Power cables - how to measure

[Guest rmpfyf]

IDK 'bout y'all though some good did come out of the recent power cable thread now locked; @Benjet suggested a paper (http://rightnote.co.uk/wp-content/uploads/2015/09/ka-paper.pdf) which whilst flawed does suggest the start of a better experiment enough that it had me rifling through the attic looking for lab equipment (only to have my wife remember she'd given my data acquisition stuff to a friend - which led to a lot of 'what do you mean you gave it away' type discussion... there was some nice kit in there too...)

So @Benjet was being constructive and me not so much (apols mate, keen on your feedback here), the thread got locked before I could add this and add something constructive, here it is. 

 

If one was able to get a hold of a decent data acquisition board (I'm thinking something with higher resolution - time and scale - than the source material, It wouldn't be too hard to script a test that checked for differences between source material and output waveform on a given bit of equipment. What would you want the test to look like?

 

What one might want

I'm thinking I'd want a mains power analyser running simultaneously enough to characterise the incoming waveform - run double-single phase - one on the incoming mains, one on the audio circuit

A few different passages of music.

The test run repeatedly throughout a day (30+ times)

 

What one might evaluate - assuming different cables are evaluated

Peak to floor differences

Comparison of gradients (easy enough on a x-y scatterplot)

Floor measurements

Mains harmonics, relative sensitivities

 

Process

Take a sample of audio, insert an impulse at the start for time alignment

Playback and record

Process data - time align then post

Maybe leave the cable in there running a day first?

 

Keen on thoughts for what people would think is of interest. Tagging a few others that might care @rawl99 @zenelectro @davewantsmoore @kukynas @buddyev @Assisi @Ittaku @Stereophilus

 

 

Sorry anyone from that last thread of interest I've left out. Anyone welcome to reply, chip in, add thoughts. 

 

Very nearly bought a mains power analyser quite inexpensively enough to have a project with my sparkie last week before I was found out (serves me right for walking away from the PC with the eBay page open), and who knows, the mate with my DAQ hardware might be inclined/coerced into sending it back...

 

Just for clarity - this isn't an opportunity for people to talk about how much cables don't do anything for them or in general, or how their kit is engineered to survive all manner of power quality issues. Just on how you'd potentially evaluate.

[bob_m_54]

I agree that @Benjet did constructively add to the thread, by illustrating that some manufacturers do show some sort of comparison, between their cables and standard cables. Which was a point brought up with regards to null tests. I did eventually get to post a reply to him on my thoughts for one of the links he posted, but wasn't able to show it on the thread unfortunately. I won't post it here and clutter up your thread though. I was also guilty of  a couple of less than constructive comments too. Apologies from me as well.

 

But yes, some hard data would be good. Even using a high res wav or flac file, through a very stable DAC, into a very stable Amplifier (which would be supplied via the power lead under test), and exploring the output across a dummy load, with a decent storage CRO.

[MLXXX]

The above approach would allow an assessment of the sensitivity of the output waveform of particular audiophile equipment to daily mains supply variations.   And it could also be used to show the sensitivity of the same audiophile equipment to daily mains supply variations when a different power cord is connected.

 

If anyone were seriously interested in investigating audiophile equipment performance variation under varying conditions, this sort of approach would be laudable.

 

Other matters that come to my mind would include how hard a power amplifier would be driven; what dummy load might be used for a power amplifier; and whether the temperature of the test room would be kept constant.

 

Given your personal/professional interest in mains supply quality @rmpfyf, I can imagine you might indeed find this type of exercise of interest.  I suspect the members of this forum who are at the subjective end of the spectrum might not be all that interested, preferring to rely very much on their auditory perceptions rather than seeking corroboration through measurements.

 

I myself would be mildly interested in the extent of variation found with mains voltage variations (which can easily be +-10V RMS at a particular power point in a dwelling over a 24 hour period). I suspect that if a modern power amplifier were not driven really hard then its power supply regulation ought to be good enough to maintain its performance at a high standard despite daily mains variations in voltage (and for that matter variations in the harmonic or other spurious content in the mains waveform), but that is to a certain extent only supposition on my part.

[pwstereo]

@rmpfyf what you are proposing sounds great.

