Meridian Unplugged - Powered by MrTechGuy.co.uk Forums Discussion HiFi & HT Chat Can sound quality be measured?

Balancing the discussion on 'those in favour of experimenting'; on the contrary, there is specific product for extracting further information and additional performance from an all digital system and I'm not referring to digital cables.

Some may want to experiment, some may not. Nothing wrong either way. But if you do start noticing or suspecting that information is missing in the soundstage, there are options to un-locking more. And believe me there is actually more!

Agreed (and also my experience), room treatment is very important and each listening room has different characteristics. Some characteristics can be pretty debilitating to the sound and enjoyment. But what you might have labelled solely 'room issues' may be a compounding of issues where source/ equipment can be treated too.

I suppose it entirely depends on whether you are happy with the sound that you have or not.

Ever wondered just how good Meridian kit can sound...?

This is my experience for what it's worth; all in the name of good healthy discussion.

Russ, bizarre but possibly true.

However, to maintain domestic harmony I spent >£1000 on funky wall panels instead.

Happy days.

In the interest of avoiding leading people astray and creating confusion, I think it's important to set some context around the following statement:
The matter of what one hears as a result of changing cables is certainly a matter of debate and always will be, as we all respond differently to different things in music.

But the idea that an electromagnetic signal can travel down a conductor and be absolutely unchanged is profoundly wrong. Even in the most basic terms of signal transmission, be it analogue or digital, several factors are at play that have a significant impact on the signal. If we keep it simple and analogue, we're looking at inevitable attenuation and group delay (different parts of the frequency spectrum moving at different speed down the cable), among other things. The impact of analogue cables on sound is well-established and many of us have experienced this.

Digital signals are more important and relevant in this debate. I'm sure some people are aware of this, but I'd guess that many aren't: Digital signals, when transmitted, travel in the analogue domain as electromagnetic waves, meaning that the signal contains both electrical and magnetic energy. Therefore, the energy in these signals is therefore subject to all the external and internal electrical, magnetic and mechanical influences that affect analogue cables and equipment. Of course, it’s a matter of extent, and a signal that is modulated digitally is fundamentally more tolerant of noise than an analogue signal, before degradation becomes noticeable

If we take a an Ethernet signal and simplify it, you can think of a square wave running at a steady base or 'fundamental' frequency (not quite true in practice, but all the principles hold true). This signal, when transmitted, is carried as a series of analogue sine waves, known as a Fourier series. In Ethernet, the fundamental frequency is 31.25MHz. The Fourier series (effectively the spectral makeup of the overall signal) will consist of a series of analogue sinusoidal signals, with the firs sine wave running at the fundamental frequency of 31.5MHz and at the same amplitude as the base square wave. Then we have a series of further sine waves at odd harmonics of 31.25MHZ (i.e. 3x31.25MHz, 5x31.25MHz, 7x, 9x and so on). The amplitudes of the 3rd harmonic (93.75MHz) will be a third of the fundamental amplitude, the 5th harmonic will be at a fifth of the amplitude and so on.

Here's a good link that gives a really nice 'plain English' explanation of what all this looks like. The video also shows how all the component sine waves add up (in a theoretically ideal world) to re-form a something approximating a square wave at the receiving end of the cable.

So, we're now in an physical reality in which our digital signal is actually made up of a long series of sine waves, at increasing frequencies and decreasing amplitudes. Here's a useful reference point that shows clearly how these basic principles of digital transmission are being applied by manufacturers of bog-standard CAT6 Ethernet cables: If you want to transmit faithfully 95% of the total Ethernet signal energy, you need a cable which can do a pretty decent job of carrying the base frequency, right out to the 19th harmonic. If model a Fourier series, you'll see that the frequency of the 19th harmonic is just under 600MHz. You may have noticed that CAT6 standards require a signal bandwidth of 600MHz - now you know why. If you look at the text printed on the sheath of a CAT6 cable, you'll probably see a reference to 600MHz.

What's the point of all this? Well, if we were to take a very fast oscilloscope and attach the terminals to the negative and positive-going wires from one of the four twisted pairs in an Ethernet cable (let's say it's 10 or 15 meters long, as is often the case in a streaming application), we'd see a surprisingly messy signal. The reason for this is that all the component sine waves have been exposed to the ravages of cable transmission characteristics. At these frequencies, attenuation is a major problem (due to skin effect) and group delay is a big factor. All this leads to a very messy signal at the receiving end.

The good news is that, in an Ethernet signal, the combination of the receiving devices and TCPIP do an incredibly good job of recovering the original square wave.

Provided specified distances for a given cable standard aren't broken, the receiving network devices and TCPIP can handle a situation where only a fraction of the original signal energy contributes to re-constituting short sections of the 'flat tops' of the original square wave.

