And was it the same amp that tripped each time?
1 Like
Hi Simon (and I_B),
Thank for your continued interest and I did eventually (several weeks later) get to the bottom of the tripping. I agree - this is not an effective start to a six person masked ‘randomised’ trial!
To answer @Innocent_Bystander question - it was the RH amplifier on the first day and the LH amplifier on the second day. As you might imagine on the second tripping I was more anxious.
The correct approach would have been to abandon the trial at this point even though I already had confirmed dates with all the listeners. Unfortunately one of the cable candidates (German - Inakustik) was on loan from a HiFi retailer and as I had borrowed this cable pair once before I was keen not to mess the retailer about by returning this pair and then asking to borrow again later.
Accordingly (on the 2nd July) I began my investigation. As I did not have much time I did not immediately try to fully replicate the tripping - I just wanted a quick fix to be able to continue with the listening tests - the next session being due the following day.
First try was to see if the fault would repeat with the cables plugged in the wrong way round (i.e. opposite to the marked ‘direction’ on the cable). See image below.
Original Image: Author provided, no reuse.
Result - eureka - no problems.
Listening tests could continue - I simply had to remember to connect the Naim NAC A5 contrary to the marked direction on the cable! To be clear the NAC A5 results in the charts have the cables oriented in the marked direction for the first two days and contrary to the marked direction on the remaining four days.
One might imagine at this point I was quite excited… I could not tell the difference when listening (if the amplifier did not trip) whichever direction orientation the cable was installed, but it seemed the Linn amplifiers might have very strong opinions on the matter!
The other amusing aspect is that test tracks that Mrs E of E and I typically use did not trip the amplifiers. The offending tracks were (first day) Simon and Garfunkel: ‘Bridge over troubled water’ and (second day) Wagner: Tannhauser, ‘Pilgrims March’, Viener Philharmonic Orchestra, George Solti conducting (Decca label, remastered at 96khz 24bit HiRes from 1971 original recording).
As a aside, I have been following the ‘AI - where is it taking us?’ thread and I briefly did consider posting a (flippant) remark that AI has even been incorporated into Linn amplifier technology such that the amplifier can form its own opinion on the music it is asked to play. 
There is quite a bit more to post on serious technical matters and to explain how I eventually found the very subtle fault in one part of a NAC A5 cable that I had assembled. I shall continue to post in a short while - but feel free to add any comment if you wish.
1 Like
Very interesting! When time permits it would be good to play with the NACA5 further, reversing direction several times, to verify if fully repeatable and actually due to a genuine cable effect. Or possibly something incidental that that was caused to change when you first reversed (but no idea what!)
1 Like
Hi @Innocent_Bystander ,
Thanks for the comment and practical suggestions for more experiments.
It may not be a surprise to you that I already tried to replicate the amplifier faulting.
At the end of July, as I was prepping for the Group B tournament and finding that very long lengths of Cardas Clear Beyond where still giving difficulty (see previous posts here and here), I took time out to see what might be happening with the NAC A5 cables at 1.5m length.
Initially I had been reluctant to repeat a situation where an amplifier might enter fault state (as one may not know if damage is occurring, either to amplifier or possibly loudspeaker).
However, given that Linn Products Ltd had already had email communications from me in 2025 summarised into technical report(s) in May 2026 - I figured I might as well carry on.
IMO, the Magico M2 loudspeaker represents a very difficult electrical load - more reactive than some test loads used in some loudspeaker manufacturers laboratories. This loudspeaker can demand very high currents both at low frequences, but also at high frequencies (beyond the audio band).