I'd be very interested in seeing this sort of testing done and the resulting data.

[bob_m_54]

10 minutes ago, MLXXX said:

snip--------------------------------------------------------------------------------------

Other matters that come to my mind would include how hard a power amplifier would be driven; what dummy load might be used for a power amplifier; and whether the temperature of the test room would be kept constant.

snip--------------------------------------------------------------------------------------

 

I hope you don't mind me truncating your post..

 

Dummy load would need to be a non inductive load, and could be just a resistor bank of carbon resistors, to give a total of 8ohms with the ability to handle maybe 50W or 100W, without getting too hot.

I did have one I made to handle 50W continuous, but don't know where it is at the moment. I may have given it away or sold it when I got rid of my other gear.

 

Amp running where it is most stable, so around the lower third of amplification, depending on the amp. And yes, ambient temperature stability would be good, but it depends how far you want to go.. Even a well insulated room without any strong air currents would be OK, although not as good as a lab obviously.

[Guest rmpfyf]

I'm curious not least because the one 'audiophile' power cable I tried made things sound (subjectively) worse and for the life of me I couldn't put a finger on why. I thought it was just me letting my biases get in on it whem my wife walked in (not aware I'd done anything), made a face and said 'Have you done something to it again? It's worse.' (she's got an interesting choice of words there however you get the sentiment). 

 

@bob_m_54 one of my guys has a very serious Yokogawa power scope that'd do this job nicely however he's more likely to connect me to mains, Taken-style, than to part with it let alone let me anywhere near it. I had some nice National Instruments stuff, 16 and 18-bit, that I need to reacquire. Or sneak one through the gate come payday. The 18-bit stuff would be ideal but is freakin' hard to get a hold of cheaply (her refrain of 'you weren't using it and it was taking up room' is hurting a bit). 

 

I was thinking of tapping my DAC's output. This changes things somewhat I'd suggest though (a) it makes for a shorter component chain to characterise and (b) I can go direct in without anything to get the signal scale to suit (downstream of an amp is going to involve some signal conditioning). The only DAC I currently have with it's own power cord is 16-bit (good) and has tubes (keen on your feelings, team). This would be testing a power cable on a DAC (which is where I had it when I tried). I'm keen on thoughts as to what's more important - DAC or amp?

 

Characterising mains is important. Keeping an eye out for suitable Acuvim/Schneider ION/etc meters (I don't know why audiophiles don't buy these before dropping $3k+ on power conditioning, but that's another discussion). 

 

I assume from comments @MLXXX and @bob_m_54 some sort of characterisation of how the amp's doing would be ideal too... environmental temp at least should be easy enough to acquire.

 

I like the idea of a dummy load... other people live here!

[Guest rmpfyf]

Wondering what could be done to measure contact/interface effectiveness also. Or some characterisation of the R/I/C/etc of any cable assy. Worthwhile?

[Stereophilus]

My thoughts, as i put in the closed thread, are that power cables have wider ranging system effects beyond the component they are attached to.  I think the only way to asses systems is with "in-room" responses.  This would neccessitate microphones, a room that is not overly damped, and some method for comparing complex music across a broad in-band and out-of-band frequency range.

[bob_m_54]

1 minute ago, Stereophilus said:

My thoughts, as i put in the closed thread, are that power cables have wider ranging system effects beyond the component they are attached to.  I think the only way to asses systems is with "in-room" responses.  This would neccessitate microphones, a room that is not overly damped, and some method for comparing complex music across a broad in-band and out-of-band frequency range.

My thoughts would be that, that would probably come under further investigation. If above tests showed a definite change in signal output, further testing would be redundant (but interesting).

 

If the above test showed no change in signal output, then further investigation should be explored, but would be much harder to set up reliably. And more very reliable equipment would be required.

[Stereophilus]

6 minutes ago, bob_m_54 said:

My thoughts would be that, that would probably come under further investigation. If above tests showed a definite change in signal output, further testing would be redundant (but interesting).

 

If the above test showed no change in signal output, then further investigation should be explored, but would be much harder to set up reliably. And more very reliable equipment would be required.

