2
\$\begingroup\$

I have built a simple bus backplane from stripboard and 40-pin female headers for a computer project I am working on. When I tested the backplane to make sure that it functioned, I found that it's soldered properly but that inside one or two of the female headers I used as connectors, there is a short that connects three lines on the backplane.

Visual inspection is of little help, the flaw isn't visible; it is inside the connector shroud somewhere.

So my question is: Is it possible to measure with a multimeter somehow, which of the seven paralleled connectors is the culprit? I would like to salvage the backplane and just replace the offending connector if possible.

back of board

front of board

\$\endgroup\$
10
  • 2
    \$\begingroup\$ So, you know what three lines are shorted together or not? If you do, get out the magnifying glass and insect all the solder joints on those three? \$\endgroup\$
    – Tyler
    Aug 12 '18 at 12:22
  • 3
    \$\begingroup\$ @Psvedman No, it's not inside one of these connectors, unless dirt somehow fell in there: these things are made from individual metal contacts; the chances of a short within a connector approach 0. On the other hand, hand soldering is not an industrial process, so I fully agree with Tyler: don't try to look at the connectors; look at what you did. \$\endgroup\$ Aug 12 '18 at 12:40
  • 4
    \$\begingroup\$ I have used high current power supplies (and once, a car battery) to make invisible shorts like this disappear. \$\endgroup\$
    – amb
    Aug 12 '18 at 12:50
  • 2
    \$\begingroup\$ @amb we shall call that method "pyrolytic debugging". On a serious note, though, carbonized PCB substrate is a mediocre conductor, and you would avoid that in systems where a mediocre undesired conductor between traces is still bad. \$\endgroup\$ Aug 12 '18 at 13:17
  • 2
    \$\begingroup\$ @amb no, seriously, especially in prototypes, that kind of debugging is totally legit! (it's just destructive) If you see smoke, something was wrong. In many cases, repairing that might have never paid, anyway, so you'd just throw away the board and test the next. In fact, for things like safety transformers for medical appliances, "isolation tests" are mandatory: you apply a voltage source (typically higher than what the device is supposed to work at), and measure the current (which should be darn close to 0). In a bind, "smoke" is a current indicator. \$\endgroup\$ Aug 12 '18 at 13:31
6
\$\begingroup\$

In all probability, it's got something to do with your soldering and not with the insides of the connectors (which were made by precision machines, and likely even tested before leaving the factory).

Take a knife and make sure to clean the gaps between your tracks. It must also fit between the solder "blobs".

Your solder joints aren't very uniform – that's really where I'd start looking. Even the tiniest little solder needle will be a short. All in all, from an effort point of view, if simply removing as much solder as possible from these and resoldering the 21 contacts in question doesn't help, redo the whole board. That time, first cleanse your copper track (slight rub with something abrasive, alcohol), so that you don't get a copper oxide layer. Apply a small amount of (preferably liquid) flux to the tracks prior to soldering in your first connector row. Use less solder overall.

\$\endgroup\$
5
  • \$\begingroup\$ I took out my microscopw a \$\endgroup\$
    – Psvedman
    Aug 12 '18 at 13:17
  • \$\begingroup\$ Sorry. I took out my microscope and inspected the solder joints. I agree that they arent too pretty, but I didnt find any solder bridges. Wjat \$\endgroup\$
    – Psvedman
    Aug 12 '18 at 13:18
  • \$\begingroup\$ Then redo the 21 solder joints. No point in trying to solve something you can't see. \$\endgroup\$ Aug 12 '18 at 13:19
  • 2
    \$\begingroup\$ What I did find was that the stripboard was badly etched, there were thin copper bridges, not visible with my magnifying glass, between some tracks. It seems that these were the culprits. I have scored all the channels on the board with a knife and the problem, and future problems from this, went away. \$\endgroup\$
    – Psvedman
    Aug 12 '18 at 13:20
  • \$\begingroup\$ Ouch! Yeah, well, maybe consider switching stripboard brands, then. Something that a reputable distributor sells shouldn't have these issues. I'd raise this with whoever sold you these PCBs (and even if they refund you consider to not buy from them again). \$\endgroup\$ Aug 12 '18 at 13:22
17
\$\begingroup\$

This is what I found, these tiny little copper bridges were all over the board. Problem solved, thank you for the help.

