Effect of Stud Torque on Main Breaker Resistance

@JamesCH95’s recent topic on terminal testing reminded me of some testing I did almost a year ago. At one point, @sanddrag told me that tighter is better when torqueing the nuts on main breaker studs, and I wondered how true that is.

Thus, I found a main breaker and an old ring terminal, and I decided to find out.

The low-resistance measurement technique was as follows: using alligator clips, I connected the + output of a calibrated Keysight N5746A to the stub of wire coming out of the terminal. I clipped the - output to the stud on the breaker. I then likewise clipped the terminals of a calibrated Keysight 34465A multimeter to the wire and to the stud, using separate leads and clips. I then used the power supply to put 10 A through the connection, and took the voltage read by the multimeter (averaged over 10 s).

It looked basically like this, assuming the ASCII art works:

Power Supply Clip                                        Power Supply Clip
|                                                        |
|        Multimeter Clip                       MM. Clip  |
|        |                                     |         |
Stub of copper wire- - - -Ring Terminal- - - -Breaker Stud- - - -Breaker

Yes, you read that right: 10 A through alligator clips. It’s probably not ideal, but nothing caught on fire. And the beauty of Kelvin connections is that none of the alligator clip contact resistances affect the measurement.

The crimp resistance of the terminal to the wire does factor into the measurement, and I didn’t attempt to quantify that: I want to measure how the resistance changes with respect to torque, and the crimp resistance should be constant.

The experimental procedure was:

  1. Torque the nut down onto the terminal, using a calibrated CTECH1MR240 torque wrench. Record the target torque, as well as the maximum measured torque.
  2. Measure the resistance, as described above.
  3. Break the nut free, and record the torque required to do so.
  4. Back off the nut fully by hand.

I repeated this procedure for 6 different torque values. I then redid the test, but abraded the contacting surface of the ring terminal gently with Scotchbrite between iterations.

Here are the data:

Obviously, I didn’t have time to really do a thorough experiment. Some things I’d like to try, if I get around to it:

  • Several different breakers
  • Several different terminals
  • Apply Braycote to threads, to see what difference lubricated torque makes
  • Torque the studs until they snap, to figure out what the breaking strength is

This is great. We see a wonderful asymptote as the contacts approach fully coupled/fully contacted. We also see surface prep improving resistance and consistency.

Data with a simple mechanistic explanation, beautiful. :heart:


A difference of 0.05mohm between 12 and 60 in-lb tells me that I definitely don’t need to mark this high on my priority list. Excellent test.


Who knows, that 5 millivolts while you’re climbing might just make a difference!


This is very nice research. I’ll add though that the reason the above is probably true is about momentary disconnections rather than increased resistance. In shock-load scenarios like you might see on a robot crashing into a wall, a contact that’s normally well connected can come loose for a few milliseconds, which is enough to force a radio or roborio reset (which then takes a good portion of the match). If the connector is loose enough that you can rotate it by hand, it’s loose enough to disconnect momentarily. My rule of thumb I try to pass on as a CSA is all the connections in the battery → PDP/PDH circuit should be tight enough that you can’t move them by hand, then a bit more. And the more pre-load torque the connection starts with, the less likely it is to loosen over time from vibrations on the robot (though you should still check it every match or two to make sure it hasn’t come loose).


In summary: more ugga dugga = good

but how much is too much?

Very nice and clear data. The spec sheet from Andymark shows 50 in-lbs as max torque.

1 Like


1 Like

You can look at this and say “Hmmm, wide range of torques produces a negligible change in resistance” or you can say “the change in resistance from 12in-lbs to 20in-lbs is 14.5 times that of that from 50in-lbs to 60in-lbs”. (initial values, not Scotchbrite)

I’d be curious to see results for torques approaching freshman-doing-their-best levels (approx 1 in-lb). :grin:


But it’s almost 7% less resistance.

93% of negligible is still negligible.

“Doesn’t rotate when I yank on it” seems good enough to me here, assuming use of a nylon locking nut for vibration resistance.

The follow-up test I’d want to see is throwing the assembly on a shake table for an hour at different torques…

1 Like

Shortening a 6awg wire by 1.4" would provide the same reduction in resistance from the best and worst non-scotchbrite tests. That’s extremely little. It’s a difference of 5mV over 100A.


I hate to say it but the test method is faulty. The cable should not be retorqued ever! The proper technique is to cut the cable damaged or molded under the lug off, strip the insulation back and torque down for each test.


I’m not sure if you are intending to say that the entire ring terminal should be cut off after a single use; it sounds like you are referring to a situation where a bare wire is directly compressed by a screw terminal, like is common in household wiring. The only part of the cable that is ever deformed or stressed would be the part inside the barrel of the terminal and only during crimping. The ring part of the ring terminal gets clamped by the nut, but shouldn’t have any visible deformation. Perhaps testing repeated clampings might be interesting, but I think the results as-is say quite a lot.


My apologies for not reading that a ring terminal was used! I thought that bare copper cable was being torqued under a lug. However, I do stand by my statement for copper or aluminum conductors. That has been a topic that has been thoroughly discussed in an Electrical Code Facebook group that I belong to.

1 Like

Ah, the reasoning behind us specifically torquing the main breaker lugs to 50 in-lbs with the CTECH1MR240 torque wrench last year and then never using any sort of torque spec for anything else on the robot.

Harbor freight $20 torque wrench that’s +/-5% out of the box: am I a joke to you?

1 Like

I found the same inaccurate torque limiting feature in a 1/4" drive HF ratchet that broke the 1st time I used it. Was from a big set, swapped the whole set out a few weeks later.