When a product needs power and signals across several subsystems, teams face a recurring design fork: put more on the PCB, or move function into a wire harness? Both are valid; the mistake is deciding by habit. Here is how we allocate function between board and harness on connected products.
What each medium is actually good at
| Aspect | PCB | Wire harness |
|---|---|---|
| High-speed / RF signals | Excellent (controlled impedance) | Poor beyond short runs |
| Power distribution at low current | Good | Good, with gauge choice |
| Mechanical routing through a chassis | Awkward, rigid | Excellent, flexible |
| Field service / replacement | Hard to swap | Easy to unplug and replace |
| Cost at volume | Low per function, high NRE | Low per unit, labour-heavy |
A rigid board is great for logic and high-frequency signals; a flexible harness is great for getting those signals to moving or distant points.
Decide by signal class first
- High-speed digital, RF, differential pairs: keep on the PCB with controlled impedance. Running them through a harness invites noise and loss.
- Low-frequency control, power, sensor leads: a harness is natural and serviceable.
- Mixed: break the boundary at a defined connector with a clear pin-out and shielding plan.
The connector where board meets harness is a reliability critical point — design it deliberately, not as an afterthought.
Where harnesses earn their place
A product with moving parts, a door, a rotating element or a remote sensor almost certainly needs a harness to bridge the fixed board to the moving point. Trying to "PCB everything" leads to flexing boards that crack. Similarly, a field-serviceable product benefits from harness segments you can unplug rather than desolder.
Where PCBs earn their place
Dense logic, any high-frequency or impedance-sensitive signal, and functions that shrink dramatically at volume belong on the board. Pushing those into a harness adds connectors, weight and failure points for no benefit.
The cost trade most teams miss
Harness cost is deceptively linear: each unit carries assembly labour. At high volume, a function that could be on a cheap PCB but lives in a hand-built harness becomes expensive. Conversely, a low-volume product does not justify a custom board for what a short harness does fine. Model both at the expected volume, not at prototype quantity.
A working decision checklist
- What is the fastest signal crossing this boundary, and what impedance does it need?
- Does any part of the path move or need field service?
- At target volume, which option is cheaper per unit?
- Where is the reliability-critical connector, and how is it locked and sealed?
- Can the harness be tested 100% (continuity + isolation) at the boundary?
FAQ: can we just use flex PCB instead of a harness?
Sometimes. A flex or rigid-flex PCB replaces a harness in tight or repeat-bend locations, but it costs more per unit and is harder to service. Use it where a harness cannot meet the bend or space requirement, not as a default.
Wrapping up
Allocate by signal class, serviceability and volume cost — not by preference. Keep high-speed on the board, use the harness for flexible and serviceable paths, design the board-to-harness connector as a critical interface, and model the cost at real volume. That balance is what makes a connected product both reliable and buildable.
For custom wire harnesses, connector assembly and 100% electrical testing, see Longkai Electronics. Company details are at link-lungkay.com.
Top comments (0)