Length Matching for High-speed Signals: Trombone, Accordion, and Sawtooth Tuning

Zachariah Peterson
|  Created: November 3, 2019  |  Updated: September 25, 2020

Layout for length matching for high-speed signals

Length matching for high-speed signals is all about synchronization...

Once upon a time, length matching for high-speed signals required a designer with enough skill to remain productive when manually applying different trace length turning schemes. Now, with the advanced interactive routing features in modern PCB design tools, designers no longer need to manually length tune every trace in their layout. The remaining choice for a designer is deciding which length matching scheme to use: trombone, accordion, or sawtooth routing.

So which of these different options is best for your high-speed design? With sufficiently wide traces (i.e., not in the HDI regime) and near-GHz bandlimited signals, you won’t have to worry about the complex resonance issues you’ll find when working with analog signals in the mmWave and sub-mmWave regimes. However, you still need to consider some important points regarding transmission line and signal integrity behaviors.

Options for Length Matching High Speed Signals

Whether you’re working with multiple signal nets that require tuning across a number of signals, or you just need to length match two ends of a differential pair, you’ll need to use some method for length tuning. At low speeds (generally TTL low power or slower), the difference between the different length matching styles is rather superficial. The differences between the two become more obvious at higher speeds.

Regardless of the speed you work with, it is generally recommended that each side of a differential pair be routed as symmetrically as possible. The idea is to maintain tight coupling between each end of the pair. However, I remember seeing an older article from Ben Jordan that showed how coupling in a differential pair can be maintained without perfectly symmetric length matching for high speed signals. This also applied when each end of the pair was routed on different layers. No matter how you choose to route your differential pairs, you should always verify the behavior of each signal in a differential pair using some simulation tools and, ultimately, using measurements.

The length matching configurations shown below are all precisely designed to ensure that impedance discontinuities are minimized along the length of the matched trace. This issue with impedance continuity throughout the trace is quite important as repeated reflections along an interconnect will accumulate, reducing the intensity of the signal reaching the receiver. This also leads to a stair-step climb in voltage at the receiver. This will always happen along a length-matched trace, although this behavior will be unnoticeable as long as your noise margin is large enough and you use the smallest number of bends as possible.

Trombone Tuning

If you are working with lower speed or lower frequency signals, you can get away with trombone tuning. You’ll notice that there are multiple 90 and 180 degree turns in this trace configuration. Using curves for these bends is preferable to using a hard right angle in that it creates a smaller impedance discontinuity. This is fine for lower speed single-ended signalling across a group of traces in parallel or for preventing clock skew in a functional block, but this can be a poor choice for length matching in differential pairs if not used correctly.

It should be obvious that the trombone portion alternates differential and common mode coupling between each side of the pair as the signal on one end moves back-and-forth through the trombone. Signals essentially switch between common mode and differential mode driving as they propagate. Just as with the other two common length matching methods, if you must use trombone tuning, then you should put it at the end of the differential pair where the mismatch arises. This is usually done near the source in order to guarantee that the signal is differentially driven at the receiver, ensuring common mode noise immunity.

Length matching for high-speed signals with trombone tuning

Trombone length matching for high-speed signals.

Accordion Tuning

Accordion tuning and trombone tuning are similar in that they both use a serpentine routing pattern. However, accordion tuning does not offset the serpentine pattern off to the side of the trace. Instead, it can be routed along the length of the desired signal trace. This is a better choice than trombone length matching for high speed signals on differential pairs as you can maintain tighter coupling along the length of the traces.

The layout shown below is fine for most applications, but it is not the best idea when you start reaching extreme signal speeds (e.g., ECL or TTL(G)). At very high speeds, you should place the length matching portion of the signal near the heavily mismatched region, i.e., closer to the driver or the receiver. This is particularly important for differential pairs as you should do your best to maintain tight coupling across the largest portion of the interconnect as possible.

Length matching for high-speed signals with accordion tuning

Accordion length matching for high-speed signals.

Sawtooth Tuning

An example of sawtooth tuning is shown below. Here, we haven’t used any smooth bends along the trace. The trace should be precisely spaced, as shown below. First, the “s-2s” rule used below is designed to ensure that 45 degree bends are used along the length of the length-matched trace. The “3w” rule (not to be confused with the crosstalk rule of the same name!) is really an upper limit; the length of the extended portion of the sawtooth should range from w to 3w, although some guidelines differ on this rule. These dimensions are used to minimize any impedance discontinuities along the length of the trace.

Length matching for high-speed signals with sawtooth tuning

Sawtooth length matching for high-speed signals: the “3w” rule.

In all three of the above methods, you should be careful not to place each section of a serpentine length matching section too close together. If you do this, portions of the serpentine length matching sections will interfere with each other through inductive crosstalk. This will distort signals as they travel along the length matching section. Howard Johnson provides an interesting explanation for this effect in this article.

Verification takes Simulations

The guidelines presented here are just that: guidelines. Anyone who has seen rules of thumb fail in certain situations knows that you should always check your layout, including length matching for high-speed signals, using post-layout simulation tools. This helps you examine important signal integrity problems like crosstalk, excessive signal reflection at bends, and skew in differential signals or across multiple traces that require precise synchronization.

The powerful interactive routing and post-layout analysis tools in Altium Designer® are built on top of a unified rules-driven design engine, allowing you to implement length matching for high-speed signals and checks for signal integrity. You’ll also have a complete set of tools for building schematics, your layout, and preparing deliverables for your manufacturer.

Now you can download a free trial of Altium Designer and learn more about the industry’s best layout, simulation, and production planning tools. Talk to someone at Altium today to learn more.

About Author

About Author

Zachariah Peterson has an extensive technical background in academia and industry. He currently provides research, design, and marketing services to companies in the electronics industry. Prior to working in the PCB industry, he taught at Portland State University and conducted research on random laser theory, materials, and stability. His background in scientific research spans topics in nanoparticle lasers, electronic and optoelectronic semiconductor devices, environmental sensors, and stochastics. His work has been published in over a dozen peer-reviewed journals and conference proceedings, and he has written 1000+ technical blogs on PCB design for a number of companies. He is a member of IEEE Photonics Society, IEEE Electronics Packaging Society, and the American Physical Society, and he currently serves on the INCITS Quantum Computing Technical Advisory Committee.

most recent articles

Back to Home