IPB

Welcome Guest ( Log In | Register )

2 Pages V  < 1 2  
Reply to this topicStart new topic
Why does the gauge of speaker wire matter?
Pio2001
post Oct 30 2003, 20:07
Post #26


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



QUOTE (lucpes @ Oct 30 2003, 02:43 PM)

File not found !

It is very important to know how much octaves the impedance falls to 1 Ohm for.
Go to the top of the page
+Quote Post
lucpes
post Oct 30 2003, 21:50
Post #27





Group: Members
Posts: 517
Joined: 9-October 01
Member No.: 254



QUOTE (Pio2001 @ Oct 30 2003, 07:07 PM)
QUOTE (lucpes @ Oct 30 2003, 02:43 PM)

File not found !

It is very important to know how much octaves the impedance falls to 1 Ohm for.

http://www.rageaudio.com.au/kappa9.jpg

Oops! fixed in the original post too... and the impedance range is more like 0.8 to 8.something ohms

This post has been edited by lucpes: Oct 30 2003, 21:51
Go to the top of the page
+Quote Post
Pio2001
post Oct 31 2003, 00:01
Post #28


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



Before starting, how do we know that this is the amplitude diagram and not the phase diagram ?
Go to the top of the page
+Quote Post
lucpes
post Oct 31 2003, 14:00
Post #29





Group: Members
Posts: 517
Joined: 9-October 01
Member No.: 254



ABX result of 2xCAT5 cables per polarity (4x8 wires/speaker) versus thicker 'lamp' cord (sorry didn't measure any gauge/diameter) (both 3m long)

playback: 'inherited' MPC (Dire Straits - My Parties) -> FB2k (24bit/dither/ASIO) @-12dB/replaygained -> M-Audio Delta 410 directly to poweramp (193W sinus power in 4Ohm)-> cables in question-> Infinity Reference standard II speakers (don't have impedance curve but I guess it varies from a near-short to 8-9Ohms).

Recorded at one speaker binding posts using 20Kohm & 1Kohm resistors to adapt the signal to line-in levels for Delta 410 (http://www.hut.fi/Misc/Electronics/circuit...er_to_line.html). Needed to make sure that the speakers are sucking the amp dry *- attenuation was a bit too much for the line in (the sound is a bit grainy) - noise level was around ~60-70dB in Sound Forge 6, signal levels around -20dB, 24bit 88khZ. Removed DC offset, normalized to ~93%, resampled to 44.1 using Anti-Alias filter/4, noise shaped & dithered down to 16bit.

* that's at normal-to-medium loud listening levels

-------------------------------------
WinABX v0.4 test report
10/31/2003 14:49:30

A file: E:\30sec_2x2_cat5.wav
B file: E:\30sec_big_cable.wav

14:49:48 1/1 p=50.0%
14:49:57 2/2 p=25.0%
14:50:03 3/3 p=12.5%
14:50:16 4/4 p=6.2%
14:50:56 5/5 p=3.1%
14:51:03 6/6 p=1.6%
14:51:16 7/7 p=0.8%
14:51:26 8/8 p=0.4%
14:51:46 test finished

Should I upload the files in case someone is interested? The difference is more than OBVIOUS!

edit: files uploaded http://www.hydrogenaudio.org/forums/index....showtopic=14732

edit2: corrected resistor values; added link to speaker to line-level adaptor

This post has been edited by lucpes: Oct 31 2003, 22:49
Go to the top of the page
+Quote Post
lucpes
post Oct 31 2003, 15:30
Post #30





Group: Members
Posts: 517
Joined: 9-October 01
Member No.: 254



QUOTE (Pio2001 @ Oct 30 2003, 11:01 PM)
Before starting, how do we know that this is the amplitude diagram and not the phase diagram ?