With my quest to find the technically correct cable for this arrangement, I had been interested in seeing how short (c.f. low loss) a cable I could use in long term use. Via many experiments I had established that 1.5m long cables such as Naim NAC A5, other brands such as Tellurium Q, or DIY home made electrical or DIY copper pipe were OK. So it was a big surprise to have the fault occur on a 1.5m long cable in the listening evaluations. What had happened was as follows…
In my effort to keep the loudspeaker cables out of sight from the listeners I had hidden the cables behind and/or below acoustic treatment. After the first day listening with the fault I had done a quick check of the NAC A5 pair, unscrewing the insulator caps from each plug or spade connection - but could not find anything untoward.
Then I put the cables back hidden behind the acoustic treatment ready for the next day - but I must have switched the cables left to right and right to left - hence the fault trip on the other amplifier on the second day. Accordingly when I tried to repeat the faulting behaviour using the faulting test tracks, the fault followed just one NAC A5 cable.
The fault turned out to be a fracture in one strand of the 19 strands (at the banana plug - LS end) on the +Ve (red) cable just inside the insulation which had been cut back by me to mount the banana plug. In other words, the amplifier was shutting down because and this open circuit single strand would probably be encouraging oscillation in the output stage. When cable was fitted in the reverse direction there is much less of an issue.
I labelled the cable to warn me to remake the termination before using (see image below).
Original Image: Author provided - no reuse.
Learning point for me:
Buy professionally made and assembled cables or get cables made to specification by people who know what they are doing (i.e. professionals).
Maybe we are all in agreement with Simon-in-Suffolk? - the loudspeaker cable evaluations need to be undertaken again, from the beginning (argh)…!
1 Like
Problem solved then! But regarding getting done by professionals, I’m sure you are aware that whilst there are excellent professionals there are also others, and the difficulty is finding the former!
1 Like
Thanks @Innocent_Bystander ,
Actually I should credit you with much of the progress on my quest re cables etc - as you have stayed with the project throughout. But there is one particular highlight observation of yours that I wish to consider once more.
When I re-published the multifrequency measurement data for Naim NAC A5 at post 405 (link here), you had previously observed…
This observation, together with the other difficulties in running effective ‘masked randomised’ evaluations of loudspeaker cable types, pushed me to undertake the modelling in LTSpice.
Furthermore, your observation of the drop in inductance for NAC A5, together with my observations of the variability of the conductor core geometric stability within the insulation of NAC A5 has caused me to go hunting for the truth to answer the following conundrum.
Is it an (unintended) consequence of manufacture that the conductor cores of NAC A5 wander in the insulation and do not have consistent separation along a given length?
or
Was it a deliberate design decision (circa 1989) to give NAC A5 this characteristic?
Either way, the outcome is the same, the high frequency losses (beyond 20kHz, but within 3dB bandwidth of audio amplifiers) for the NAC A5 cable are higher than some other commercial cables (e.g. Chord Sarum-T) or other lower cost DIY approaches (e.g. my Copper Pipe). The excess losses in NAC A5 vs other types listed are a consequence of more loss due to skin effect which is present in NAC A5 (particularly at ‘pinching’ points when the two conductors are closest together).
Higher losses in this region (10kHz - 100kHz) will tend to curtail the amount of power getting to the tweeters (beneficial for not stimulating tweeter resonance) and also damp any HF return reflections (and even damp any potential HF oscillations should they occur due to clipping).
To consider the above, I have found two articles of potential relevance in the public domain.
One article online claims to be text extracted from an interview published in Hi-fi Review magazine in April 1989 with Malcom Stewart and Julian Vereker (Hi-fi Review magazine in April 1989). In that article, JV talks about the consequence of poor impedance mismatches (of interconnect cables) and the importance of getting that correct. In the same article, JV shares his thoughts and frustrations re NAC A4 and why NAC A5 came into existance.
The other article is from HiFi News & Record Review (August 2000) in which it is clear that Naim were using the B2 Spice circuit simulation package to design their products.
My uninformed opinion from all of the above is that the geometric wandering of the conductors in NAC A5 was/is deliberately introduced at manufacture using some kind of eccentric feeder as the conductor pairs are sheathed in insulation. The suggestion of ‘reliable and predictable directionality’ in NAC A5 is a ‘smokescreen’.
