Unbalanced vs Pseudo-balanced interconnects (inc. DIY)

Huge posted:

Timeo techncratos et donae ferentes!

Quidquid latine dictum, altum videtur.

Et quod scripsi, scripsi.

Hi Huge,

I'm interested in reading the results of your listening tests with the DIY cable.

Preferably in English.

Interesting bit of work there Huge. I'll be interested in what you find when you make up the cables and get to listen to them. 

James

How did you model the difference between Mogami and other cables of similar geometry?

Or - on what basis did you pick Mogami?

Mogami W2549's excellent cable. I've tried various other brands for interconnects but to my ears it's the best for DIY. 

Hi Adam,

To answer the last question first - because of recommendation from Mike-B, followed by research on t'internet.

So far I haven't modeled any other cables, but I may do so.  (Any cables you specifically want? Any other cables you'd recommend? Do I detect a hint here?)

The model for the Mogami cables was based on:
1  the specs they give for capacitance (core to screen and core to core) and core to core mutual inductance
2  other inductances calculated from the cable geometry
3  the assumption of 8mm of screen being unwound to connect to the solder buckets of the DIN connector.

I also did a calculation of the capacitances from first principles just to check for cheating in the specs, and got values within 20% of those quoted by Mogami in their spec sheets, so I have confidence that they're not 'fudging' their values significantly (actually the calculated values were about 5-15% lower than the quoted values, so Mogami may be being a little bit conservative to make some degree of allowance for manufacturing tolerances).

Latest piece of modelling was to vary the source parallel capacitance, load capacitance and stray capacitance from the ground plane of the receiving amp and see how stable the solutions were to varying circumstances of usage (some show worrying resonance peaks where they are acting as tuned circuits interacting with the source and load reactances).

Update on the modelling exercise...

Results of additional modelling

I’ve tested models covering most combinations of the following conditions

Source output impedance
600Ω with 10pF to 10nF + 0.01Ω to 50Ω in parallel

Receiver input impedance
50kΩ with 2pF to 470nF in parallel

Single-ended source with unbalanced differential receiver
Unbalanced differential source with single-ended receiver
Unbalanced differential source with unbalanced differential receiver *

Coax cable shield terminated both ends
Pseudo-balanced, shield terminated source end only
Pseudo-balanced, shield terminated receive end only *
Pseudo-balanced, shield terminated source end, 22pF to 1nF terminator (RF bypass) capacitor at the receive end

* For these configurations, only a reduced set of source and receiver terminating impedances were tested.

There are five points that have emerged quite clearly but as this is theoretical only, they are no more than theoretical guidance notes:

• At any end that has a single ended configuration, the shield should normally be terminated to 0V / Earth.

• At any end that has an unbalanced differential configuration, the shield should normally not be connected to 0V and should be left unterminated.

• If both ends have unbalanced differential configuration, a balanced or pseudo-balanced configuration should be used. 1) If either end provides a facility to terminate the screen to earth (e.g. the casework) the screen should be terminated to the casework at that end.  2) If no earth connection is available, the connection should be wired pseudo balanced with the screen connected to zero volts at the source end only.

• When a capacitor is used to terminate the shield to 0V at an unbalanced differential receiver, the protection at high RF (above about 200MHz) is improved at the expense of protection at lower RF. The precise frequency / amplitude / phase response is highly dependent on the terminator capacitor, the receiver input impedance and stray capacitance 0V to Earth within the receiver component itself.  As such the performance of this arrangement is extremely difficult to predict in the real world and its use without extensive testing in situ, is not recommended.

• Doing a modelling exercise on this scale makes my brain hurt.

N.B. If both ends have single ended configurations, then you have an earth loop, and the type of interconnect used is probably going to be the least of your worries!

Dear Huge, thanks for posting.

Have sense to apply your considerations to a din/xlr cable that someone have between pre- and power-amplifier?

bicela posted:

Dear Huge, thanks for posting.

Have sense to apply your considerations to a din/xlr cable that someone have between pre- and power-amplifier?

In general terms, a 5 pin DIN to 3 pin XLR when used as a stereo interconnect behaves in the same way as a 5 pin DIN to 5 pin DIN when used as a stereo interconnect.

However an interconnect between a Naim preamp and a Naim poweramp is classed as a SNAIC, and it's against the forum AUP for me to comment on that, even in the Padded Cell.

