Effect of speaker cable on THD

[Alan0354]

I finally did the comparison of THD with speaker cables of different length and parallel. I compare to the bare amp.



This is taken from the output of the amp directly at the output transistor before even the cable going to the panel connector. This is the true distortion of the amp. Test frequency is 20KHz, amplitude is 40Vpp which is about 50W into 4ohm. The load is 4ohm. You should expect the THD going down to half for standard 8ohm. The measurement is 0.0055% which is pretty good.




This is measure on the load end (eg. at the speaker) with only a 3feet long Monster type 12gauge cable. THD up almost 10 times to 0.048%. I have to say just going through the connector measure about 0.019%, so the cable add about 0.03%. But you get the picture.



This is measure at the load end with 10ft monster like 12 gauge cable. THD is whopping 0.14%. This already is a big cable and 10ft is not uncommon in real system. You see how bad the cable make?




This is adding a 6feet 12 gauge monster like cable in PARALLEL to the 10 feet cable. Look at the improvement of THD by paralleling cables. THD back down to 0.065%



I hope this convince people that keep claiming you don't need big cables for speakers. That they pull out some charts by some amp company and claim it's ok to use 16 gauge cable if it is short. I long experience cable is one of the utmost important thing in a good system. I don't want to repeatedly writing my experience when I first bought my JM Lab speaker that was over 3 times the cost of my Kef and not sounding better until I use 4 pairs of monster like 12 gauge cables on each speaker.

Also, how do you judge amps with low THD if the cable mask the signal with distortion?

 

[Alan0354]

In case you complain I did the test at 20KHz. Here is the test at 5KHz

This is directly for output of transistor like before. I don't think you can sneeze on 0.0018% for 4ohm load, 50W output into 4ohm load with 31V rails.




This is with 10ft monster 12 gauge cable. Need I say more.......other than 0.0156%



OK, I say more. The most offensive is it's the odd harmonics the cable introduces. Look at the higher odd harmonics.

 

[BinaryMike]

It is essentially impossible for pure copper wire and resistance wire (inside the 4R load resistor) to distort the signal because pure metals don't exhibit nonlinear resistance. That's why wirewound resistors are the standard of excellence. It's far more likely that a poor connection is responsible for the nonlinearity. I could also imagine thermal modulation in the wire being responsible, but that seems unlikely at such high frequencies. Other remote possibilities: mechanical/acoustical interaction, inductive coupling.

 

[Alan0354]

I used gold plated banana plugs. How do you explain that 10ft is a whole lot worst than 3ft? I cannot emphasis enough. I am using Monster cables, 12 gauge. I pay money to get a 100ft roll to build the cables for my amp comparison test I did before and posted here. I spare no money to do this. I spent over $150 and a few days of building for the amp comparison test and things are not useful after that.

As I said, I heard the difference 20 years ago when I bought my JM Lab, I almost cry when I first put it on to replace my Kef floor standing Uni-Q. Every pair of 12 gauge I added, it open the sound more until the 4th pair. It's the airyness, openess and the biggness that comes out with more pairs of cables. That was with the Acurus 200X3 amp only.

I am an EE for so many years. I was skeptical about this before, I know how you feel. It's the result in the hearing test that I cannot dispute. Now I have the empirical data to support this. This is scientific.

 

[BinaryMike]

I used gold plated banana plugs. How do you explain that 10ft is a whole lot worst than 3ft?

One possible explanation is nonlinearity at the load end of the cable, assuming your measurements were taken at that end.

 

[for_p1]

I used gold plated banana plugs. How do you explain that 10ft is a whole lot worst than 3ft? I cannot emphasis enough. I am using Monster cables, 12 gauge. I pay money to get a 100ft roll to build the cables for my amp comparison test I did before and posted here. I spare no money to do this. I spent over $150 and a few days of building for the amp comparison test and things are not useful after that.

As I said, I heard the difference 20 years ago when I bought my JM Lab, I almost cry when I first put it on to replace my Kef floor standing Uni-Q. Every pair of 12 gauge I added, it open the sound more until the 4th pair. It's the airyness, openess and the biggness that comes out with more pairs of cables. That was with the Acurus 200X3 amp only.