Thats fine.  As long as it is acknowledged.  I agree it is going to be harder to reliably measure.

[bob_m_54]

15 minutes ago, Stereophilus said:

Thats fine.  As long as it is acknowledged.  I agree it is going to be harder to reliably measure.

I actually contradicted my own thoughts, when I mentioned exploring it further. Any further testing would have to be in the area of physiology and mental  processes (what the brain is doing) in the listener, and is way beyond the scope of this experiment.

 

What we are talking about here is measuring whether the cable actually does anything or affects the audio in any way. So it is purely physics based.

 

Sorry @rmpfyf for muddying the waters.

Edited October 5, 2020 by bob_m_54

[davewantsmoore]

2 hours ago, rmpfyf said:

Just on how you'd potentially evaluate.

The first thing to do is to assess the "repeatability" of your test setup.   ie. using the same system/cables, run the test 10 times over, and do some in-depth evaluation of the output signal errors.  This result is critical to the evaluation of anything else. (eg. different cables).

 

If you do not see identical results from test to test .... then this has to be at minimum factored into your method, analysis, etc..... and "ideally" (nothing ever is) actually traced back to a cause and understood.

 

 

... but other than that.    It isn't hard (just tedious).   Capture the output and check for differences.

[Ittaku]

System measurements vary wildly even between tests with sensitive equipment. Running two sweeps on my system with my test equipment in my room yields a slightly different result every single time. If you wish to do system tests, you need to do a lot of sample runs and characterise what the per-measurement variation is in the first place before you can determine what would qualify as a significant difference beyond that error when changing something in the system.

Edited October 5, 2020 by Ittaku

[davewantsmoore]

Quote

The above approach would allow an assessment of the sensitivity of the output waveform of particular audiophile equipment to daily mains supply variations.

In order to do this on any sort of accurate scale you will need to synchronise the data of the mains power analyser, with the output of the DUT.

 

On a longer time scale (hours, days, etc.) then it will very much be reading tea leaves.

2 hours ago, rmpfyf said:

Amplifier

 "dummy load" for an amplifier is unlikely to reproduce real world power draw.

 

I'd stick to the output of a DAC or preamp to begin with.

[davewantsmoore]

1 hour ago, rmpfyf said:

Wondering what could be done to measure contact/interface effectiveness also. Or some characterisation of the R/I/C/etc of any cable assy. Worthwhile?

You're getting ahead of yourself.

 

Measure  some differences in the output signal first, and then work to be sure you have controlled sufficiently for where they are coming from  (eg. is it really the mains variation.... or is it really the diffrnt power cables.... or)

[davewantsmoore]

1 hour ago, Stereophilus said:

My thoughts, as i put in the closed thread, are that power cables have wider ranging system effects beyond the component they are attached to.  I think the only way to asses systems is with "in-room" responses.  This would neccessitate microphones, a room that is not overly damped, and some method for comparing complex music across a broad in-band and out-of-band frequency range.

Yes... but this is a complex and difficult to control test.

Starting first with the output of a DAC or preamplifier is much simpler...... and the hypothesis of "it's affecting the poweramp and/or speakers" ... can be left unresolved for the interim.

[bob_m_54]

9 minutes ago, davewantsmoore said:

In order to do this on any sort of accurate scale you will need to synchronise the data of the mains power analyser, with the output of the DUT.

 

On a longer time scale (hours, days, etc.) then it will very much be reading tea leaves.

 "dummy load" for an amplifier is unlikely to reproduce real world power draw.

 

I'd stick to the output of a DAC or preamp to begin with.

Dummy load connected to the amplifier, rather than speakers. That would still give you realistic loads on the amplifier. But you could still use speakers, depends how well your test room is insulated, or how forgiving your neighbours are LOL.

 

Edit: and it's the affect on the signal at the output of the audio chain, that needs to be measured, not any effects of the room on the audio output..

Edited October 5, 2020 by bob_m_54

[Guest rmpfyf]

1 hour ago, Stereophilus said:

My thoughts, as i put in the closed thread, are that power cables have wider ranging system effects beyond the component they are attached to.  I think the only way to asses systems is with "in-room" responses.  This would neccessitate microphones, a room that is not overly damped, and some method for comparing complex music across a broad in-band and out-of-band frequency range.

