MHz, but this can vary slightly with frequency. For microstrip calculations, you might use a value closer to 4.0 or 4.2 for higher frequencies. Thickness of the Substrate (h): This is the thickness of the FR4 board. Common thicknesses are

1.6 mm (0.063 inches) or 0.8 mm (0.031 inches). Trace Thickness (t): Typically, copper thickness might be 35 µm (1 oz), 70 µm (2 oz), or 105 µm (3 oz) for PCB manufacturing. Trace Width (W): The width of the trace is what we're primarily solving for to achieve 50 ohms impedance.

Using these parameters, you can use formulas or online calculators for microstrip impedance. Here's a simplified approach:

Formula for Microstrip Impedance:

Z0˜87?r+1.41ln?(5.98h0.8W+t)

Where: Z0 is the characteristic impedance (50 ohms in this case) ?r is the effective dielectric constant h is the board thickness W is the trace width t is the trace thickness

For a rough estimate with:

?r=4.2 h=1.6 mm t=35 µm (1 oz copper)

Solving for W:

50˜874.2+1.41ln?(5.98×1.60.8W+0.035) 50˜872.54ln?(9.5680.8W+0.035) 50˜34.25ln?(9.5680.8W+0.035) 5034.25˜ln?(9.5680.8W+0.035) 1.46˜ln?(9.5680.8W+0.035) e1.46˜9.5680.8W+0.035 4.305˜9.5680.8W+0.035 0.8W+0.035˜9.5684.305 0.8W+0.035˜2.222 0.8W˜2.187 W˜2.734 mm So, for a 50 ohm microstrip line on FR4 with these specifications, the trace width would be approximately 2.73 mm.

However, these calculations can be quite sensitive to manufacturing tolerances and material variations. It's recommended to use a precise microstrip calculator or simulation software for exact dimensions, and possibly to verify with a test board or adjust based on actual measurements if possible.

*******************************************************

It's good that it shows its methodology for arriving at this figure, IMO.

The Saturn pcb toolkit seems to be very good. We use that and the traces are right in real life.

It does other stuff, like calculating trace capacitances and inductances and via properties, lots of useful calcs.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required