I am an EE for so many years. I was skeptical about this before, I know how you feel. It's the result in the hearing test that I cannot dispute. Now I have the empirical data to support this. This is scientific.

This is interesting observation. Can you repeat it at output terminal of your amplifier? That way you will exclude output circuit city likely coil and relay there. What was the load type you use?

 

[Djcoolray]

Thanks for the information.....

So running parallel cables works better than a single cable to the speakers. I say it's something new to checkout and see for ourselves. So much for the same shoot'em up and the same rodeo !

 

[ConradH]

Can you modify the load resistor to look more like a speaker using one of the models published for the purpose? I'm guessing the results will be a lot worse, though I don't understand why you're getting what you're getting. I gotta try this!

 

[Alan0354]

Sorry I should have explain the testing a little more. This is really a continuation of my thread on my amplifier. The detail is explained in post 4 where you see I measured THD and different point from output of the transistors to the banana connectors. Big difference even between inside of the connector to the outside of the connector.

You see picture of the output part of my amp, the wires are so big, two 8" 10 gauge red wires from the output of the pcb to connector added so much distortion.

I want to show you how much effort in doing this and amp comparison.

This image show the cables I used for measuring THD today, everything is with gold plated banana plugs shown, notice the big box of monster like 12 gauge cable, I bought that just for the amp comparison test I posted before and I used the same cables made for today's testing. No mickey mouse at all here.



This picture shows I use 4 pairs of those monster cables from each speaker. I separate two for the woofers and two for the mid and tweeter.



The cable really make a big difference if you have a 4ohm speaker like mine. I am convinced that the cable was the bottle neck of my amp comparison test last time. All the effort designing an amp with very low distortion doesn't mean anything with those speaker cables. Mind I used 4 pairs from the relay to the speakers that have to travel long distance.


This is really bad news for me, not an achievement on this finding. This means this is pretty much the end of the road for my amp design. I see no point in keep designing different amps as it might not matter. From my comparison experience, I highly question question the accuracy of people's opinion that they can hear the difference. Until people gone through the extend I've done in building the big relay switching fixture that can go back and fore, huge cables for speakers, you cannot even start to compare. For more info, refer to this thread:
http://audiokarma.org/forums/index.php?threads/does-a-good-and-expensive-amp-matters.757185/
http://audiokarma.org/forums/index.php?threads/does-a-good-and-expensive-amp-matters.757185/ Post 52 show the test fixtures I built.

 

[Alan0354]

One possible explanation is nonlinearity at the load end of the cable, assuming your measurements were taken at that end.

Yes. I literally have a dedicated gold plated connector cable connecting to the QA401 that I plugged into the spot I measure. I just plug onto the dummy load end. No short cuts.

 

[Alan0354]

Can you modify the load resistor to look more like a speaker using one of the models published for the purpose? I'm guessing the results will be a lot worse, though I don't understand why you're getting what you're getting. I gotta try this!

Yes, I think you should try this to confirm. My experiment is repeatable. Just too bad you live so far away, or else we can do it together. You can look and use all my cables.

I am sure if the load is reactive like speakers, it can only get worst.

The worst is you can see the new odd harmonics that are not there when I measure at the output of the transistors.

 

[Alan0354]

It is essentially impossible for pure copper wire and resistance wire (inside the 4R load resistor) to distort the signal because pure metals don't exhibit nonlinear resistance. That's why wirewound resistors are the standard of excellence. It's far more likely that a poor connection is responsible for the nonlinearity. I could also imagine thermal modulation in the wire being responsible, but that seems unlikely at such high frequencies. Other remote possibilities: mechanical/acoustical interaction, inductive coupling.

It's not the resistance, it's the inductance and capacitance............nah, it's not capacitance as even 500pF won't make a dent on a 4ohm load. More like inductance or something I don't know. Remember the speaker is 4ohm, distortion is say 0.01% from the cable. All it takes is change of 4 X 0.0001=0.4mohm to create the distortion. Let's just look at the effect of inductance. Say at 20KHz, calculate with inductance to have a change of 0.4mohm to create 0.1% variation(just an example).

