PCB Trace Width: IPC-2221 vs IPC-2152

IPC-2221 vs IPC-2152: Which Should You Use?

So you're sizing traces and wondering which standard to follow. Here's the deal: IPC-2221 came out in 1998, but it's actually based on measurements from 1954. Yeah, 1954. The formula is simple and conservative:

The k factor is 0.048 for external traces (the ones exposed to air on top or bottom layers) and 0.024 for internal traces buried in the stackup. ΔT is your temperature rise in °C, and A is the cross-sectional area in mil². It works, but it's overly cautious by modern standards.

IPC-2152 showed up in 2009 and changed the game. They actually ran new experiments with modern PCB materials and layer stackups instead of relying on data from the Eisenhower era. The result? You can use narrower traces for the same current, or push more current through the same trace width. For something like a 10A external trace with a 10°C rise, IPC-2152 lets you get away with a trace that's roughly 30–40% narrower than what IPC-2221 demands. That's real board space you're getting back. For new designs, use IPC-2152. Period. The only reason to touch IPC-2221 is if you're dealing with a customer or certification body that specifically calls it out by name in their requirements. Otherwise, you're just wasting copper and board real estate.

Temperature Rise Budget

Your trace doesn't exist in a vacuum (well, unless you're doing space hardware). The actual temperature it reaches is the ambient temperature plus whatever rise your current causes:

FR4, which is what most of us use, has a glass transition temperature (Tg) somewhere between 130°C and 170°C depending on the grade. You really don't want to get close to Tg — the board starts getting soft and weird things happen mechanically. Stay at least 20°C below it, more if you can.

Here's where it gets tricky. Say you're designing something that goes inside an enclosure with other heat-generating stuff. Your ambient might be 70°C, not the 25°C you're used to thinking about. If your board's Tg is 130°C and you keep that 20°C safety margin, your maximum trace temperature is around 110°C. That leaves you with only 40°C of temperature rise budget to work with. Not a lot of headroom.

Most engineers aim for these targets depending on application:

• Consumer electronics: 10°C rise — keeps things cool to the touch and maximizes reliability
• Industrial gear: 20–30°C rise — still reasonable, components are rated for it
• Power electronics: 30–40°C rise — you're pushing it, but sometimes you need every millimeter of board space

These aren't hard rules, just what tends to work in practice. I've seen power supply designs that accept 50°C rise on short trace segments because the thermal mass is low and it doesn't matter. Context is everything.

External vs Internal Layers

This is where people often get surprised. Internal traces — the ones sandwiched between layers in your stackup — run significantly hotter than external traces carrying the same current. Why? Heat dissipation.

External traces can dump heat directly into the air (or into your thermal camera when you're debugging why something is melting). Internal traces are surrounded by FR4, which is a terrible thermal conductor. We're talking about 0.3 W/m·K for FR4 versus roughly 150 W/m·K for copper. Heat has to conduct through multiple layers of fiberglass and epoxy to escape, and it does so grudgingly.

The IPC-2221 formula captures this with that k factor: 0.024 for internal versus 0.048 for external. That's a 2× difference. In practice, internal traces need roughly twice the cross-sectional area to carry the same current at the same temperature rise. If you calculated a 20 mil external trace, plan on 40 mils (or more) if you have to route it internally.

Most engineers try to keep high-current paths on external layers whenever possible. If you absolutely must route power on an internal layer, be generous with the width. I've debugged enough boards where someone assumed internal and external were equivalent — they're not, and your nose will tell you when you power it up.

Copper Weight and Cross-Section

Copper weight is one of those specs that seems simple until you start doing the math. The industry uses ounces per square foot, which is wonderfully unintuitive. Here's what it actually means for your trace dimensions:

Standard PCB fab houses do 1 oz copper by default. It's cheap, well-understood, and works for most stuff. But look at that table — going from 1 oz to 2 oz doubles your cross-sectional area for the same trace width. That means you can carry twice the current (roughly) without widening the trace. Or you can halve the trace width for the same current capacity.

For power supplies and motor controllers, I usually spec 2 oz copper. The cost bump is minimal unless you're doing huge production runs, and it gives you so much more flexibility in routing. Just watch out for minimum trace width and spacing — thicker copper is harder to etch cleanly, so your fab house might push back on 4 mil traces with 2 oz copper.

Resistance and Voltage Drop

Here's something that bites people: even if your trace stays thermally happy, you might still have a problem. Voltage drop is real, and it's proportional to resistance:

Copper resistivity ρ is 1.72×10⁻⁸ Ω·m at 20°C, and it increases with temperature — the coefficient α is 0.00393 per °C. That term in brackets accounts for the resistance going up as the trace heats up.

Let's work through a real example. You've got a 100mm long trace, 1mm wide, using standard 1 oz copper. You're pushing 3A through it. The cross-sectional area is 1 mm × 0.035 mm = 3.5×10⁻⁸ m². Plug in the numbers:

• R = (1.72×10⁻⁸ × 0.1) / (3.5×10⁻⁸) = 0.049Ω
• V_drop = I × R = 3A × 0.049Ω = 0.15V
• P_loss = I² × R = 9 × 0.049 = 0.44W

That's 0.15V dropped across your trace. If you're running a 3.3V rail, you just lost 4.5% of your voltage budget before you even got to the load. For a 5V rail it's more tolerable, but for anything precision or low-voltage, it's a problem.

The power dissipation is 0.44W, which doesn't sound like much, but it's spread over a small area. That's what causes the temperature rise we calculated earlier. Long high-current traces need to be wider, or you need to jump to 2 oz copper. Sometimes both.

Practical Tips

Okay, enough theory. Here's what actually works when you're laying out boards:

Pour copper on power rails instead of routing traces. Seriously. A 10mm wide copper pour at 1 oz can easily handle 20A or more with less than 5°C rise. It's lower resistance, lower inductance, and you don't have to worry about calculating widths for every segment. Just flood the area and call it done. I see people routing 100 mil power traces when they could pour a polygon and have better performance with less effort. Use thermal vias under hot traces to spread heat. If you've got a high-current trace on an external layer, drop an array of vias under it to pull heat into the internal copper layers and spread it out. Space them every 0.5 to 1mm along the trace. Use 10 or 12 mil vias — bigger is better for thermal transfer. This is especially important if the trace is long or if you're routing near thermal limits. The internal copper planes act as a heatsink. Verify everything with an IR camera on your first prototype. I can't stress this enough. All these calculations assume ideal conditions: uniform current distribution, no adjacent heat sources, specific airflow, perfect copper plating thickness. Real boards are messier. That trace might run cooler because there's a ground plane nearby acting as a heatsink, or it might run hotter because it's next to a linear regulator that's dumping 2W. The IR camera tells you the truth. Flir makes a phone attachment that's good enough for most work — I've caught so many issues with mine that it paid for itself on the first project.

One more thing: if you're doing anything with serious current — motor drives, power supplies, battery charging — consider getting your fab house to do a cross-section analysis on your first run. They'll cut through your board and measure the actual copper thickness and trace geometry. Plating thickness varies, and that 1 oz copper might actually be 0.9 oz or 1.1 oz depending on how the plating bath was running that day. For critical designs, knowing the real dimensions matters.

Calculate your trace dimensions with our PCB Trace Width Calculator — it shows both IPC-2221 and IPC-2152 results side by side so you can see the difference and make an informed choice. Plug in your current, temperature rise, and copper weight, and it'll give you the trace width you need. Way faster than doing the math by hand, and it's easy to try different scenarios to see what works best for your layout.