 

Wouldn't be too hard (just annoying for anyone in the house) to put a mic in place and simultaneously measure the output (most DAQ cards are 8-channel, there's a bit to do in unpicking channel multiplexing but it's all feasible). Getting a suitable noise floor would be an interesting challenge (my current spare machine is not 'quiet'). Got a calibrated condensor collecting dust somewhere.

 

I'd prefer to keep it upstream as to not need unpick and characterise further variables, though I get what you're on about.

 

22 minutes ago, davewantsmoore said:

The first thing to do is to assess the "repeatability" of your test setup.   ie. using the same system/cables, run the test 10 times over, and do some in-depth evaluation of the output signal errors.  This result is critical to the evaluation of anything else. (eg. different cables).

 

I'd script it to go off every half hour (or more) for a day - would need more than 10x for decent averaging. Probably more. As per @Ittaku there's a *lot* going on at mains level. 

 

I'm feeling DAC/pre output is going to be easier and likely more scientifically insightful.

 

6 minutes ago, davewantsmoore said:

You're getting ahead of yourself.

 

Dunno. If I were writing a report I'd want some basic characterisation of what's under evaluation.

 

1 minute ago, bob_m_54 said:

or how forgiving your neighbours are LOL.

 

They've a newborn so we can rule a bit of that out.

[MLXXX]

5 hours ago, Stereophilus said:

My thoughts, as i put in the closed thread, are that power cables have wider ranging system effects beyond the component they are attached to.  I think the only way to asses systems is with "in-room" responses.  This would neccessitate microphones, a room that is not overly damped, and some method for comparing complex music across a broad in-band and out-of-band frequency range.

Unfortunately, introducing loudspeakers, room acoustics and microphones would very possibly mask the small changes in the waveform output by a power amplifier caused by changing the mains input voltage at the power point, or changing the power cord from the power point to the power amplifier.  

 

The voice coil of a speaker driver can get vey hot when driven even at moderate levels, significantly increasing its resistance, and thereby significantly reducing the current flow, with the result that the sound pressure level is momentarily reduced. There is no guarantee the voice coil will behave in the same way each time a piece of music is played, even if the one track on a CD were replayed continuously.  One reason is that the air flow created by speaker cone movement is to some extent  turbulent, thus introducing random variation.  Also, over a period of an hour or more, if a CD track were played continuously, the housing of the speaker would gradually rise in temperature; air trapped inside the speaker enclosure would gradually get hotter; and air in the room would tend to rise in temperature.

 

But that would not be the end of the matter. Convection air currents would be stimulated and these would to some extent be random.  Microphones would receive sound waves passing through layers of air at different temperatures.  On the other hand, air conditioning if used, and in particular if it involved high air volume continual circulation of air in the room, could help create a homogenous body of air.

 

Then there is the question of humidity. High frequencies are attenuated as they pass through moist air and the rate of attenuation is sensitive to temperature and humidity. Domestic air conditioners tend to reduce humidity if operating in cooling mode.  Air-conditioning for a museum or an art gallery might be designed to maintain a particular humidity level. 

 

Anyone entering the room would disturb the pattern of standing waves, and the reverberation characteristics of the room.  Street noise and noise from within the dwelling would be picked up by the microphone(s).

 

The microphone placement relative to the loudspeakers would be very critical and would need to be maintained to a high precision. Room dimensions become critical in a non-anechoic room and these dimensions will vary with temperature as the building structure contracts and expands affecting the way reflected sound reinforces or cancels the direct sound, especially at short audio wavelengths (high audio frequencies). 

 

Because of the above sorts of issues, if trying to measure very small changes in the output waveform of an audiophile device it is generally going to be more successful if the measurement is done directly from the output connector of the device than to attempt to include the potential masking effects of speakers, room acoustics, and microphones.

 

Of course if the differences were profound, audibly "night and day",  then the differences could indeed be heard in recordings with microphones in an ordinary listening room. And such recordings could be uploaded for interested parties to listen to for themselves. 

