Zip cord for speaker test

John & Pano,
because in the previous cable thread Pano mentioned to have little time at the moment for experimenting, I proposed to compare different cables tied to one amp to do a listening test as you proposed to check whether the image stayed exactly between the speakers.
Two things happened afterwards:
A) Despite his busy job, Pano could find some spare time and started this thread, already showing some results with several initial tests.
B) I did my test as promised, found a revealing difference, but stumbled much to my surprise on having LS cables (from a well know brand) with a huge capacity, but despite this producing a perfect sound image.

Now because of this very unexpected extreme cap value I'm going to dig deeper to find out more details about this cable, but this deviates quite a bit from your objective with Zip cords.
My question therefore is, should I start a new thread or do you prefer to keep my testing results in this thread.

Hans
 
Hans,
I just put a spreadsheet together to calculate L and C vs cable RFZ.
A cable with a dielectric coefficient of 5 with an RFZ of 1 ohm has a capacitance of 2273 pF per foot, 10 feet is about 22 nF.
C=2273.8/Z


All I could possibly think of is a zobel.
Jn

edit: titanium dioxide has a permittivity of 80, and an 8 ohm cable made with that as the insulator would have a capacitance of 103 nF for a 10 foot length. Just don't know if a cable could be made with it.
 
Last edited:
Any chance of splitting one set and re-trying IMD? Or do you want to keep the zips together for the listening test?
Actually, if you regroup one set of zips with two zips connected to pos and two to neg, you duplicate the split geometry. And retain the construction for the later listening.

jn
 
I am open to either. Both make sense on the topic we are discussing.
I will be happy to engage either way.

John

O.K. As long as it may help to trigger new ideas, here is part one.
With my VNA I measured Zo of my LS cable with Zo=sqrt(Zopen*Zshort).
Image below shows that Zo is quite a bit different when comparing it to coax or Zip cord.
At LF Zo is significantly lower, quite comparable to my ESL63 in the Audio range and at HF Zo keeps rising.
Tomorrow I will show the ESL's impedance for the same FR and try to calculate the Reflection coefficient rho=(Zl-Zo)/(Zl+Zo).

Hans
 

Attachments

  • Z0 MH770.jpg
    Z0 MH770.jpg
    60.7 KB · Views: 239
The first 3 sentences clearly explain why that formula is good.

Well, except he used lightspeed instead of actual wire prop velocities, but he stated that in the 3rd sentence.

Remember, nothing he does goes out past ten microseconds or so.

As I stated in this thread (again), the t-line analysis falls apart after 10 or 20 microseconds. It is because it is a short form analysis. But 20 microseconds is enough. The short form used is a perfect match to a 200 element distributed RLC model.

jn
 
Last edited:
Administrator
Joined 2004
Paid Member
Actually, if you regroup one set of zips with two zips connected to pos and two to neg, you duplicate the split geometry. And retain the construction for the later listening.
Not quite sure I know what you mean there. So far I've cut only four 5m runs. I'll drag them up the mountain with me for further tests and measurement. I suppose another four runs of 5m will need to be cut for the listening tests.
 
Here's the second part with measurement results.
First image shows the Impedance of my ESL63 speaker together with the cable impedance in one plot.
When matching impedances is important, they look as being made for each other.

The second image shows the reflection coefficient, where Cyril Bateman found the cause of amplifier destruction in specific cable/speaker situations.
As can be seen, no Zobel network at the LS side will be needed where from 60Khz to 6Mhz the Refl. Coeff is slightly negative meaning harmless minor reflections in opposite phase, where in Batemans paper it was almost +1 without R/C termination.

Since I subjectively prefer the sound of my LS cable above a simple Zip cord, it could be that matching impedances between cable and speaker is more important than I ever thought.
So if true, reflections must be playing an important role.

Hans
.
 

Attachments

  • Cable vs Speaker Impedance.jpg
    Cable vs Speaker Impedance.jpg
    85.3 KB · Views: 332
  • Reflection Coefficient.jpg
    Reflection Coefficient.jpg
    69.6 KB · Views: 335
Here's the second part with measurement results.
First image shows the Impedance of my ESL63 speaker together with the cable impedance in one plot.
When matching impedances is important, they look as being made for each other.

The second image shows the reflection coefficient, where Cyril Bateman found the cause of amplifier destruction in specific cable/speaker situations.
As can be seen, no Zobel network at the LS side will be needed where from 60Khz to 6Mhz the Refl. Coeff is slightly negative meaning harmless minor reflections in opposite phase, where in Batemans paper it was almost +1 without R/C termination.

Since I subjectively prefer the sound of my LS cable above a simple Zip cord, it could be that matching impedances between cable and speaker is more important than I ever thought.
So if true, reflections must be playing an important role.

Hans
.
Ah, MIT-MH-770, I couldn't find "LS", so had no idea. That cable has that bol on the end, I'm sure it is a high capacitance cable set for low impedance in the audio band, and a zobel on the end to prevent the amp from seeing a capacitive load when the speaker climbs to high impedance.

Seems the high capacitance reading thing is solved.

Thanks Hans :up: Can you explain how you measured the impedance of the ESL and the cable? You are able to measure very high in frequency. Can you also explain the reflection coefficient?
if you go back to that paper referenced earlier, Cable Impedance he has some nice diagrams showing what happens in the line as a result of a reflection coefficient.
When the load is below the impedance of the line, the reflection coefficients (negative values) at both ends of the cable conspire to make the load current look like a slowly rising RC waveform, although it doesn't cleanly fit an exponential it does look like one. If the load is much higher than the line, the reflection coefficient is positive, and the resulting waveform oscillates. (the second graph)

When a step signal (0 to 100%) is applied to the line by the amp, that signal travels down the line at prop velocity. If it hits a short at the end, the reflection will be -1, and there will be a step signal of opposite polarity returning to the amp, the line voltage will become zero behind the returning signal. (the third graph)

jn
 
Last edited:
It is both. If you examine Cyril Bateman's papers, you see that he has done both, calculated the coefficient for a speaker line and a speaker load, and measured it by use of a reflection bridge.
Hans used a VNA to measure.

If you look at the high frequency reflection coefficient he posted, note that it doesn't stray too close to +1 at very high frequencies. If remained high positive at a frequency where the amplifier gain goes through unity, the amplifier could oscillate. A zobel is used to keep that reflection coefficient low, as close to zero as possible at the high frequency.

jn
 
Last edited:
Thanks Hans :up: Can you explain how you measured the impedance of the ESL and the cable? You are able to measure very high in frequency. Can you also explain the reflection coefficient?

Hi Pano,

Basically a VNA puts a resistor R in series with a the DUT with impedance Z.
When the voltages on both sides of the resistor R are called resp. A and B, then B=A*Z/(Z+R), or Z=R*B/(A-B), that’s how the impedance is measured for resp open cable, shorted cable and LS.
Cable impedance is then calculated with Zo=Sqrt(Zopen*Zshort)
The reflection coefficient is calculated from the first image, being Rho=(Zl-Zo)/(Zl+Zo)

Hans
 
Last edited: