HDMI Investigated - Are expensive cables a scam?

What's the difference in performance between a £5 HDMI cable and a £100 one? We find out

8 Feb 2011


Manufacturers use two key techniques to ensure the quality of their HDMI cables. Bit Error Rate (BER) tests use a signal generator to output an HDMI signal, which is sent down the cable. The other end is connected back to the test machine, which compares the data sent and received to identify errors.

Any more than one error per billion bits – which equates to somewhere between one and ten per second, depending on the data rate – means the cable doesn't meet the specifications of the HDMI standard.

According to documentation published by AudioQuest, errors increase sharply as the cable exceeds a certain length. For example, a 19m cable may exhibit only one error per minute, but this rises to one per second at around 22m. At 23m, errors measure in the tens of thousands per second.

So why does the signal degrade so dramatically? The other test that manufacturers use provides some insight. The eye pattern test uses an oscilloscope to examine the signal on individual pairs of wires in an HDMI cable. In theory, a digital signal should appear as a square wave as the voltage jumps from maximum to nothing to represent digital ones and zeros. However, limitations of the cable and other components cause the verticals to become sloped, giving an eye-shaped pattern as seen on the image above. With increasing interference the eye-shaped pattern becomes messier to the point that the shape is lost completely.

As long as there is empty space in the centre of the oscilloscope – as long as the 'eye' is open – the signal will be successfully transmitted. However, if interference is so great that the eye closes, either vertically due to a lack of signal amplitude, or horizontally due to timing jitter, errors will occur.


The eye pattern test makes it easier to understand why some cables introduce hardly any errors while others produce a lot. It suggests that it's statistically unlikely that a cable will suffer just enough interference to introduce significant but not critical levels of errors. It also helps to explain why higher data rates might cause a cable to fail – with a higher rate, the eye is compressed horizontally, increasing the chances of it closing and introducing errors.

These tests make compelling arguments for both camps. BER and eye pattern tests are important for product development and quality control, but this equipment is expensive. The Tektronix DSA8200 used for eye pattern tests costs $140,000. Premium HDMI cable manufacturers argue that the quality control they employ can't be matched by companies that sell cables for just a few pounds each. However, the tests themselves indicate that there's a very small margin between a cable introducing one error per second and tens of thousands. This supports the theory that, for a given data rate at least, a cable is likely either to work well or fail completely.


To test the cables we played two Casino Royale Blu-ray discs in PlayStation 3 consoles. We used an Onkyo TX-SR308 receiver and Q Acoustic Cinema 2000 speakers to test audio and two Panasonic TX-P42V20B TVs for video. You can see which cables we tested in the table (top right).

If we were to see any differences in quality between the various cables, we'd have expected the Ecosse cable, which is 1m long, very expensive and received glowing reviews elsewhere, to perform the best. The cheap Premier AV cable is 10m long and we left it coiled on top of a four-way mains extension lead, so it should have performed the worst.


We had nine volunteers for our audio tests and 10 for our video tests. For each test, we asked our guinea pigs to decide which speakers sounded best, or which TV had the better picture. They were given three options: A, B or No difference. The results appear in the table (middle right).

If we remove Test 3 from each set – where we used identical cables – we can see how the perceived performances of the cheap (Linx and Premier AV) and expensive cables differ in the table.

Read more