 

(Even dummy loads for power amplifiers will rise in temperature somewhat during use, and the resistance therefore rise somewhat, which will slightly reduce the current flowing through the dummy load, all other things being equal.  However, the ADC connected to the dummy load, typically via an attenuation pad, will be monitoring the voltage across the dummy load, not the current flowing through it; and modern power amplifiers usually have quite low output impedances.)

Edited October 5, 2020 by MLXXX

[Guest rmpfyf]

@MLXXX it'd of course be interesting to study the cumulative effect to see whether it is, as you suggest, night and day or whether, as a Prof of mine used to suggest, its 'down in the flys**t'.

 

I'd agree though that for a study of power cable effects it's best to run a limited variable domain... there is a lot to consider going beyond this.

[aussievintage]

28 minutes ago, MLXXX said:

Unfortunately, introducing loudspeakers, room acoustics and microphones would very possibly mask the small changes in the waveform output by a power amplifier caused by changing the mains input voltage at the power point, or changing the power cord from the power point to the power amplifier.  

 

Take a step back then, and consider what that really means.  As I said elsewhere, any real effect has to be small, very small, or else there would not be so much debate.  It also means there are better ways to spend your money.

[davewantsmoore]

3 hours ago, bob_m_54 said:

That would still give you realistic loads on the amplifier.

Even a dummy load with a complex impedance (tailored to look like a speaker impedance) doesn't really represent a speaker... a speakers impedance fluctates with cone position.

 

Eg:  8" woofer impedance (cone rest, in, out)

 

3 hours ago, bob_m_54 said:

But you could still use speakers, depends how well your test room is insulated, or how forgiving your neighbours are LOL.

... but then it becomes very very difficult to control.   The repeatability of microphone measurements of sound in a room is low when talking about a very high resolution.

 

It is best to begin with the electric signal.

 

If it cannot be shown (for example) that there is error introduced in the electrical signal (eg. in the DAC or the amp, or wherever) .... then a new hypothesis can be made which is that there wont' be any error at the speaker....  (and then you can worry about testing that)

 

3 hours ago, bob_m_54 said:

Edit: and it's the affect on the signal at the output of the audio chain, that needs to be measured, not any effects of the room on the audio output..

Sure.   You can capture the sound coming out of the speaker in a way which maximises it's resolution and repatability.... but it is much lower reslution than looking at electrical signals.

[Stereophilus]

21 minutes ago, rmpfyf said:

@MLXXX it'd of course be interesting to study the cumulative effect to see whether it is, as you suggest, night and day or whether, as a Prof of mine used to suggest, its 'down in the flys**t'.

 

I'd agree though that for a study of power cable effects it's best to run a limited variable domain... there is a lot to consider going beyond this.

I am in agreement that there is a lot going on at "system level".  It is part of the reason that in my view measuring the effects of power cables is going to be challenging.  

 

Limiting the measurements to output from power amplifiers will make measurement much easier.  I would just caution about drawing conclusions too broadly in doing so.

 

17 minutes ago, aussievintage said:

 

Take a step back then, and consider what that really means.  As I said elsewhere, any real effect has to be small, very small, or else there would not be so much debate.  It also means there are better ways to spend your money.

Can we please move on from the "value" argument? We are talking about cause and effect.  Even a small, statistically significant measured effect is useful in our further understanding.

Edited October 5, 2020 by Stereophilus

[davewantsmoore]

3 hours ago, rmpfyf said:

Dunno. If I were writing a report I'd want some basic characterisation of what's under evaluation.

It doesn't matter what is under evaluation if you can't reliably show a difference in signals.    I think you will have your work cut out for you there.

[davewantsmoore]

22 minutes ago, rmpfyf said:

I'd agree though that for a study of power cable effects it's best to run a limited variable domain.

No, it's essential.

 

After you characterise what differences (if any) the incoming mains power make to the signal you are capturing.....   You will know how much diffeernce it is going to make as to when you run the test.

 

What that is going to tell you is, how valid will a test be where you use power cord A.... and then some time later use power cord B....   If you're getting errors of "X magnitude" running the same cord, ever few minutes ..... then unless these errors can be controlled for, or unless they are much smaller than the differences you get when using two differnent power cables..... then you can't see beyond them.