Discrepancy in Microstrip Impedance Calculation: 2.08mil/4.06mil/5.4mil/3.84 Dielectric

447 13

#1 21661131 08 Feb 2012 11:32

Rick Santarpio Rick Santarpio

Anonymous

Post #1
21661131 08 Feb 2012 11:32

I used this website to calculate microstrip impedance and found it to be inaccurate. The paramameters were: Trace Thickness = 2.08 mil, Substrate Height = 4.06 mil, Trace Width = 5.4 mils, Substrate Dielectric = 3.84 mils. The calculated impedance was 36.3 ohms but I believe it should be 50.6 ohms. When I google "impedance calculators pcb" this website is the first hit. The websites associated with hits 2-5 all calculate the microstrip impedance for this stackup to be 50.6 ohms. Can anyone explain this discrepancy?

ADVERTISEMENT
#2 21661132 10 Feb 2012 23:09

Joe Wolin Joe Wolin

Anonymous

Post #2
21661132 10 Feb 2012 23:09

Rick,

We are checking into this now. Thanks for bringing it to our attention.
ADVERTISEMENT
#3 21661133 11 May 2012 13:25

Chris Lee Chris Lee

Anonymous

Post #3
21661133 11 May 2012 13:25

I've questioned the accuracy too. See my comparison of three different online calculators.

chris

distances are in mils

[code lang="php" line="none"]source w s t h er Zd (ohms)
Mantaro 38 5 0.7 62 4.3 100.3
EEWeb 26 5 0.7 62 4.3 99.7
Saturn 37 5 .5oz 62 4.3 100.4

Mantaro 44 10 0.7 62 4.3 100.1
EEWeb 46 10 0.7 62 4.3 100
Saturn 43 10 .5oz 62 4.3 100.2

Mantaro 22 5 0.7 31 4.3 99.3
EEWeb 23 5 0.7 31 4.3 99.7
Saturn 21 5 .5oz 31 4.3 99.5

Mantaro 27 10 0.7 31 4.3 99.9
EEWeb 36 10 0.7 31 4.3 99.6
Saturn 26 10 .5oz 31 4.3 100.1[/code]
#4 21661134 12 May 2012 17:01

Cody Miller Cody Miller

Anonymous

Post #4
21661134 12 May 2012 17:01

Some online calculators use a very simplified approximation to calculate impedance. These formulas work if your looking for a ball park impedance. Then a few years ago a committee put together the IPC-2141A Design Guide for High-Speed Controlled Impedance Circuit Boards. This effort was to come up with a set of approximation formulas in one standard. This is what EEWeb used for this impedance calculator.

To get back to the comparison that you made we have some data from Polar's impedance solver a high end simulation tool.

W=38, H=62, Er=4.3

Polar Zo=87.91
EEWeb Zo=90.2
Mantro Zo=90.23
ADVERTISEMENT
#5 21661135 12 May 2012 17:06

Cody Miller Cody Miller

Anonymous

Post #5
21661135 12 May 2012 17:06

In Regards to Ricks Questions I think it may be that you used the wrong calculator. Use this link for "Microstrip Impedance":http://www.eeweb.com/toolbox/microstrip-impedance/

I calculate 50.6 ohms

see attached screen shot
#6 21661136 15 May 2012 07:35

Rick Santarpio Rick Santarpio

Anonymous

Post #6
21661136 15 May 2012 07:35

Thanks Cody,

I agree, it now seems to work. I can't explain the former discrepancy. Thinking I might have been using the wrong calculator, I went back and tried using the only other impedance calculator with 4 input parameters (Symmetrical Stripline) and got 33.1 ohms rather than 36.3 ohms. At this point I wish I took a screen shot of what I saw a few months back. The good news is that the calculator appears to be correct.
#7 21661137 12 Jun 2012 23:58

Ian Lewis Ian Lewis

Anonymous

Post #7
21661137 12 Jun 2012 23:58

Have you verified your embedded impedance calculator?

Using http://www.eeweb.com/toolbox/embedded-microstrip-impedance/ with
T: 1.4, H1: 4, H2: 5.7, W: 5, Er: 4.2

I get 34.6 Ohms.

