Derating a wire for high freq

[rdelauter]

I have been trying to find a formula or table on how much to derate a wire or cable when the freq increases in the conductor. I have a bonding wire that has 25 amps of current at 8kz. I want to go back to the manufacture with some more information about the problem and a possible solution, however, I can not locate a derating factor for this. Any help or suggestions. The wire is a 10AWG.

 

[petersonra]

I have been trying to find a formula or table on how much to derate a wire or cable when the freq increases in the conductor. I have a bonding wire that has 25 amps of current at 8kz. I want to go back to the manufacture with some more information about the problem and a possible solution, however, I can not locate a derating factor for this. Any help or suggestions. The wire is a 10AWG.

I have a Xerox of an article that talks about derating for various size cables and frequencies.

It claims the skin effect coefficient is related to the product of the cir. mils of the conductor and the frequency. There is a chart with it that gives the coefficients for various products.

10,000,000 = 1.000
20,000,000 = 1.008
30,000,000 = 1.045
.
.
.
80,000,000 = 1.138
90,000,000 = 1.195


#10 is about 10380 circ mils, so the product is about 83,000,000.

Using the coefficient for 80,000,000 gives you a current carrying capacity of 1/1.138 or about 88% of the capacity at DC.

For the 90,000,000 product, the current carrying capacity is 1/1.195 or about 83% of what it is at DC.

BTW, why would a bonding wire carry 25A of current?

 

[rdelauter]

Not sure why it is carry current, we are asking the same question with no response from the manufacture. We believe maybe induced from the nearby filter chokes, the inverter freq is 8kz, and the current is 8KZ, not a coincidence.

 

[winnie]

The entire VFD system is capacitively coupled to ground, and _must_ carry _some_ current at the switching frequency, even with everything working perfectly and no leakage through the insulation system.

I am surprised as all get-out by the _25A_ number, but you haven't told us anything about how large the drive is. Also, this capacitive coupled current should be _very_ 'spikey', with jolts of current right at the switching transitions. Because this is such an ugly waveform, it is quite possible that your current measurement is being skewed rather higher than reality.

-Jon

 

[rdelauter]

It is actually a 120 KVA UPS system and did not look at the waveform. I will the next time I get to a site that has one of those units but not sure when that will be. Thanks for the input.

 

[ramsy]

There is a remark in the extensive help section of the Electrical Design Reference (EDR), a shareware App. with 30 day trial. http://www.edreference.com/

If the wire is smaller than #2 AWG, the skin effect is usually negligible and can be ignored for practical design problems. Otherwise, designers working with high-harmonic environments are advised to turn to more sophisticated calculations such as Neher-McGrath.

 

[Mike03a3]

There is a remark in the extensive help section of the Electrical Design Reference (EDR), a shareware App. with 30 day trial. http://www.edreference.com/

Originally Posted by EDR
If the wire is smaller than #2 AWG, the skin effect is usually negligible and can be ignored for practical design problems. Otherwise, designers working with high-harmonic environments are advised to turn to more sophisticated calculations such as Neher-McGrath.

That is not an unreasonable statement for 60Hz, where the skin depth of copper wire is greater than the diameter of #2. However, it is certainly not true at 8KHz, which is the freqency we are talking about in this thread. I calculate the skin depth of copper to be ~0.738mm at 8KHz, significantly less than the 6.543mm diameter of #2 or the 2.588mm diameter of the OP's 10AWG.

 

[Smart $]

That is not an unreasonable statement for 60Hz, where the skin depth of copper wire is greater than the diameter of #2. However, it is certainly not true at 8KHz, which is the freqency we are talking about in this thread. I calculate the skin depth of copper to be ~0.738mm at 8KHz, significantly less than the 6.543mm diameter of #2 or the 2.588mm diameter of the OP's 10AWG.

Perhaps. But you are assuming the conductor to be solid. Do the math for stranded. Diameter of an individual strand of 7-strand #10 is 0.0385" (0.9779mm). Skin depth of ~0.738mm @ 8kHz x 2 > 0.9779mm. Skin effect is negligible.

 

[LarryFine]

Perhaps. But you are assuming the conductor to be solid. Do the math for stranded. Diameter of an individual strand of 7-strand #10 is 0.0385" (0.9779mm). Skin depth of ~0.738mm @ 8kHz x 2 > 0.9779mm. Skin effect is negligible.

Yes, but can we ignore the intimate contact among strands? Wouldn't a 7-strand conductor have the effective cross-section of, say, a 6-leaf clover?

 

[Smart $]

Yes, but can we ignore the intimate contact among strands? Wouldn't a 7-strand conductor have the effective cross-section of, say, a 6-leaf clover?

Well, to get slightly more technical, no we can't ignore the intimate contact. But at the same time, we can't ignore the increase in surface periphery, inner spacings, and the reduction in core heating (more and smaller cores). Yes, a litz-wire bundle (i.e. insulated conductor strands) would be more effective, as currents would not be able to cross among the strands and migrate outward.

Admittedly, my understanding of skin effect is very weak, and I stand to be corrected as necessary...

 

[Mike03a3]

Perhaps. But you are assuming the conductor to be solid. Do the math for stranded. Diameter of an individual strand of 7-strand #10 is 0.0385" (0.9779mm). Skin depth of ~0.738mm @ 8kHz x 2 > 0.9779mm. Skin effect is negligible.

It would be if it was Litz wire, with insulated strands. However, when it isn't insulated the 7 strands are in physical contact, allowing the current to jump from strand to strand and continue to favor the outside of the bundle. This is called bundle effect. This could be mitigated somewhat if the strands were braided, but 7-Strand #10 is just six strands surrounding a center strand. There are additional complexities involved in calculating the effective conductivity of stranded wire caused by the voids between adjacent strands, which is why there are so many type of "stranded" wire for various applications.

.