Cable differences are real

[mansr]

Cable differences are real, and they can be measured. Rummaging through my cable box, I selected a few and measured their frequency response up to 125 MHz using white-ish noise.

 

These are the cables I used:

• RG-58 coax, 0.5 m, 4.8 mm outer diameter, black
• RG-58 coax, 2 m, 4.8 mm, black
• No-name cheap-looking single RCA, 5 m, 2.6 mm, black
• No-name not-quite-as-cheap dual RCA, 10 m, 2.8 mm, black with red/white plugs
• No-name video (yellow plugs) single RCA, 1 m, 2.8 mm, black
• Maplin Pro dual RCA, 0.75 m, 5.9 mm, black + black/grey braid
• Maplin Pro dual RCA, 3 m, 5.9 mm, black + black/grey braid
• Maplin (non-pro) single RCA, 1.5 m, 3.7 mm, black with black/blue plugs
• AudioQuest Tower dual RCA, 1 m, 4.7 mm, black with grey stripes
• iFi branded dual RCA, 0.5 m, 3.7 mm, purple
• 21 AWG zip cord, 13 m, 2.0 mm, black

 

Below are spectrum analyser screenshots for each of these cables, in random order. Your task is to guess which spectrum goes with which cable, as many or as few as you like. It's also OK to simply point and laugh at the more silly-looking ones. The noise generator spectrum is inherently a little rolled off, so bear that in mind.

 

1. 
2. 
3. 
4. 
5. 
6. 
7. 
8. 
9. 
10. 
11. 

 

Have fun!

[mansr]

16 minutes ago, Arpiben said:

For the fun be complete do you mind sharing instruments used: Noise generator or AFG and Spectrum Analyzer ?

Both are part of a Tektronix MDO3024 scope.

[mansr]

1 hour ago, jabbr said:

Could you list the outside jacket diameters and colors for each cable?

I've updated the list with that info.

[mansr]

1 hour ago, Arpiben said:

how it has been measured @mansr knows

Each cable was connected directly from the AFG output to the spectrum analyser input on the Tektronix MDO3024. Any settings not shown in the screenshots are instrument defaults.

[mansr]

4 minutes ago, Ralf11 said:

BTW, did you guys know that Optical Cables are Directional?

That's because impurities in the glass act as tiny one-way mirrors.

[mansr]

4 hours ago, tmtomh said:

If I had to guess, I'd guess 9 or 11 is the Audioquest. But I freely admit that's based largely on my bias against Audioquest (although a bias based in large part on the poor measurements of some of their equipment).

I like the way you're thinking, whether your guess is right or wrong.

[mansr]

4 minutes ago, jabbr said:

My guess is that the cables with thicker and multicolored jackets will have worse performance  

That would be a fair assumption, but in this instance it isn't entirely accurate.

[mansr]

2 minutes ago, Ralf11 said:

price/foot on the cables will be a strong correlate of poor performance

I don't remember the cost of most, and several were "free," the Audioquest included.

[mansr]

11 hours ago, Arpiben said:

Clear that you didn't use any directional coupler, thanks.

Direct from AFG Output (BNC F) to RF Input (N F) it is simply not possible without adaptors.

On top of RF Input N->BNC  adaptor you probably also used  banana/BNC and RCA/BNC ones.

I obviously used adaptors as required, N-BNC and BNC-RCA, all 50 Ω. The AFG output and RF input are both 50 Ω as well.

[mansr]

The score so far is 6 correct guesses and 3 incorrect. You guys are doing well.

[mansr]

Just now, Arpiben said:

May I ask why you are AFG outputting Noise rather than Sine or Bursts?

It doesn't have a sine sweep option. I guess they want you to buy a separate AFG for a full feature set.

[mansr]

I'll make it (slightly) easier for you. Since several people have already guessed them correctly, I can confirm that 3 and 6 are indeed the RG-58 cables.

[mansr]

1 hour ago, tmtomh said:

I just re-read the descriptions and noticed that the zip cord is 13m long.

The best part about that zip cord is that I ordered 10 m. After using some of it, there's still 13 m left. Thanks, Farnell.

[mansr]

OK, I'll reveal one more. #4 is the 10 m generic RCA.

[mansr]

2 minutes ago, The_K-Man said:

HOW is this 'off-topic'?   Ave Maria...

The topic is the interconnect cables I tested, not speaker wire or music mastering.