Result 1 Update:

I've now made and fitted the Mogami W2549 interconnect 400mm long, wired as a traditional twinax arrangement, that is the two cores carry the L&R lines and the screen caries the zero volts return current.

Conclusions, compared to the 'Lavender' interconnect:

The sound seems better organised.  Intelligibility of voices is improved as is timbral accuracy.  Both these qualities are improved when the relationships of harmonics within a tone are better preserved, so the brain can associate the right set of harmonics with the fundamental, and so get a clearer picture with less work.  I believe this may be the case here.  These changes are very beneficial and the system now clearly distinguishes more subtle shading in the music as well as making lyrics clearer.  The improvements to clarity here are similar in magnitude to adding an external power supply to the pre-amp.

There have also been two very slight shifts in overall balance:  The HF seems a little sharper, and the bass seems a little lighter.  However these are quite minimal changes, and I've adapted to them in less than an hour of listening.

Result 2 Update:

Mogami W2549 interconnect 400mm long, directionality:

I think there may be a very slight difference in the way the assembled cable behaves in each direction - intelligibility seems slightly better in the direction of the writing on the sheath, but these tiny differences could easily be accounted for by differences in soldering, or my head being in a different place by a few cm while listening.  Remember the cable is hand soldered, so there will be differences at each end (e.g. small residual stress).

Given the small nature of the difference: either the bulk cable isn't significantly directional (what I actually believe), or a difference does exist but it's still very slight and could easily be dominated by differences in sound quality between different manufacturing runs or different assembly techniques.

Hi Huge,  interesting listening results,  I think I understand, but to be sure & that we all do,  can you clarify the IC connection type(s)  & what they are connecting on your system.  

My NDX & NAT IC's to Supernait (DIN-DIN) are s short as possible @ 28cm & 37cm  & the cable type & config is per your Result-1   .........  I agree all the notes except bass   ".....bass seems a little lighter"     I found the opposite,  bass was a touch stronger in the mix.    

The first interconnect (the only one I've tested so far) is using 2549 twinax.  The two cores carry the L & R signals connected to pins 3 & 5 (Naim call these Ch1 & Ch2).  The screen carries the 0V reference (and hence earth) connection, connected to pin 2 (this is the connection Naim call -ve).  This is the classic 'unbalanced' form of interconnect cable.

The IC I'm testing is from the ND5 XS to the preamp.

I have assembled the pseudo-balanced interconnect, but I'll wait 'til tomorrow so I can test it earlier in the day when my senses are fresher.

Mike,

There may be a difference in our perception of the bass because my system uses a sub - it has flat FR down to 15Hz.
I could have shortened the interconnect to about 35cm, but I allowed 40cm to give a little slack rather than risking putting force on the connectors.

I had forgotten your sub Huge,  but still interesting why your perceive a lighter bass.    My speakers go to 32Hz & are still moving the SP meter around at 25Hz..     My initial experiments were comparing various cable makes - Lavender, AR Sounds, Chord & FBS,  & three makes of bulk cable & thats how I ended up with Mogami W2549.  My notes for 2549 bass include firm, texture & tone.     The length thing was my OCD about untidy cables looping around & also getting as low C,L&R as possible.       

Mike,

Now that's most interesting, I'd agree with the comment of firm texture in the bass.  In my system, although the bass appears lighter, I would consider it to have retained it's firmness of texture, despite the perception of lighter presentation.  Turning the sub up a tiny bit (probably no more than 1 or 2 dB needed) would restore the perceived level and increase the sense of firmness above that of the Lavender.  So really we're splitting hairs here (as opposed to splitting infinitives).

Result 3 Update:

I've now made and fitted the Mogami W2534 interconnect 400mm long, this is wired in a stereo pseudo-balanced configuration, viz-
Pin 3  >   Core 1 - Signal 1               >  Pin 3
Pin 5  >   Core 3 - Signal 2               >  Pin 5
Pin 2  >   Core 2 - Return 1 (0V)     >  Pin 2
Pin 2  >   Core 4 - Return 2 (0V)     >  Pin 2
Pin 2  >   Screen -                             (no connection)

Conclusions, compared to the Mogami W2549 (unbalanced twinax) interconnect:

The differences here are quite a bit more more subtle but still similar to the improvements of the W2549 interconnect over the 'Lavender' interconnect.  The biggest gain (for me) is a significant further increase in intelligibility.  There are also some minor gains in sorting out different elements within the overall sound, for example, it's easier to hear when a vocalist is singing in unison with herself - the two separate instances of the same voice singing the same note with the same shaping and timing are more easily distinguished as separate entities contributing to the overall sound, rather than just melding together.   There may also be a slight decrease in background clutter, but I'm not at all sure if this is a significant change or not - it's a suggestion of difference, rather than a definite difference; so could easily be a subconscious expectation bias.  The overall presentation, bass & HF response are identical as far as I can tell.

