2
\$\begingroup\$

I have a copper trace on a PCB board, but it terminates in an annular ring. My question is, can I trace the path of the trace? I need to determine if it is connected to ground. Thank you in advance.

PS: Apologies for any potential errors in my English; I'm learning the language

\$\endgroup\$
4
  • \$\begingroup\$ DMM in resistance mode? Ground is usually connected to multiple points on the board, it should be easy to find one and measure resistance to the trace. \$\endgroup\$
    – Lior Bilia
    Commented May 15 at 22:33
  • \$\begingroup\$ @Lior Biliá In resistance mode? Wouldn't measuring continuity be sufficient? In the scenario you mentioned, it's possible if there's only one ground plane, but I understand that there can also be multiple isolated ground planes. Could you please explain? \$\endgroup\$
    – condor12
    Commented May 15 at 23:08
  • \$\begingroup\$ Continuity is often sufficient, but be aware that it's generally based on a threshold, so you may be presented with false-positives for low resistance paths... Is it a two layer board, or more? (for 2-layer, just look at the opposite side!) \$\endgroup\$
    – Attie
    Commented May 15 at 23:14
  • \$\begingroup\$ The continuity measurement just indicates that the resistance between the meter leads is below some low value (about 35 Ohms for one of my meters). It does not indicate a direct zero-Ohm connection. \$\endgroup\$ Commented May 15 at 23:33

2 Answers 2

3
\$\begingroup\$

Sure, there are many ways. Some more tedious than others...

By hand: simply check every other pad on the board for continuity!

(Simply, he says...)

Note that trace resistances are typically single ohms or less, so beware of a DMM's continuity range. Make note of components connected, particularly low-value resistors that can influence the measurement.

Beware, there are occasionally features called "net bridges", where a trace might connect to other objects on the board, but needs to be treated differently until then: for example, a trace might enter a polygon pour, only joining with it at a critical point. A typical application is Kelvin terminals for a power resistor. The polygon would pour over the same net trace, but placing the bridge under the resistor, the trace connects as close to the component as possible, minimizing error. You'll measure the connectivity, of course, but not why it was arranged the way it was: that will require more careful inspection.

You may need to remove components, perhaps all the way to a bare board (but, do make note of any bodge wires along the way, and inspect traces for cuts or patches), to get the most accurate picture.

By machine: it can be done by wiring to every pad and testing the connectivity matrix.

(Simply, he said?..)

Crazy, but it has been done: MAPPING THE NINTENDO SWITCH PCB | uSoldering via Hackaday for example.

Alternately, if you can make a table of pad locations, a flying probe tester could be used to check every pad, exhaustively, with automation assistance.

Without board files in hand (the usual case for these machines -- and also the faster one, as only every net needs to be checked (all pads connected to the given net, all nets disconnected from all others), not literally all pads), I suppose the best possible case would be something like:

  • Strip the board of components; clean off solder as well as possible, leaving a smooth and level board with bright finish on the pads. (Optional: dissolve the solder to expose bare copper for better contrast.)
  • Photograph the board, using a low-distortion camera, or flatbed scanner. Scale to actual size, and rotate and deform as needed to correct imaging errors.
  • Adjust image thresholds until the solder (or bare Cu) pads are highlighted. This extracts the soldermask pad layer.
  • Run a program to find pad centroids, and tabulate them.
  • Feed the table to the flying probe machine and wait.

An adaptive algorithm of course can reduce solution time by mapping out individual nets (you don't have to check between all pads, but one pad from a given discovered net and all others, and so on until all nets have been found), but will require more programming, and more interaction with the machine.

There might well be ready-made tools for this, out there; I don't know. The main industry need, as far as I know, is just to test fabbed boards -- the netlist is known -- but I've never operated such a machine myself, and don't know what features are typical.

By inspection: X-ray the board and follow traces on various layers. Discriminating layers in a radiogram can be very difficult, impossible even, say when traces align vertically (in the stackup) so you can't tell which one is splitting off where. Perhaps with tilting the board through a few angles, a rudimentary CT (computed tomography) approach can be used, even just visually (enough tilt/parallax to tell apart some traces, but not all; but by checking various angles, you figure it out).

Advanced equipment isn't necessary here; specialized, yes, but not wildly so. A simple mail scanner can prove useful for electronics work -- also for inspecting soldering, etc. The density of materials (lots of metal, including heavier ones like tin and perhaps lead) does make this more challenging -- ideally a higher energy source would be used (50keV+?), but weaker sources can still do the job with patience (i.e., very long exposure times).

By visual inspection: For 2-layer boards, this is... I'm not going to say trivial, but given that you may need to remove some components to see everything, it's nothing out of the reach of an electronics assembly/repair tech, for example. For 4-layer boards, if there is no, or minimal, routing on inner layers, traces can be followed around just as well, and only a few default nets (VCCs/GNDs) will be hidden inside -- which will also be apparent as their vias have "dead end" traces on the outer layers (only connecting to non-signal ends of components, power pins). Whereas signal vias will always have traces connecting them top and bottom. If they are using inner layer(s) for routing, the seemingly-dead-end traces will become apparent on analysis; then, you can see how problematic it is, if it's just a few and easy enough to probe out, or many nets and is kind of a lost cause.

If a destructive test is permissible, one can sand or mill down the board, layer by layer, photographing progress. Also very tedious, and messy, but grants a direct view of the whole thing.

\$\endgroup\$
2
\$\begingroup\$

If you have a 2 V power supply with adjustable current limit, you can connect it to the track and GND and watch the current path with a thermal camera.

The current should not overload the trace and melt it, but be high enough to produce a temperature difference. I would use 1..2 A for a typical 0.2 mm width trace.

Inner traces are visible as well, until the board temperature rise makes the picture blurry.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.