Here's the bad news. Let's say that 25% of the original signal energy is 'seen' by the receiving network device as the wanted flat tops (and bottoms). What happens to the other 80% of the original signal energy ...?

This is what buggers our systems. All the distorted harmonics are still in there and they're all firmly in the band of nasty RFI frequencies. For all it's cleverness, is utterly transparent to RFI.

So we have a situation in which Ethernet cables contribute positively to RFI noise in a hi-fi system - you could call it self-noise. As for why higher quality Ethernet can sound better in the real world, it's down to better design and materials that minimise attenuation, group delay and several other unwanted factors, meaning that there is less overall diction and less self-noise.

Sorry for the long post. I've taken some time and effort to write this and I've done it because I hate seeing the wrong message going out and putting people off. Without a shadow of a doubt, cables can and do make a difference. They may not have a 'sound' as such, but they most certainly can influence sound. To argue that this is wholly impossible in an analogue scenario would literally require Ohms law to disappear.

The idea that cables can't make a difference in the digital world is much more understandable, but the basic physics of transmission described above mean that digital cables really can make a difference, and an unexpectedly big one at that because of the frequency range that Ethernet operates at (and, unavoidably, all higher frequency harmonics that go with it).

By the way, if you're in doubt about the validity of what I've said, there's a plethora of information on the web about digital signal transmission, both generally and in the context of Ethernet. This is all based on long-established signal transmission principles.

Rolski, I mean no offence at all and all credit to your friends in the space and automotive industries: what they say about deign and fitness for purpose is absolutely true. But remember, in the audio world, the end outcome isn't purely about moving packets of data from A to B - it's about music and music is affected heavily by things like RFI that don't matter (at least not as much) to cars and space ships ... :-)

Since RFI into control systems can lead to serious injury / death(s), I'd say it was more carefully looked at / controlled than in an audio system.

Is this another attempt to call Ethernet data analog (i.e. The LAN isolator thread)?

Adding RFI to Ethernet discussions seems odd, since the data itself is RFI, other than basic precautions (cable shielding, appropriately sized ferrites to prevent pickup from, or interference to other devices), the only people that can deal with this correctly are the product designers, since you can't isolate the data - which should be swamping the noise, and doesn't just disappear at the receiver which still needs internal shielding/segregation. (transmitting via optical is not complete isolation, since this is being converted back to electrical within the end devices, giving you back the unwanted RFI).

The basic design of the transmission system is that you are literally spraying crap into your endpoint, changing the materials of the hose, isn't going to audibly influence much, the best you can hope for is to make sure you don't have a leaky hose.

Discussions about bits not being bits tend to be by people that don't seem to be running systems with ECC memory / checksummed storage, and since a single bit out of place in the bits of the computer running at nearly 100,000 times the speed of the audio output can lead to crashes / mass data corruption it seems odd at why the hard bit is assumed to be working fine, but the 'easy' bit needs desperately to be fixed.

A single bit out on a firmware/software update can cause major problems - however, we are happy to send these files around (or across, depending on belief) the earth.

Lots of technical information, that is potentially technically correct isn't necessarily applicable, the analog behaviour of Ethernet is a fixed issue, although you can deliberately degrade the signal (I.e. Repeatedly adding additional isolators to the ones already present as part of the fundamental design - You have have a minimum of two - adding external ones doesn't make the analog properties of the signal any better, trying to argue it's analog and then deliberately degrading it seems odd for anyone not trying to sell both isolators and Magic cables)

I see references to experimentation - unfortunately (assuming the listener is human) to actually test the audibility of a change is extremely difficult unless you can get rid of the subconscious bits of the mind and any other external factors (like getting your ears back to within a few millimetres of the last test).

Non-scientific experimentation can also lead to serious problems (I.e. People playing electrician and removing / uprating safety fuses or playing with mains polarities defeating the internal fuse protection, leading to potentially fatal results) I also seem to remember reading here that LAN cable shielding was good at hiding interference created by loose connections behind ring-main sockets - personally I'd prefer the 'your house is in danger of burning down' audible alert so I could fix the problem, not the symptom.

Whenever I hear someone tell me to trust my ears, I assume I am being accused of either not being human, or not having a brain. (Or someone is trying to use the known biases/deficiencies of the human auditory system to try to sell me something)

I like the human auditory system - for zero conscious processing, you get position, speed, direction and a ton of other information dumped into your brain, my ability to enjoy music comes from being able to relax and not have to worry about manually processing the sound.

When I really want to listen and enjoy something, I have a quiet room I can completely black out - so the only thing is the sound - however, not being able to see what they are listening to seems somehow to stop some peoples enjoyment (although this isn't blind testing, since I know what I am listening to)

Long ramble over, always get annoyed when people seem to claim that computer data (travelling at gigahertz and not tolerant to bit errors - the data, not the protocol) is somehow different to computer data (audio) that is being received by something a lot less picky about timing and precision.