enter image description here

\$\endgroup\$
4
  • 6
    \$\begingroup\$ ooooooh. That's nasty.! \$\endgroup\$ Aug 12 '18 at 13:39
  • \$\begingroup\$ You can also accept your own answer. Please mention what your action was to solve the problem? A knife as well as suggested by Marcus? Or did you just apply some voltage to melt them away? \$\endgroup\$ Aug 12 '18 at 18:22
  • \$\begingroup\$ A knife. It cut off all the invisible little copper bridges. \$\endgroup\$
    – Psvedman
    Aug 12 '18 at 20:15
  • \$\begingroup\$ @MarcusMüller I concur with your assessment OOOOOH. If solder joints aren't the issue maybe carbon deposits between the tracks are. Carbon is highly conductive...maybe the smoke test is in order to burn away the carbon. \$\endgroup\$
    – Old_Fossil
    Aug 13 '18 at 6:09
6
\$\begingroup\$

Although you've already found the problem (well done), here are a couple of options for people with similar problems in the future:


Dedicated milliohm meter

A special kind of meter could have been used, but not many people have one - a milliohm meter. These are capable of measuring resistance with much greater sensitivity than typical multimeters.

Starting the measuring at one end of the "shorted" tracks and moving towards the other end, the measured resistance would have steadily decreased as the milliohm meter probes got closer to the location of the short between two tracks. This would tell you that you were getting physically closer to the short.

Continuing moving in the same direction along the tracks, the resistance would eventually then started to increase, which would tell you that you have now passed the physical location of the short. You're looking for the lowest resistance reading, to tell you that you are closest to the physical location of the short.

After removing one short, you would need to repeat the process in case there were multiple shorts between the same two tracks (as there were on your board).


Home-made milliohm meter

Another approach may be possible in your situation of a "bare" board (i.e. no components) but only if you have a variable current, current-limited power supply (e.g. a typical bench supply) and a multimeter with millivolt sensitivity. This becomes a home-made milliohm meter (not very accurate, but as I explain below, that doesn't matter in your case).

You would connect the power supply between the two "shorted" tracks, but not with the intent to burn out the short. The current is intended to cause enough of a voltage drop across even low value resistances, to have a measurable voltage drop on a good multimeter e.g. in the millivolt range.

So with the power supply set to current-limiting at say 1A († see below), you use the millivolt range of the multimeter, moving the probes along the two shorted tracks, looking for the point of lowest voltage. By Ohm's law, the point of lowest voltage, with a fixed current from the power supply, would also be the point of lowest resistance i.e. you're probes are closest to the short.

Depending on the accuracy of the current limiting from your power supply, the millivolt reading on your multimeter might not be able to be accurately converted into milliohms, but for finding a short you don't care about the absolute reading. As described above, you only care about the relative changes which occur as you move the probes along the tracks i.e. is the reading getting higher (you're moving away from the short), or is it getting lower (you're moving towards the short).

(†) As kindly pointed out by glen_geek in a comment, there is a risk with this technique that if (a) the "short" is formed with a weak-enough material, and (b) the test current is high enough to "blow out the short", then this process of trying to find the short, might remove (or partially remove) it. That prevents you actually finding it, or assessing whether it may reform in future.

Therefore it would make sense to minimise this risk, by starting with the lowest current that still allows a difference in the voltage drop to be measured along the "shorted" tracks.

\$\endgroup\$
3
  • 2
    \$\begingroup\$ Yes, this does work. But applying enough current (like 1 Amp) to trace the current-path may be enough to blow out the short - been there,done that. Once its blown...it can't be found, and one wonders if in future the short may re-bridge. \$\endgroup\$
    – glen_geek
    Aug 12 '18 at 16:36
  • \$\begingroup\$ @glen_geek - Thanks for the helpful comment. I haven't been so unlucky to have that happen to me (I've found solder shorts using the above technique, which were not going to blown out by the test current), but I certainly see the risk. I'll update my answer to add that point. Again, thanks. \$\endgroup\$
    – SamGibson
    Aug 12 '18 at 20:09
  • \$\begingroup\$ Dont have that yet :) Its one of the things Id like to buy/build when funds become aviable in the future. \$\endgroup\$
    – Psvedman
    Aug 12 '18 at 20:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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