That is an older version of the LMS (loudspeaker measurement system) software. When you clicked on the phase button (right of impedance) it would display the phase graph. The proggie didn't excel in usability, but that is the impedance graph in the link.
Go to the top of the page
+Quote Post
Continuum
post Oct 31 2003, 18:32
Post #31





Group: Members
Posts: 473
Joined: 7-June 02
Member No.: 2244



QUOTE (lucpes @ Oct 31 2003, 03:00 PM)
Should I upload the files in case someone is interested? The difference is more than OBVIOUS!

edit: files uploaded http://www.hydrogenaudio.org/forums/index....showtopic=14732

The samples are misaligned by ~1049 samples. I hope this didn't influence your test.
Go to the top of the page
+Quote Post
Pio2001
post Oct 31 2003, 20:08
Post #32


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



Thank you very much for sharing your results, Lucpes.

@Continuum, no, the difference is clearly audible, and measurable :



+2 dB above 10 kHz !
I've superimposed the two spectrums, and fit the levels for medium frequencies. In color, the differences between the cables, red=CAT5 louder, blue=Lamp cord louder.

What the hell is the gauge if this Cat5 cable ?
I'd like to know how the resistors were used, do you mean one time 30 kOhm and one time 10 kOhm in serial, so as to get 40 kOhm ?


Edit : picture link

This post has been edited by Pio2001: Nov 17 2008, 21:37
Go to the top of the page
+Quote Post
lucpes
post Oct 31 2003, 20:42
Post #33





Group: Members
Posts: 517
Joined: 9-October 01
Member No.: 254



The offset did not influence the test:

ABC HR config file
---------------
TestName = Cables Listening Test
Original = 30sec_big_cable.wav
Sample1 = 30sec_2x2_cat5.wav
Sample2 = 30sec_big_cable.wav

Offset1 = 0
Offset2 = 1049

test results
----------------
ABC/HR Version 0.9b, 30 August 2002
Testname: Cables Listening Test

1L = 30sec_2x2_cat5.wav
2R = 30sec_big_cable.wav

---------------------------------------
General Comments:

---------------------------------------
ABX Results:
Original vs 30sec_2x2_cat5.wav
8 out of 8, pval = 0.004

Speaker-to-line in converter:
http://www.hut.fi/Misc/Electronics/circuit...er_to_line.html

Sorry, I used 20kohm & 1kohm resistors (was not sure as I built that some 3 months ago to RMAA the modifications I did to the amp - no measurable anomalies (would need better soundcard) except for added crosstalk around -90dB) when using a 5W 100 ohm resistor as load (no speakers).

CAT5 (standard UTP Network cable): used 4 lenghts (with the outer jacket) of 4-twisted pair per speaker: so would be an equivalent to 24 gauge*16wires for one polarity... dunno how much that gets to...

However, there is no point in getting worried about cables if you don't have such a pathological case as mine: 4ohm nominal impedance, 86dB/W/m sensitivity & serious impedance swings so lots of current @constant voltage needed (no 2-way bookshelf speaker would exhibit such behaviour).

The infinity Kappa 9 I mentioned earlier is not my speaker, however it was the only impedance graph I could find for such a case. I guess my RSII's are somehow similar.

This post has been edited by lucpes: Oct 31 2003, 21:21
Go to the top of the page
+Quote Post
ff123
post Oct 31 2003, 21:11
Post #34


ABC/HR developer, ff123.net admin


Group: Developer (Donating)
Posts: 1396
Joined: 24-September 01
Member No.: 12



QUOTE (lucpes @ Oct 31 2003, 11:42 AM)
The offset did not influence the test:

ABC HR config file
---------------
TestName = Cables Listening Test
Original = 30sec_big_cable.wav
Sample1  = 30sec_2x2_cat5.wav
Sample2 = 30sec_big_cable.wav

Offset1    =  0
Offset2    =  1049

If the offset was 1049 samples, then the value you used is incorrect (offsets in abchr are defined in milliseconds). It should be more like:

1000 * 1049/44100 = 24

Also, this assumes that Sample2 is the one which starts later.