[mansr]

I think we've waited long enough. Here's the list:

1. Maplin Pro dual RCA, 0.75 m, 5.9 mm, black + black/grey braid
2. iFi branded dual RCA, 0.5 m, 3.7 mm, purple
3. RG-58 coax, 2 m, 4.8 mm, black
4. No-name not-quite-as-cheap dual RCA, 10 m, 2.8 mm, black with red/white plugs
5. Maplin Pro dual RCA, 3 m, 5.9 mm, black + black/grey braid
6. RG-58 coax, 0.5 m, 4.8 mm outer diameter, black
7. No-name video (yellow plugs) single RCA, 1 m, 2.8 mm, black
8. AudioQuest Tower dual RCA, 1 m, 4.7 mm, black with grey stripes
9. 21 AWG zip cord, 13 m, 2.0 mm, black
10. Maplin (non-pro) single RCA, 1.5 m, 3.7 mm, black with black/blue plugs
11. No-name cheap-looking single RCA, 5 m, 2.6 mm, black

[mansr]

22 minutes ago, tmtomh said:

It's also interesting that - if I read the images right - the 1.5m and especially 3m Maplin cables' frequency responses are a bit more ragged than most of the others.

That's largely down to impedance mismatches. The long Maplin has an impedance of about 75 Ω whereas the AFG and spectrum analyser are 50 Ω. This results in reflections which cause varying amounts of cancellation. An RG-59 coax looks similar.

[mansr]

22 minutes ago, Jud said:

Can someone say why we should be looking for level response up to 125mHz, or am I reading the scale wrong?

Because that's where the differences are.

[mansr]

5 minutes ago, tmtomh said:

So then would all the cables' responses look different when used in an actual analogue audio interconnect situation?

Without impedance matched inputs and outputs, you'll always get some reflections. That's only a (potential) problem if the cable length is a significant fraction of the signal wavelength. The wavelength of a 100 kHz signal in a typical cable is 2 km. If you're wiring up a stadium, you should perhaps care. At home, it makes no difference at all.

[mansr]

8 minutes ago, d_elm said:

Could it be for 10MHz square waves for clock signals ?

Such signals should use properly impedance matched connections.

[mansr]

1 minute ago, Jud said:

Yeah, wondering about any possible effects of those differences.  Is the dip in the AQ response, for example (1) completely insignificant for audio, (2) at a frequency range where there are common sources of interference and thus possibly beneficial, or (3) the result of an intentional design decision in shielding, dielectric, material, etc., thought to be otherwise beneficial or neutral in the audio range?

1. It's at approximately 48 MHz, so entirely irrelevant for audio.

2. I have never heard of any particular problems near 48 MHz. Besides, the location of the dip (probably) depends on the length of the cable.

3. It is obviously somehow a result of the cable construction. Why they thought plain coax wasn't good enough is anyone's guess.

[mansr]

59 minutes ago, mevdinc said:

See what @mansr and others make of  the article below.

http://audiofi.net/2019/01/audio-engineer-claims-his-null-tester-settles-the-debate-on-wires/

I agree with Ethan Winer.

[mansr]

Let's look a bit more closely at reflections. One way to observe them is by sending pulses into one end of a cable, leaving the other end unconnected, and measuring what comes back. To this end, I have configured the AFG to generate 6 ns pulses. The AFG output is connected via a short RG-58 cable and a T coupling to the scope input (1 MΩ) and the cable under test. We start with a 2-metre RG-58.

 

The first pulse seen here is the feed from the AFG. The second is the reflection from the open end of the cable. Since the round-trip distance is 4 m and the signal takes about 5 ns/m, the delay is 20 ns.

 

If the far end of the cable is shorted instead, the reflected pulse switches polarity:

 

Terminating the cable with a matched 50 Ω load, the reflection pretty much vanishes:

 

A 75 Ω termination causes a small positive reflection:

 

In general, the reflection is positive if the terminating impedance is larger than the characteristic impedance of the cable, negative if it is smaller.

 

With a 4 m RG-59 cable, we get multiple reflections:

 

The reason for this is that the cable from the AFG has a characteristic impedance of 50 Ω while the RG-59 is 75 Ω. This causes the first reflection (the second pulse in the scope image) to be reflected down the cable with flipped polarity, then once more at the open end (preserving polarity) before we see it 80 ns after the initial pulse. This repeats to produce a small positive pulse at 120 ns, and so on, each time attenuated a bit more. The closer the two impedances, the smaller the reflection will be. At the extremes of open or short circuit, all energy is reflected.

[mansr]

Just now, semente said:

Worse can be euphonic?

Only if it makes an audible difference.

[mansr]

Time for another demo. This time we're using the 4 m RG-59 between the noise generator and the spectrum analyser. Remember, the impedance of both of these is 50 Ω. With the 75 Ω cable we will thus get a small reverse-polarity reflection in either end of the cable. If cable round-trip is an odd number of half wavelengths long (0.5λ, 1.5λ, etc), the reflections will be 180° out of phase with main signal and partially cancel it out. The cable is 4 m long, so the round trip is 8 m which is half a wavelength at 12.5 MHz with a signal speed of 5 ns/m.

 

 

As expected, there is a dip in the spectrum at 12.5 MHz and every 25 MHz above.