Nevertheless, with this cable construction there is risk of putting little ‘nicks’ into strands when preparing cable ends for soldering. These nicks can turn into fractures and even open circuits with frequent (and possibly rough) handling!
1 Like
Hi @BBK ,
How are you getting on with your Vector Network Analyser?
I have been busy repeating many of the REW measurements I had previously undertaken on my system using various candidate loudspeaker cables.
(Readers may recall that I needed to repair one end of one Naim NAC A5 cable).
Following the previously reported RED card situation and subsequent email exchanges - where Linn Products Ltd and I suggested alternative brands of loudspeaker cable to try - I evaluated two more USA branded loudspeaker cable types.
Unfortunately, although those cable performed very well and I was also able to obtain all the necessary electrical data, the two alternative USA branded cables did not satisfy the ‘critically damped transient response’ assessment criteria.
Accordingly, I decided to eliminate all USA loudspeaker cable brands from the competition.
Below is an image of the remaining (five) cable pairs in the competition.
Original Image: Author provided, no reuse.
Below is a screenshot of the summary REW files of L+R test signal stimulation (FYI, I have separate files for R only and L only test signal stimulations).
Original Image: Author provided, no reuse.
Further analysis and results to be posted soon.
ATB
E of E
2 Likes
Does the “English” in the thread title need changing to British, or European, or perhaps NAW (Non American World)?
And BTW, something I meant to say before: wouldn’t the more usual term along the lines of ©️EoE be simpler and more readily recognised by others than the footnote you add to images?
1 Like
hello Edmund,
the VNA arrived , but I was busy with something far more interesting. I was trying to modify the playback chain’s kernel so it works like analogue tape Kernel, because that closes the gap between digital and analogue playback. It is actually something you may verify with REW - from green to blue…
1 Like
HI @Innocent_Bystander ,
Thank you for the suggestions, particularly regarding marking the images I upload with ©️EoE instead of my previous method.
I would like to retain ‘English’ in the thread title, as in ‘English Premier League’, for the football reference. Furthermore, because I have edited the title so many times for humour and updates, keeping the first part consistent helps.
I recognise that the title might seem less accurate now the North American brands of loudspeaker cable have been eliminated from the cable competition. However, the world is a very big place (c.f. International), I still have Magico loudspeakers (and an ATI amplifier for AV use), and I am pretty sure some electronic circuit components in my main system are sourced from the USA.
Hi @BBK ,
Thanks, great post!
Are you able to describe in more detail about how you are modifying the
“playback chain’s kernel so it works like analogue tape Kernel”
in your system?
From your user profile it is not clear to me which tools or techniques you might be using to control the impulse response in the way you have indicated. Also, do you have the corresponding amplitude and phase response vs (log) frequency for the two example impulse responses posted?
FYI Simon-in-Suffolk and I were in a discussion (nearly a year ago) about digital filtering and the impact on listener perception.
Link to that discussion (which includes example impulse responses) is post number 324 here.
A link to a follow on discussion from February this year is here (post number 372).
FYI I have been using the impulse response views in the REW tool to make judgments about which cables are superior. I am trying to get to the point where I can post such graphs in a way that makes it obvious as to which cables are superior and why.
It may be of interest that I have found looking at the relative magnitude, ‘clarity’ and consistency of the impulse from the main reflection from back wall - to - front wall and back again to the microphone (which corresponds to about 26ms in my room) to be informative of cable differences. I have developed a few ideas as to why this might be - but I need to be more certain before posting.
Other analysis in REW that I have found helpful to clearly and consistently separate cables by performance in my system is the (acoustic) Distortion views.
Thank you for pointing out your earlier discussion. I had missed it, but the direction is very much as you suggest there.