Probably the easiest thing to do would be to use Schnofler's abchr-java to automatically determine the offset.

ff123
Go to the top of the page
+Quote Post
Continuum
post Oct 31 2003, 21:33
Post #35





Group: Members
Posts: 473
Joined: 7-June 02
Member No.: 2244



I didn't think that you were hearing the offset. It was just something I noticed when I created the difference-wave. wink.gif
Go to the top of the page
+Quote Post
lucpes
post Oct 31 2003, 21:37
Post #36





Group: Members
Posts: 517
Joined: 9-October 01
Member No.: 254



QUOTE (Continuum @ Oct 31 2003, 08:33 PM)
I didn't think that you were hearing the offset. It was just something I noticed when I created the difference-wave. wink.gif


Well, I'm not that handy with Sound Forge & don't have any Java package installed...

Lamp cord similar to one used here: this page (bottom right one) 12 gauge: http://home.earthlink.net/~rogerr7/wire.htm#alllowcost

CAT5e more details: 4x8 twisted pair 24 awg per speaker. All white plus coloured ring wires are connected together for the (+) connection to the tweeter and all coloured wires as (-) connection.

This post has been edited by lucpes: Oct 31 2003, 21:44
Go to the top of the page
+Quote Post
Pio2001
post Oct 31 2003, 22:50
Post #37


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



So it is quite a big gauge after all.
24 AWG is 0.20 mm2
16x0.2=3.2 mm2, that is around 12 AWG.

This is surprising... Maybe recording the signal at the ampli plugs and comparing the 4 results (the two cables, speaker-side, and the two cables, ampli-side) could give more info. If the two ampli-side recordings are similar, the speaker-side recording that is the closest should be the best.

By the way, I find the idea of an attenuator for speaker - line connection strange. The author assumes that the output of an ampli is 3 to 20 Volts. This is true when the ampli is completely overdriven, and the speakers maybe broken.
Actually, I plugged directly the speaker output of my ampli into the line input for my recordings, and I had to set the volume so that I was worried about the neighbours, because it was late, so as to reach a proper line in level.
Your attenuator was 20:1, thus without it, you'd have got a signal 26 dB louder. You say that it was -20 dB, thus it would have clipped by 6 dB.

I recall again for everyone else : this experiment is DANGEROUS for your amplifier : a short circuit, a loose contact, a jack plugged off... and your ampli is fried !
Go to the top of the page
+Quote Post
_Shorty
post Nov 1 2003, 01:54
Post #38





Group: Banned
Posts: 694
Joined: 19-April 02
Member No.: 1820



unless it has protection for those situations, many amps have over-current protection so they don't get cooked if they're accidentally short-circuited. Would hate to find out the hard way though if I were unsure if it had that kind of protection.
Go to the top of the page
+Quote Post
saratoga
post Nov 1 2003, 04:49
Post #39





Group: Members
Posts: 4715
Joined: 2-September 02
Member No.: 3264



QUOTE
- Although it should be the case - I can tell you that in practice it is not always that way. Prime example was my last MB which was getting random data errors from the VIA chipset - because of what i guess was interference the digital data was getting corrupted.


Just an aside: If your board doesn't work, its broken. No amount of interfernce you're going to get outside of a labratory (let alone your home) is going to effect your motherboard.
Go to the top of the page
+Quote Post
KikeG
post Nov 1 2003, 19:31
Post #40


WinABX developer


Group: Developer
Posts: 1578
Joined: 1-October 01
Member No.: 137



High freq. attenuation of CAT5 cable is quite probably due to high capacitance of this cable, together with low impedance of load at high frequencies. This can cause a high-frequency rolloff. For this reason, CAT5 cable would not be good for tough loads such as this.
Go to the top of the page
+Quote Post
lucpes
post Nov 1 2003, 22:04
Post #41





Group: Members
Posts: 517
Joined: 9-October 01
Member No.: 254



QUOTE (Pio2001 @ Oct 31 2003, 09:50 PM)
So it is quite a big gauge after all.
24 AWG is 0.20 mm2
16x0.2=3.2 mm2, that is around 12 AWG.

This is surprising... Maybe recording the signal at the ampli plugs and comparing the 4 results (the two cables, speaker-side, and the two cables, ampli-side) could give more info. If the two ampli-side recordings are similar, the speaker-side recording that is the closest should be the best.

