I use relatively thick traces on all my boards, even for signals that could probably use 5–8 mils. I usually go with 12 mil traces and 12 mil clearance as my minimum, and for power lines (except ground) I often use 40–50 mils, wide enough to match an IC pad. Ground traces are usually 24 mils so they can run between pins. Most of my boards are single-sided and made using the photoresist method. I’ve found that thicker traces help compensate for my low-end printer, which sometimes botches thin traces during artwork printing.
The circuits are typically low-power (5V or 3.3V logic, max 1.5A draw from a 7805 regulator). I do have a few traces carrying audio signals and one 24 MHz crystal going to a microcontroller, but those traces are short (under 0.5") and wide (about 40 mils).
I’m curious, am I being overly conservative with trace widths? Are there any good rules of thumb or practical upper limits for trace width that I should consider especially when working with single-sided boards?
First, You are talking about track widths, not thickness.
You are conservative in this regard. I remember working on boards some 15+ years ago that were all though hole, DIP packages where the routing was all 10~12 mils wide and could be ran between the package leads.
Using today’s fabricators, going that wide is ok., but not required except for low capability fabrication methods (not common).
If the board has any higher speed type signals, then defining track width based on impedance becomes more critical. It does not sound like you are doing anything critical if you are getting by with single or even double sided boards.
If you’re not dealing with high-speed signals, there’s generally no downside to using wider traces, especially if it helps with fabrication reliability, as in your case with the photoresist method and print quality. That said, one thing to watch for (even at relatively low frequencies) is trace capacitance on longer runs. I once had a 10 MHz parallel bus where the traces were long enough that even 6 mil widths introduced too much capacitance, causing faults. So even below the GHz range, width and length together can matter, though that’s more the exception than the rule.
For high-speed signals (above ~1 GHz), you’ll definitely want to switch to controlled impedance, where the trace width is determined by your stack-up, not just convenience. But in your current use case, your 12/12 rule seems totally reasonable.
For homemade PCBs and even for some professionally made ones, using wider traces can improve reliability. It increases the chance that all connections will survive the manufacturing process, especially when using lower-end tools like basic printers and photoresist techniques. However, wider traces can make routing more difficult, particularly when you’re trying to fit them between pins on through-hole components. It’s even more of a challenge on boards with tighter layouts.
In your case, using wider traces makes sense given your process and tools. As long as everything fits on the board and functions correctly, your approach is reasonable and practical.
As a side note, making SMD boards at home can be easier than through-hole, even with fine-pitch parts like SSOP (0.5 mm). Not having to drill hundreds of holes really speeds things up and simplifies the process. For anyone hand-fabricating boards, SMD can be surprisingly accessible and may give you more routing flexibility even with wider traces.
For low-speed circuits and home fabrication, using wider traces that your process handles reliably is perfectly sensible. While it’s possible to get down to 8 mil traces with toner transfer, there’s usually no need to push for that unless you’re running into routing constraints. For high-speed designs you’d definitely need to consider impedance matching, which would also necessitate more advanced manufacturing processes.
One thing to keep in mind with wider traces on single-sided boards is that if you end up running them parallel and close together for long distances, it can introduce some unwanted coupling between signals. This usually isn’t a big issue at low frequencies, but if you move into more complex analog or mixed-signal designs, it’s something to watch for.