A brick in the time domain corresponds to infinite extent in the frequency domain, and a brick wall in the frequency domain corresponds to infinite extent in the time domain. Conventional FIR/IIR filtering inevitably reflects this trade-off as pre-ringing, post-ringing, or both, particularly when the filters are designed for very steep frequency-domain transitions.
Relaxing the transition bandwidth from, say, the conventional ~5% to around 38% dramatically reduces the sidelobe amplitude and the energy outside the main lobe.
Have a look at the metrics.
Importantly, you do not have to sacrifice the audio band from 10 kHz upward, as you suggest. You can first upsample the signal to the maximum rate supported by the DAC, say 384 kHz, using sinc reconstruction (for example in Audirvana), and then apply a custom FIR at that higher sample rate. That is essentially what I have been experimenting with this summer.
Most conventional FIR families are primarily optimized in the frequency domain, with narrow transition bands and strong stopband rejection as the main objectives. I have instead built a new family of FIR filters optimized primarily for time-domain concentration, together with a plugin that works in Audirvana as an apodizing pre-filter ahead of the DAC.
Within the audio band, amplitude and phase remain flat. The FIR is symmetric and therefore linear-phase.
The nice part is that none of this requires new hardware. It works perfectly well with the DAC and system I already have.
1 Like
Hi @BBK
Thanks for sharing this work of yours…
I think what you are doing is very interesting, particularly the degree of control over the sound it facilitates with the Naim based system you have assembled.
I will share a little more about my technical backgrounds as it may save us both a bit of time in posting as we converse about what we are doing with our audio systems.
Full Disclosure:
A major part of my PhD in electronic communications required me to study and design digital filters for implementation on a real time FIR DSP card for a specific audio baseband signal processing application. This work was undertaken prior to 1985. There are pictures of the test kit at post numbers 294 and 298. The actual kit appears in the centre of the image at post number 311 (link here).
For added credibility I include a picture below of one of the relevant pages of the PhD.
©️EoE
There are an earlier group of posts (starting at 101 on this thread) which give some information about digital signal processing system I designed a few years later (link here), with the original posting on a different thread (here).
This post is too long now to go into the ‘choosing loudspeaker cables’ work.
Feel free to comment or just wait a while as I add more information on my work to ‘correlate’ the electrical parameters (of LS cables) with LT Spice simulations of the LS cables in my system with the REW analysis of acoustic measurements at listening position AND comments by listeners (other than me) of user selected test track replay.
4 Likes
Hi @Edmund-of-Essex ,
Your 1985 study is very interesting and, in my view, supports the thesis that the focus was, and perhaps still largely is, on the frequency domain. As your study illustrates, the key optimization criteria were stopband and passband ripple as a function of transition width.
If we compare one of the best analogue players with almost any high-quality digital DAC today, digital outperforms analogue in virtually every conventional measured parameter. Yet there remains one conspicuous difference in the time domain: the impulse response.
For me, this is the last major gap to be closed, and that has been the objective of my work.
2 Likes
That is wrong and need to correct it. The temporal coherence only works for native (not up-sampled) material. Thus @Edmund-of-Essex you are right!
Today managed to finish the virtual cable link between Audirvana and REW scope. The up-sampling with the sinc function already introduces long ringing.
I had assumed that a 44.1 or 48 kHz signal could first be transparently up-sampled to a higher sample rate and then processed with the wide-transition Black-19 filter to obtain the same temporal advantage as with native high-rate material.
That is not correct.
Band-limited upsampling is based on sinc-derived interpolation. The ideal sinc kernel has infinite time support, and practical implementations use a finite, windowed approximation. In either case, the interpolation stage itself introduces a long oscillatory impulse response with pre- and post-ringing.
I confirmed this directly by measuring a 48 kHz single-sample impulse upsampled 4× to 192 kHz. The resulting 192 kHz samples already contain the symmetric ringing of the interpolation filter before Black-19 is applied.