X=2 /pi f L or L=X/(2 /pi f) where X is the reactance of the inductance of the wire. X=0.0004ohm, f=20KHz, so L=0.0004/(6.28 X 20000Hz)=3.18nH. I am pretty sure the inductance of a few feet of wire is more than that. Someone might want to verify this. Now, the inductance can be a funny thing as this is a wire. So all it takes is reactance of the inductance to vary and you get 0.01% variation.

I am sure there are other factors I don't even know, but this kind of give an idea how little it takes to change 0.01%.

 

[gridleakrick]

What instrument are you using to measure the distortion?
Rick

 

[Alan0354]

Can I ask, have you compared to a spade connection or bare wire? Bananasare great for quick plugging and unplugging but may not be the best, in general.

Actually it's a very good idea. I am going to put two set of banana connectors per channel on my amp so the connector is less of the bottle neck. But using spade might be even better. I intentionally drill the spacing a little off so it's tight to plug in for better contact, but still not as good as spade.

But now, the connector is definitely not the main source of problem, I demonstrated that the 10ft cable is a whole lot worst with the same amount of connectors tightly fitted.

 

[Alan0354]

What instrument are you using to measure the distortion?
Rick

It is QuantAsylum QA401:https://quantasylum.com/collections/frontpage/products/qa401-audio-analyzer

But I don't think it is the bottle neck as I can measure much better result by measuring in other spots.

 

[loudnoises]

Endless questions Alan, but here's maybe a simple one.

Can you compare the same 3' (or 10') length of monster cable with the conductors removed from the outer jacket, and separated?

Thinking about those folks who like to run one wire for pos and one for return, often having them lifted up on little risers.

 

[BillWojo]

I don't like banana connectors, they are basically a barrel shaped pin going into a straight round hole. There is only point contact connection being made. I use spade or ring lug as used forever in industrial applications. I can't imagine the amount of failures that would occur if control systems were wired together with banana plugs and sockets.
They look cool, that's where it stops for me. For the most reliable, gas tight, low resistance connection I use ring lugs.

BillWojo

 

[ConradH]

Bill is right, there are bananas and there are bananas, and not all are created equal. I use them all the time with no effect on measured THD, but I'm only using good lab quality ones, not imports and audiophile stuff. Ring lugs under a nut are hard to beat and I use those for metrology work.

Pure capacitance or inductance shouldn't affect the results much, but if it changes, if there's a voltage coefficient or if the current flow affects the wire spacing, that might do it. Just speculating. I've had output inductors affect distortion in blameless amps, and it doesn't take much core to do it. Only air core inductors have no effect. Even winding an inductor on a resistor body, as is common, is no good. I assume any other inductor in the output loop would be subject to the same effect. This stuff is pretty easy to test on the bench with short heavy leads and by adding known components to the loop.

 

[Ray Gianelli]

Makes a good case for powered speakers.

 

[for_p1]

Bill is right, there are bananas and there are bananas, and not all are created equal. I use them all the time with no effect on measured THD, but I'm only using good lab quality ones, not imports and audiophile stuff. Ring lugs under a nut are hard to beat and I use those for metrology work.

Pure capacitance or inductance shouldn't affect the results much, but if it changes, if there's a voltage coefficient or if the current flow affects the wire spacing, that might do it. Just speculating. I've had output inductors affect distortion in blameless amps, and it doesn't take much core to do it. Only air core inductors have no effect. Even winding an inductor on a resistor body, as is common, is no good. I assume any other inductor in the output loop would be subject to the same effect. This stuff is pretty easy to test on the bench with short heavy leads and by adding known components to the loop.

You may be getting to something here. If real speakers were used as load - they contain inductors with ferrite core. They are known to be non-linear. That would show significant raise of harmonics. Also soldering quality is important - poor soldering can result in non-linear conductivity. I feel the need to replicate this experiment myself.