By the way, I find the idea of an attenuator for speaker - line connection strange. The author assumes that the output of an ampli is 3 to 20 Volts. This is true when the ampli is completely overdriven, and the speakers maybe broken.
Actually, I plugged directly the speaker output of my ampli into the line input for my recordings, and I had to set the volume so that I was worried about the neighbours, because it was late, so as to reach a proper line in level.
Your attenuator was 20:1, thus without it, you'd have got a signal 26 dB louder. You say that it was -20 dB, thus it would have clipped by 6 dB.

I recall again for everyone else : this experiment is DANGEROUS for your amplifier : a short circuit, a loose contact, a jack plugged off... and your ampli is fried !

As KikeG said, the attenuation might be due to the high capacitance/low impedance (which would act as an RC filter? ) - twisted pairs for all length - (+that would add a meter or two to the effective length of the cable...).

About the attenuator - my speakers have <<90/W/m efficiency and the Gemini might be in the 95dB range (http://www.bamberglab.com/s3.htm

See a SPL calculator here: http://www.doctorproaudio.com/doctor/calcu...uladores_en.htm

For 111dB SPL / 95dB/1W/1m speaker sensitivity a rough calculation would yield 25W. For 86 dB/1W/1m & the same perceived loudness one would need around 200W. Go figure that I want a beefier amplifier just after doing that smile.gif

So i'd need lots more amplifier power/voltage/amperage/whatever for same SPL (also taking in account the supposedly impedance swings which would need lots of current). The lowish sensitivity & impedance troubles is due to 2*10" woofers, 3*5" mid drivers and two tweeters to be driven through what is seems to be a major network of capacitors, transformer-type coils, shunt LC filters and long lengths of wire. However, the sound is pure bliss (okay, I'm going to the dark side here...)

Crossover here: http://oellerer.net/infinity_classics/RS_I...nical_sheet.pdf

I'll try to redo the test with amp side recordings next weekend. Ugh... I'd also use my smaller 3-way bookshelf speakers to provide data on 'speaker load/cables/amplifier matching'. Until then I'll keep kicking myself in the nuts for staying three months with CAT5 cable... as the difference while listening is humongous (at least that's what I'm hearing on the other side of the speaker).

edit: Added link to SPL calculator
edit: added 'low impedance needs lots of current' claim.

This post has been edited by lucpes: Nov 2 2003, 02:12
Go to the top of the page
+Quote Post
sshd
post Nov 2 2003, 00:43
Post #42





Group: Members
Posts: 207
Joined: 16-June 03
Member No.: 7218



What type of cat5 cable was used in this test?

I know of two types (probably noy described very good):

A: Solid core type -- used for installations.
B: Soft type -- used in network cables with plugs.
Go to the top of the page
+Quote Post
lucpes
post Nov 2 2003, 01:24
Post #43





Group: Members
Posts: 517
Joined: 9-October 01
Member No.: 254



El cheapo CAT5 solid core type used in the test.

Solid core is used for PC-PC or PC-switch direct connections.

Stranded or soft is used for patch-pannel connections.

I won't bother to use stranded, nor teflon insulated CAT5 (which some claim should sound better theoretically due to not having dielectric losses), no Belden-gaga whatever type to repeat the test.

edit: typos

This post has been edited by lucpes: Nov 2 2003, 02:24
Go to the top of the page
+Quote Post
sshd
post Nov 4 2003, 00:24
Post #44





Group: Members
Posts: 207
Joined: 16-June 03
Member No.: 7218



QUOTE (lucpes @ Nov 1 2003, 10:04 PM)
Until then I'll keep kicking myself in the nuts for staying three months with CAT5 cable... as the difference while listening is humongous (at least that's what I'm hearing on the other side of the speaker).

It seems you have decided that big lamp cord sounds better that cat5 cable. Has this been established as a fact?

Some questions:

1. Pio2001 made an interesting graph comparing the signals to eachother. The signals are different, but which one is more transparrent? I believe the samples should be compared to the original instead of eachother.

