Twisted pair cable attenuation

Attenuation per 100 m by frequency and category, and the three-term law behind it — each term a different physical loss mechanism.

Attenuation per 100 m at 20 °C

f (MHz)	Category 6	Category 5e	Category 6A
1	2.04	2.13	2.08
4	3.81	4.01	3.80
10	5.99	6.33	5.93
16	7.60	8.05	7.49
20	8.52	9.03	8.38
31.25	10.74	11.41	10.50
62.5	15.48	16.52	14.99
100	19.92	21.35	19.13
125	22.49	—	21.51
155	25.31	—	24.09
200	29.15	—	27.58
250	33.04	—	31.07
300	—	—	34.27
400	—	—	40.05
500	—	—	45.26

Values in dB per 100 m of solid horizontal cable. A dash means the frequency is above that category's specified range, where the coefficients are not valid.

The three-term law

α(f) = a√f + b·f + c/√f (dB per 100 m, f in MHz)

Each term is a different mechanism. The √f term is copper loss from skin effect, which dominates across the band. The linear term is dielectric loss in the insulation. The 1/√f term accounts for the shield where one is present and is negligible above a few megahertz.

Category	a	b	c	Specified to	Provenance
Category 6	1.8200	0.0170	0.20	250 MHz	Verified against a public source
Category 5e	1.9108	0.0222	0.20	100 MHz	Unverified, see note
Category 6A	1.8200	0.0091	0.25	500 MHz	Unverified, see note

Where these come from

This page is sourced differently from most of this site. TIA-568 and ISO 11801 are paid standards, so nothing here is quoted from them. Category 6 coefficients come from public IEEE 802.3 working-group documents, which reproduce the cabling specification for the benefit of the Ethernet standards work. Category 5e and Category 6A are widely published figures that could not be verified against a primary source, and are labelled that way rather than presented as equal. Your cable's datasheet beats all of them.

From cable to channel

A channel is the sum of its parts. Stranded cordage loses up to 50 % more per metre than solid horizontal cable, which is where the familiar 1.05 channel factor comes from: the standard 90 m + 10 m model is electrically 90 + 10×1.5 = 105 m of cable.

IL = α(f)·L_cable/100 + α(f)·1.5·L_cords/100 + N·k√f

Temperature

Insertion loss rises roughly 0.4 %/°C for unscreened constructions and about half that for screened ones. A cable in a hot ceiling void at 60 °C carries about 16 % more insertion loss than the same cable certified on a cool day. A bundle carrying PoE heats itself, so the temperature to design against is not the ambient one.