Whether these lesser gains are worth the hassle of trying to solder two cores and the screen to the 0V pin on the DIN plug is another question (pin 2 is in the middle, and so is the most awkward position).

I'll try to post pictures of the soldering job later on.  Going back to the music now.  

OK, further listening has revealed another difference - one in presentation of musical tension.  It's difficult to describe exactly, so I'll provide two paradigms from different genres...

1  Edvard Grieg, "1st Norwegian Dance".  In the second passage Grieg introduces a new theme, he develops this through the passage gradually building tension, then resolving it, before reverting to the first theme to end the dance.  With the W2534 interconnect, the building of tension is faster and more powerful, and the resolution more dramatic.

2  Ari Mason, "The Curse".  This is one of my normal 'test tracks': The whole track is built on tension and the effect is easily lost on lesser systems.  With the 'Lavender' interconnect I can feel it cause my chest and my breathing to tighten up.  With the W2549, the effect was enhanced a bit, extending the tension to my neck and shoulders.  Using the W2534 interconnect it actually starts to induce adrenalin shock: in addition to breathing and muscular tension, the tension causes all the hairs on my head, neck and arms to stand up, I even start to sweat and hyperventilate.  It's quite freaky actually.

More (personal and anecdotal) listening results comparing the Mogami W2534 used in pseudo-balanced configuration, against the other two cables.

Note that I’ve only documented where I found differences, for many pieces of music I found the two Mogami cables virtually indistinguishable.  Indeed in terms of resolution, background clutter and overall balance I could discern no differences between them; however both were almost always preferred to the ‘Lavender’ interconnect, and the ‘Lavender’ was never found to give the better result.

3  Antonio Vivaldi, “Quatro Stagioni, ‘Printemps I’”.  Drottningholm Baroque Ensemble.  When the rain shower hits suddenly, it’s now such a sharp dynamic contrast that I jump reflexively.  Previously I haven’t experienced this effect with this recording.  However the recording is very bright, and that appears highlighted to a point that’s only just acceptable; it should be noted though that this is a period instruments recording and that is the nature of this ensemble.  (This is the same for both Mogami cables and a similar brightness is also noted in the brass section of the Hallé orchestra in the Manchester Free Trade Hall – a characteristic I remember from live concerts; so I believe this to be accurate rather than an accentuation.)

4  Josef Haydn, “Symphony No. 104, ‘London’”, Orch 18^th C / Bruggen The sheer ‘joie de vie’ of this music is conveyed in a more complete and more definitive manner.  The contrast to the more stately second movement is also more pronounced.

5  Gustav Holst, “Suites for Military Band, ‘Chaconne’ (& others)”, Cleveland Symphonic Winds / Fennel (Telarc).  The reproduction of the bass drum seems to be more even through the piece and the ‘skin slap’ from striking the drum is more easily distinguished within the initial transient.  The strength of the transient has less masking effect on other bass instruments as there appears to be less audible ‘smearing’ of the transient.  The balance of the piece seems better preserved.

5  Beethoven “Symphony No. 7”, SCO / Mackerras.  In the first movement the contrast between first stanza (the conflicted theme) and the soaring theme that overtakes it, is more clearly defined than previously.  The extension of the conflict continuing under some (but not all) of the reiterations of the soaring theme is also much clearer.  The brooding mechanistic darkness of the second movement is again more starkly portrayed.  These changes serve to further highlight the battle to get the strength to overcome in the third movement, and the final victory of an indomitable spirit in the fourth.

6  Hector Berlioz, “Sinfonie Fantastique”, Concertgebouw / Davies.  Here again, the contrasts of mood are portrayed with significantly greater differentiation, making listening even more involving and pressing home the meaning of the piece more dramatically.