2. The recording is made behind the speaker. Is this relevant to how the speaker sounds?

3. Does the signal cable from speaker terminal to line-in affect the sound?

4. A lossy codec + DSP operations was used to provide the source. I should prefer a lossless codec without any replaygain/DSP operations done as the source for such an experiment (just to be sure we are testing the right thing).
Go to the top of the page
+Quote Post
Pio2001
post Nov 4 2003, 00:37
Post #45


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



QUOTE (sshd @ Nov 4 2003, 12:24 AM)
3. Does the signal cable from speaker terminal to line-in affect the sound?

Not under the assumption that only resistance (Gauge) matters, because the line in impedance is huge (47,000 Ohm). However, it can add very much capacity.
Go to the top of the page
+Quote Post
lucpes
post Nov 17 2003, 15:05
Post #46





Group: Members
Posts: 517
Joined: 9-October 01
Member No.: 254



QUOTE (sshd @ Nov 3 2003, 11:24 PM)
It seems you have decided that big lamp cord sounds better that cat5 cable. Has this been established as a fact?

Some questions:

1. Pio2001 made an interesting graph comparing the signals to eachother. The signals are different, but which one is more transparrent? I believe the samples should be compared to the original instead of eachother.

2. The recording is made behind the speaker. Is this relevant to how the speaker sounds?

3. Does the signal cable from speaker terminal to line-in affect the sound?

4. A lossy codec + DSP operations was used to provide the source.  I should prefer a lossless codec without any replaygain/DSP operations done as the source for such an experiment (just to be sure we are testing the right thing).

No, CAT5 sounds worse in my particular case. With speakers which do not draw lots of current (eg. bookshelf or small 3-way) it may even sound better or there would be no measurable or audible difference.

1. The CAT5 sounds muffled & no dynamics. The lamp cord brought the sound close to my Sennheiser HD600.

2. Yes, it's what actually gets into the speaker as a signal.

3. I don't think so (10kOhms+ line in) versus 4 Ohms for speaker.

4. It's enough to test with a lossy codec. It would make a bigger difference with a full-bandwith (uncompressed lossless) encoding.

I won't do another test, as my amp got fried somewhere in the middle of the process... 3/6 output transistors gone (literally blew up), one driver transistor gone, two diode rectifiers, one trannie from the feedback circuit, two resistors that controlled the bias current and now I'm to emitter resistors which went gaga. Hope I pull it through somehow... sad.gif

DO NOT CHANGE CABLES/DO ANY OPERATIONS ON YOUR AMP IF IT IS WORKING & TRIPPLE CHECK EVERYTHING BEFORE STARTING IT.

edit: it works 'ok' now but the side that blew gets too hot... back to the hacking table.

This post has been edited by lucpes: Nov 17 2003, 15:15
Go to the top of the page
+Quote Post
Pio2001
post Nov 17 2003, 23:36
Post #47


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



Sorry to hear about this.
Either there was a short circuit somewhere, which is likely with thin coaxial cables, especially with one end stripped out for the speaker connection, or with the plugging-unplugging (center of the cinch accidentally in contact with any grounded metallic part), or even the line input muting (can be internally muted by short-circuiting it)... Either the capacity of the coaxial cable added to the speaker cable drove the output stage of the "audiophile-kind-sound-is-more-important-that-security" amplifier into positive feedback.

If all is repaired and one side is abnormally hot, there might still be partially burned transistors : try to play some pure sound at a loud level (but don't blow your speakers, it would be very annoying ! ), and watch for the hot channel to sound clipped.
Go to the top of the page
+Quote Post
lucpes
post Nov 18 2003, 13:59
Post #48





Group: Members
Posts: 517
Joined: 9-October 01
Member No.: 254



Phew, I fixed it... turned out that the potentiometer that controlled the bias current was a little oxidized too due to high current that passed through when a diode blew, cleaned it, adjusted quiescent current as per the other channel setting & everything is OK now.

Cost: 15 euros + 8 hours of 'debugging'

Now to use banana plugs on the cables instead of bare coax wires for line input purposes... but some other time...

Horror pic here (23cm diameter for the transformer can in the middle; you can see on the right board two white resistors that replaced the old ones which caught fire)

This post has been edited by lucpes: Nov 18 2003, 20:49
Go to the top of the page
+Quote Post

2 Pages V  < 1 2
Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 18th April 2014 - 15:42