That seems like way too much drop from the basic surface calculator (http://www.eeweb.com/toolbox/microstrip-impedance/), which gives 54.7 Ohm for the same basic stackup.

In fact, it is very close to what the symmetrical calculator (http://www.eeweb.com/toolbox/symmetric-stripline-impedance/) gives for a 4 mil separation and other parameters the same.

It is also much lower than any other calculator I have found. Saturn and the old WindRiver spreadsheet give ~46 Ohms. Other web-based calculators give about the same. They all seem to be based on the same formula.
ADVERTISEMENT
#8 21661138 13 Jun 2012 00:07

Ian Lewis Ian Lewis

Anonymous

Post #8
21661138 13 Jun 2012 00:07

Assuming I am thinking about this correctly, this is looking really unlikely: Set embedded microstrip to match the embedded stripline using 1.4, 4, 9.4, 5, 4.2 and compare that to embedded stripline with 1.4, 4, 5, 4.2, physically exactly the same thing, but the second has a piece of metal on top, and the embedded microstrip gives a lower value.

Embedded microstrip: 32.9 Ohms
Embedded stripline: 34.5 Ohms

That is not physically possible. The copper layer cannot increase the impedance.
#9 21661139 14 Jun 2012 00:07

Cody Miller Cody Miller

Anonymous

Post #9
21661139 14 Jun 2012 00:07

The copper layer absolutely changes the impedance. The additional copper layer increases the capacitance of the trace which decreases the impedance. A general formula for a transmission line is:

[tex]Z_{O}=sqrt{frac{L}{C}}[/tex]

Since C is in the denominator as C increase Zo decreases. Check out this "link":http://en.wikipedia.org/wiki/Telegrapher's_equations for more information.
#10 21661140 14 Jun 2012 17:49

Ian Lewis Ian Lewis

Anonymous

Post #10
21661140 14 Jun 2012 17:49

I understand that the copper layer decreases the impedance. That was exactly my point.

The EEWeb embedded microstrip gives a lower, not higher, impedance than the stripline.

That claims the impedence is increased by the copper layer. I see no physical mechansim by which that could be correct.

More than that, I would expect the embedded microstrip impedance to be much higher than the stripline with comparable spacing to the plane layers as I believe I set up in my example parameters.

One case has a single plane at 4 mil and the other has two planes each at 4 mil. Other than that, I think I configured every other physical characteristic identically from the point of view of the equations.
#11 21661141 20 Jul 2012 12:05

Rainer Kordmaa Rainer Kordmaa

Anonymous

Post #11
21661141 20 Jul 2012 12:05

these calculations are approximations at best even if everything is calculated correctly, but yeah i suspect there are errors in this calculator
http://www.eeweb.com/electronics-forum/stumed...dge-coupled-microstrip-impedance-calculation/

i posted about this, in the formulas presented in the web calc explenation i found several differences from the IPC-2141A
#12 21661142 17 Mar 2017 12:54

Shen Zhao Shen Zhao

Anonymous

Post #12
21661142 17 Mar 2017 12:54

Attached screen shot shows a different result from it was 4 years ago. It is now 58.0 ohms. What I'd like to say is that this is not a reliable source, and the reason I found this posting is I had a question about the accuracy of its "Asymmetric Stripline Impedance Calculator". For a board we designed, the EEWeb calculates 47.9 ohms, while Mantaro gives 29.967 ohms (see other two screen shots). Quite a difference!

Unfortunately, we made the PCB based on the EEWeb calculation, and the impedance of the trace turns out to be close to the Mantaro's calculation.
#13 21661143 17 Mar 2017 22:55