Therefore:
48 kHz→conventional band-limited up-sampling→does not provide the same temporal concentration as applying the custom FIR directly to genuinely high-sample-rate material.
With the virtual cable however was able to test the idea without the hardware convolution and the temporally optimized Kernel i got is this:
The input signal was 0 0.5 0 in the digital domain.
2 Likes
Hi @BBK
Thank you for sharing all this highly technical activity of yours! I am enjoying reading about your endeavours with your Naim system. I guess we (on the forum) are just trying to get ‘into the music’ in the most realistic way achievable with the kit we have got in our home environments.
The convergence of recognising how something sounds (using only one’s own ears) with the measurement data from some of the tools you and I are using does take some dedicated study and practical experimentation - and really is a sub niche within a niche of the world of HiFi.
Also, thank you for crediting my earlier posting 
.
I fully agree with your point about the importance of considering the time domain waveform for listener perception and listening experience. FYI, I have posted previously on this topic about one year ago on another thread,
The link to the specific posting is here.
I suspect you are already aware that transformations between time and frequency domains need careful consideration. However, because I have been doing this sort of thing since the very early 1980s, I can generally undertake such transforms ‘pictorially’ in my head (obviously to a very low level of accuracy).
Periodically I write HiFi manufactures about the need to present time domain information as well as frequency domain information in their tools. For example I wrote to Linn Products Ltd (~two years ago) in respect of the Space Optimisation tool:- “Only the amplitude response of correction filters is visible to users - not the impulse response, or phase.”
PS: Naim are just as guilty in the lack of information as Linn Products Ltd in this respect in the Focal/Naim CI-102 management software.
2 Likes
I have a problem - lots of information to post, but maybe some information (for example notes taken by listeners) is more interesting (to forum members) than the technical information which I usually post?
The following are photos of the notes taken by one listener on three different days of a total of 5 listening sessions. These notes have had listener identification redacted (by orange paper) and subsequently annotated to show which loudspeaker cables were in place for each session.
PS: This particular listener was aware that loudspeaker cables where the subject of assessment - but was not aware of any information beyond that.
©️EoE
©️EoE
©️EoE
I plan to provide more context and linkage with technical and measurement information over the next few posts.
1 Like
Hi All,
Highly technical posts to come - read on if interested and please do comment if you see any problems or questionable assumptions I have made in the presentation of this work.
Input from @BBK , @Innocent_Bystander and @Simon-in-Suffolk would be greatly appreciated. There are likely quite a few others who will have relevant knowledge - I would definitely appreciate input from all quarters.
As previously established, the objective is to find the loudspeaker cable (when fitted in the system between the Linn amplifiers and Magico M2 loudspeakers) that provides the best ‘critically damped transient response in sound pressure’ at the listening position.
The four steps in the process to achieve the objective are:
-
Obtain the electrical parameter data for the cable (either by obtaining the data from the manufacturer or measuring using LCR meter).
-
Construct a sufficiently valid circuit model of the output stage of amplifier, loudspeaker cable (from data at Step 1) and loudspeaker in simulation tool such as LT Spice. Using risetime controlled square waves as simulation input, observe the details in the rise of power delivered into the simulated loudspeaker, paying particular attention to the top of the square wave.
-
Undertake sine sweep acoustic measurements at listening position of the actual system with the candidate cables. Analyse measurements using tool such as REW with assessment criteria focused on Distortion, Group Delay and Impulse Response.
-
Undertake sufficiently ‘masked’ human listening tests of the actual system with the candidate cables to note perceptions of variety of users to selected music tracks.
Table below is the electrical data (corresponding to Step 1 of the process) for the final five loudspeaker cables that remain in the competition.
©️EoE
Technical Disclaimer: Whilst all data presented in the table above were obtained during the last 11 months with a calibrated LCR meter, please note that the entries should not be taken as official measurement data or representative of any official specifications of the cables.
For that data or information - please contact the manufacturer (or in country distributor) directly.
1 Like