3
\$\begingroup\$

If the positive and negative terminals of certain DC supplies are brought together under certain conditions, the wires will "stick together" with a slight force, feeling almost like they are magnetically attracted. The force is perhaps only slightly stronger than the weight of a paper clip, but it's obvious if the two wires are hanging freely when they are pulled apart.

This effect can be easily reproduced with a bench supply. 12V and a current limit of about 2A will do. I get good results with alligator or banana clips. Even 12V with a current limit of 100mA seems to be enough. The issue here is the response time of the supply, which is another variable. Perhaps 100mA is meaningless in this case, because upon contact, a much larger current has already passed? If I try 3V and 4A, the effect seems weaker.

I assume this is the same phenomenon that causes broken filaments in incandescent bulbs to re-stick once contact is made. When conventional car headlights "burn out", often the filament is left dangling and can be temporarily restored by banging on the headlight or slamming the hood. The headlight may then work until car shuts off, or even days after.

I also assume this is the same phenomenon that damages relays that close under high loads.

What is the mechanism behind this?

test setup

I am doubting magnetism. My guess is micro-arcing (welding).

My reasoning:

  • Once the two wires are connected, they are essentially one wire. I'm not keeping them running along each other in parallel. They are making a large, 1 meter circular loop to the supply. Keep in mind that the wires themselves do not appear to attract or repulse, it's just the ends.
  • If I make contact and then move the wires along each other (while keeping contact), I no longer notice an attractive force when separating them.
\$\endgroup\$
  • \$\begingroup\$ You can calculate the Magnetic field of a long wire and use F=BIL to work out the magnetic force between the wires. Does it seem to be of the same magnitude as you feel? \$\endgroup\$ – Andrew Morton Dec 4 '18 at 16:34
  • \$\begingroup\$ Reminds me of the wire around the nail thing we would do in high school. Depending on how the amount of turns that nail had, the force would become stronger and it would become more of a resilient magnet. \$\endgroup\$ – KingDuken Dec 4 '18 at 16:38
  • \$\begingroup\$ For wires, the force is magnetism. Relays get damaged by arcs that form during disconnects. This includes momentary disconnects as a result of contact bounce when the circuit is closed. DC is much harder on switch contacts than AC. When the AC waveform passes through the zero current point, any arcs will be extinguished, and will not reform across an air gap. \$\endgroup\$ – mkeith Dec 4 '18 at 16:50
  • 1
    \$\begingroup\$ Bort, your observations and imagination are important. Rather than just asking others at the outset, you did a variety of tests and have taken time to think on your own. It's wonderful that you took that opportunity to think about the world. Now, re-read your writing and distill out what you think are the significant observations -- not just ones that appear to be mutually supporting. (Also, don't assume some are meaningless and discount them by not writing them down. Keep everything on the table and in view.) Come up with new experiments to try. It's not as simple as you've been told so far. \$\endgroup\$ – jonk Dec 4 '18 at 17:04
  • \$\begingroup\$ magnetic force calculator \$\endgroup\$ – Andy aka Dec 4 '18 at 19:22
2
\$\begingroup\$

What is the mechanism behind this?

Microwelding.

The point of contact between the two leads has a significantly higher resistance than the wires themselves, or their connection to the power supply. Passing a current through this circuit will cause heat to be generated at that point of contact, melting some of the metal around that point and causing the leads to stick to each other.

If you're using an inexpensive bench power supply, I wouldn't be surprised at all if its current limiting were not responding fast enough to a sudden short like this. The actual current that flows when the leads first make contact may be significantly higher than the set limit -- depending on the exact architecture of your PSU, it's possible that its output smoothing capacitors are being discharged directly into your test leads, bypassing current limiting entirely.

\$\endgroup\$
0
\$\begingroup\$

The aligned magnetic fields of closely spaced wires serve to draw them together, but only if the currents are running in the same direction. In the case of a loop, where the current is running one direction in one wire and the other direction in the other, that force will push them apart. The amount of force you're describing sounds about right.

This is not how incandescent bulbs "self-heal." In that case, the loose, unconnected filament is disturbed mechanically, and if it closes the circuit, the resulting momentary arc can "weld" the filament back together. It doesn't last long, though, since that weld is generally now the weak spot on the filament.

Relay contacts suffer from initial arcing, bounce arcing, and particularly release arcing. When switching an inductive load, release arcing can be significant.

\$\endgroup\$
  • \$\begingroup\$ What are you concluding? And what specifically are you saying in relation to magnetism? Once the two wires are connected, they are essentially one wire. I'm not keeping them running along each other. Keep in mind that the wires themselves do not appear to attract or repulse, it's just the ends. \$\endgroup\$ – Bort Dec 4 '18 at 19:24
  • \$\begingroup\$ I read the updated post...I didn't understand originally, I thought you were seeing the wires attracted to each other as the current ran through them. The other answer is correct, you're seeing microwelding, and it's the same thing that happens in an incandescent bulb. \$\endgroup\$ – Cristobol Polychronopolis Dec 4 '18 at 21:29

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.