Kevin Angelo Ma Kevin Angelo Ma

Anonymous

Post #13
21661143 17 Mar 2017 22:55

I think the source of EEWeb comes from _he IPC-2141A (2004) “Design Guide for High-Speed Controlled Impedance Circuit Boards”_ . It is quire bothersome to see that there is quite a difference in the answers. Here is a "link ("")":http://www.ipc.org/4.0_Knowledge/4.1_Standards/2141A_Errata.pdf to the errata of the IPC2141A. Like what Cody send "IPC standard based their work off of the Transmission Line Design Handbook and that the equations in the handbook is what they intended."
Also referring to previous comments in this forum, https://www.eeweb.com/electronics-forum/stume...edge-coupled-microstrip-impedance-calculation
#14 21661144 07 Apr 2017 09:57

zeropond kumar zeropond kumar

Anonymous

Post #14
21661144 07 Apr 2017 09:57

Hello
I am new member of this forum site. But, I have known that Microstrip Impedance Calculator Inaccurate

In about my numbers are:
Dielectric constant - 5.58
Dielectric height - 21mil
Signal line width - 23 mil
Copper thickness - 11z - 3.4mil
Create an account, log in here. You will receive points by participating in discussions.
Join this discussion.

Install Elektroda application

Didn't find an answer? Ask Artificial Intelligence

*I agree to send the question to OpenAI, Anthropic PBC, Perplexity AI, Inc., Kagi Inc., Google LLC - owners of language models in order to prepare the best response. The companies may monitor and log information entered into the form.

*I agree to publicly display my question and answer. The question and answer will be publicly available to everyone. The process may take a few minutes. Upon completion, you will be redirected to the page with the answer.

Wait...(2min)

Topic summary

✨ A discrepancy was reported in microstrip impedance calculations using an online calculator with parameters: trace thickness 2.08 mil, substrate height 4.06 mil, trace width 5.4 mil, and dielectric constant 3.84, yielding 36.3 ohms instead of the expected ~50.6 ohms. Multiple online calculators and references, including Mantaro, Saturn, and others, consistently calculate around 50.6 ohms for similar stackups. The discrepancy was attributed to possible use of incorrect calculators or simplified approximation formulas. The EEWeb calculator is based on IPC-2141A standard formulas, which are approximations and may differ from high-end simulation tools like Polar’s impedance solver. Discussions highlighted that embedded microstrip impedance calculations can be lower than expected due to additional copper layers increasing capacitance and thus lowering impedance, consistent with transmission line theory (Z0 = sqrt(L/C)). However, some EEWeb embedded microstrip calculators showed physically inconsistent results compared to embedded stripline calculators, suggesting potential errors in the formulas or implementation. The IPC-2141A standard and its errata were referenced as sources of the formulas used, but differences remain between calculators. Users noted that these calculators provide approximations and recommended caution, especially since PCB fabrication results sometimes align better with alternative calculators like Mantaro. Overall, the consensus is that microstrip impedance calculators vary in accuracy due to differing formula implementations, assumptions, and approximations, and high-fidelity simulation or measurement is advised for critical designs.
Generated by the language model.

FAQ

TL;DR: One stackup returned 36.3 Ω, but the correct microstrip calculator gave 50.6 Ω; the issue was “used the wrong calculator.” [Elektroda, Cody Miller, post #21661135]

Why it matters: Picking the wrong field‑solver or geometry model can misroute budgets and ruin impedance on finished PCBs.

Quick Facts

Correct tool/inputs for 2.08 mil T, 4.06 mil H, 5.4 mil W, εr=3.84 → 50.6 Ω. [Elektroda, Cody Miller, post #21661135]
IPC‑2141A equations underpin many web tools; a Polar check showed 87.91 Ω vs ~90 Ω from two calculators. [Elektroda, Cody Miller, post #21661134]
Using a stripline calculator instead of microstrip led to 33.1 Ω earlier; user confirmed resolution after switching. [Elektroda, Rick Santarpio, post #21661136]
Embedded‑microstrip example: surface ≈54.7 Ω, embedded ≈34.6 Ω for similar stackup—flag for validation. [Elektroda, Ian Lewis, post #21661137]
Field report: EEWeb 47.9 Ω vs Mantaro 29.967 Ω; measured board matched Mantaro’s prediction. [Elektroda, Shen Zhao, post #21661142]

Why did I get 36.3 Ω when others show about 50.6 Ω for the same stackup?

You likely chose the wrong calculator geometry. Re‑running as a microstrip with the same dimensions produced 50.6 Ω. The poster who flagged the mismatch confirmed it after using the microstrip tool rather than a stripline calculator. “Use the wrong calculator” was the root cause. [Elektroda, Cody Miller, post #21661135]

Which formulas do many online impedance calculators implement?