Matt.

Whilst the cable will affect a number of variables, a good 75ohm cable will transmit the signal with little to no errors, as the data is digital it cannot be altered by a cable without introducing errors. The signal is also transmitted in the MHz domain well outside the audible range.

I read a definitive test recently concerning HDMI cables where the test involved a section of video where each frame had an md5 hash generated so that the captured frames could be compared with the original to determine whether or not expensive cable made grass any more vibrant or flesh tones more natural, the result bar one or two frames that had single pixel errors showed that a £1 cable was as good a transmitter of data as a £100 cable.

T

A few of the points in this thread have been about room acoustics. I'm in strong agreement that they overwhelm differences in equipment in terms of listening pleasure (at least for me).

One listening/measurement approach that I've found useful is LEDR

Give it a shot. While it does require an ear/brain combo for the measurement (as opposed to lab measurement equipment), it gives pretty unambiguous results when positioning speakers and acoustic treatment.

Hi Big T,

You are correct in that a good Ethernet cable will transmit the Ethernet signal without data errors. In fact, with the help of TCPIP, it will do that all day, every day, year after year, as long as you don't try to use too long a length of cable. But even in the absence of bit errors, the RFI carried by the distorted analogue sine waves I mentioned will pass through the end of the cable and into the rest of the system unhindered.

This is the key point: all can appear entirely well from a data integrity point of view, as error correction will mask the inevitable distortion of the analogue sine waves that I descried above. This distortion is an unavoidable fact, driven by how an electromagnetic signal interacts with a conductor.

So, the cable will run free of data errors, but the underlying RFI is still there in spades and will inflict significant damage to the sound.

I can't comment on HDMI cables as I have no experience with them. I'd hazard a guess that TV pictures are less sensitive to RFI.

Thanks for posting. The LEDR test works fine on my desktop PC. It's amazing to hear the sounds floating over and behind while sitting at my desktop. I'll try later to download the files, create a CD, rip it on dBpoweramp and see how my big M system does. Seems like a lot of work but might be worthwhile.

This is factually incorrect. Whilst TCP/IP does allow for error correction;

1. TCP/IP is used only used for some communication, notibly it is NOT used for Sooloos audio data communication except for very coarse integrity checking.

2. IP networking has error detection built in for other types of communication and this statistic is available at NIC level.

3. Lengths of cable are part of specification so gigabit over CAT5e/6 is good for 100m ish.

If your NIC on PC or Switch supports it, check out its error count. If it's anything other than 0 errors, something is broken with the network. Providing all parts of the network are built using in spec components, and the network is designed to be in spec, the network will be verifiably error free, TCP/IP error correction or not.

Hi Ian,

I agree with all three of your bullets - points well made.

To step back a bit, the very point of digitally encoding any kind of signal for transmission is to make the base signal more robust in the presence of noise and distortion that always affect analogue electromagnetic signals to some degree. I think we all agree on that

TCPIP, where it applies, adds further robustness to this picture by preventing transmission of data packets containing errors.

Back with the Ethernet signal, there is some inevitable distortion of the component analogue waveforms, certainly including some degree of attenuation and some temporal distortion. That these things will happen to some degree is a certainty, based on long established principles of signal transmission and properties of conductors. Making this un-true would require some basic laws of physics to be broken.

The cleverness in Ethernet design is that it creates a remarkably robust environment that allows data to be passed reliably over distances of upto 100m. The day job of digital encoding and TCPIP (where applicable) in this context is to allow continuous accurate data transmission in the presence of some noise and distortion on the base waveform. This ties back to the basic objective behind digitally encoding a signal.

In Ethernet, whatever percentage of the underlying analogue signal is distorted, while it does no harm to the data stream, contributes additively to RFI noise inside the cable and is passed on relatively transparently down the food chain.

It's worth having a look online for oscilloscope images of Ethernet signals at the downstream end of a cable - you'd be surprised by how much distortion of the original square wave there can be without causing big errors. Just make sure you're looking at images showing a signal after transmission, rather than a fresh signal on the output of a switch, where there would be no meaningful distortion.

It's also quite helpful to look for screen grabs from spectrum analysers, where you can see what I described earlier re the base signal being made up of energy at increasing frequencies and decreasing amplitudes.

Regards,
Dave

Hi Akimo,

Very interesting link. Is this something you have been able to apply in planning your room treatment? I'm lucky in having a fairly well behaved room, but I know it has issues and that I need to plan some room treatment.

Dave

I think LEDR came out of the research that the Canadian Government funded back in the 80's.

For me, it helped identify the ceiling of my room as a problem spot. I played around with acoustic treatment on the walls until things sounded right. Diffusion won out over absorption.