Several sites implement approximation formulas collected in IPC‑2141A, a design guide for controlled‑impedance boards. One contributor noted their calculator used IPC‑2141A, aligning with common industry practice for quick estimates rather than full 2D/3D solvers. [Elektroda, Cody Miller, post #21661134]

Are IPC‑2141A‑based results close to field solvers?

They can be close but not exact. A Polar impedance solver returned 87.91 Ω, while two web calculators gave ~90.2 Ω and ~90.23 Ω for the same case. Expect a few ohms difference without detailed field solutions. [Elektroda, Cody Miller, post #21661134]

How does a nearby copper plane change microstrip impedance?

A nearby plane increases capacitance, which lowers characteristic impedance. As one expert said, “The additional copper layer increases the capacitance of the trace which decreases the impedance.” That’s consistent with Z0 ∝ √(L/C). [Elektroda, Cody Miller, post #21661139]

Why did an embedded‑microstrip calculator give ~34.6 Ω while the surface version gave ~54.7 Ω?

The embedded geometry includes dielectric above the trace, raising capacitance and lowering Z0. A user observed 34.6 Ω embedded versus 54.7 Ω surface for similar inputs, prompting a validation check of that tool. [Elektroda, Ian Lewis, post #21661137]

Can an embedded microstrip ever show lower Z0 than a symmetric stripline with equal plane spacing?

One report showed an embedded microstrip calculated lower Z0 than an equivalent stripline, which contradicts physical intuition. The user highlighted this as a calculator anomaly needing review. Treat such outputs as red flags. [Elektroda, Ian Lewis, post #21661138]

What real‑world risk comes from trusting one calculator?

A team built a PCB using one web calculator predicting 47.9 Ω. The finished board measured closer to another tool’s 29.967 Ω prediction. Cross‑check before fabrication to avoid expensive respins. [Elektroda, Shen Zhao, post #21661142]

Quick 3‑step: how do I sanity‑check a microstrip impedance number?

Run two independent calculators (e.g., EEWeb, Mantaro, Saturn) using identical inputs.
Compare with a reference from a field solver or vendor stackup note.
Reconfirm geometry (microstrip vs stripline vs embedded) and copper thickness units. [Elektroda, Chris Lee, post #21661133]

Which inputs dominate microstrip impedance?

Conductor width (W), dielectric height (H), copper thickness (T), and dielectric constant (εr) dominate. A shared comparison table varied W, spacing, and thickness, showing Z0 near 100 Ω when tuned across those parameters. [Elektroda, Chris Lee, post #21661133]

What’s the difference between microstrip, stripline, and embedded microstrip?

Microstrip runs over a reference plane with air above. Stripline sits between two planes inside the dielectric. Embedded microstrip runs over a plane but is buried under a dielectric cover. The thread discusses all three geometries and their calculators. [Elektroda, Ian Lewis, post #21661137]

Why do different tools disagree by several ohms?

Most web tools use simplified approximations, and some implementations deviate from published formulas. One poster found several formula differences from IPC‑2141A in an online explanation, suggesting possible source errors. [Elektroda, Rainer Kordmaa, post #21661141]

Did the calculator’s output change over time for the original case?

Yes. A later screenshot showed 58.0 Ω for the same scenario discussed years prior. This variability underscores why you should verify results across tools and time. [Elektroda, Shen Zhao, post #21661142]

How do I avoid the stripline vs microstrip mix‑up?

Match the geometry to your stackup. If your trace is on an outer layer above a single plane, use microstrip. One user resolved a 36.3 Ω vs 50.6 Ω discrepancy by switching to the microstrip tool. [Elektroda, Rick Santarpio, post #21661136]

Do calculators handle copper thickness consistently (e.g., 0.5 oz)?

Not always. One comparison used “0.5 oz” and “0.7 mil” entries across tools that still converged near 100 Ω. Always confirm units and thickness mapping in each interface. [Elektroda, Chris Lee, post #21661133]

Generated by the language model.