7  Hungry Lucy “Pulse of the Earth”.  Changing to the W2534 cable intensifies the brooding mood of inevitable tragedy that suffuses the tracks and builds through the album, making the final dislocation even more poignant.

I’ve concluded, that for me, the way this cable preserves mood is well worth the extra hassle involved in soldering the connections at the shielding end.  Here’s two photos:

(But it seems that Hoop.La no longer supports access to Google Drive!, however the links do work)

Huge posted:

More (personal and anecdotal) listening results comparing the Mogami W2534 used in pseudo-balanced configuration, against the other two cables.

Note that I’ve only documented where I found differences, for many pieces of music I found the two Mogami cables virtually indistinguishable. 

Hi Huge,  interesting re your comment on "I found the two Mogami cables virtually indistinguishable".    Happy with that as it confirms my findings those years ago & has not giving me cause to think about changing my Mogami W2549 IC's ..... phew

Using my last bits & pieces of left over Mogami W2534,  I've just made up a 2 metre RCA to DIN to connect a Chord 2Qute to 252 for a friend.  I stripped back the RCA end by 20cm & tight twisted each of the channel pairs in the same direction as in the 4-quad config.  I did nothing with the screen,  leaving it floating at both ends.  I finished the RCA ends with expandable nylon braid & tidied & held in place by shrink wrap at each RCA & the junction of the 2534 cable outer.  The DIN end with both -ve channels on pin-2 was a bit of a fiddle,  but I found cutting back aprx 50% of each of the flex strands worked OK.

My buddy had the 2Qute connected with a 1m long Chord  but it left the 2Qute balanced (hanging) on the side of the rack.  The new 2m long IC allowed the 2Qute to mount on a different shelf.   I did not spend a lot of time listening, just making sure it was installed & worked,  but he was on the phone last evening extolling how much better it is.   (not sure myself, but he sure likes it)    Now he's asking if I can make a special length BNC-BNC coax & have the 2Qute on another rack.  OK if I can find a pukka digital cable  ..........  need to git some book learning dun.  

Mike-B posted:

Huge posted:

More (personal and anecdotal) listening results comparing the Mogami W2534 used in pseudo-balanced configuration, against the other two cables.

Note that I’ve only documented where I found differences, for many pieces of music I found the two Mogami cables virtually indistinguishable. 

Hi Huge,  interesting re your comment on "I found the two Mogami cables virtually indistinguishable".    Happy with that as it confirms my findings those years ago & has not giving me cause to think about changing my Mogami W2549 IC's ..... phew

Using my last bits & pieces of left over Mogami W2534,  I've just made up a 2 metre RCA to DIN to connect a Chord 2Qute to 252 for a friend.  I stripped back the RCA end by 20cm & tight twisted each of the channel pairs in the same direction as in the 4-quad config.  I did nothing with the screen,  leaving it floating at both ends.  I finished the RCA ends with expandable nylon braid & tidied & held in place by shrink wrap at each RCA & the junction of the 2534 cable outer.  The DIN end with both -ve channels on pin-2 was a bit of a fiddle,  but I found cutting back aprx 50% of each of the flex strands worked OK.

My buddy had the 2Qute connected with a 1m long Chord  but it left the 2Qute balanced (hanging) on the side of the rack.  The new 2m long IC allowed the 2Qute to mount on a different shelf.   I did not spend a lot of time listening, just making sure it was installed & worked,  but he was on the phone last evening extolling how much better it is.   (not sure myself, but he sure likes it)    Now he's asking if I can make a special length BNC-BNC coax & have the 2Qute on another rack.  OK if I can find a pukka digital cable  ..........  need to git some book learning dun.  

Mike the first part of "for many pieces of music I found the two Mogami cables virtually indistinguishable." is also quite important, as for some other pieces the differences were considerable.  Essentially I found that, for my system at least, the W2534 configured as pseudo-balanced, preserved the mood of the music better where the piece was attempting to portray a particular mood.  In all other ways they were virtually indistinguishable.  I suspect the difference is not the cable itself, but the pseudo-balanced construction.

In respect of a BNC - BNC digital coax, you could try using RG6 - it's considerably better than RG59 in all ways and will preserve clean digital edges better.

Use of 2534 to do a DIN-2xRCA is an interesting exercise - soldering 2534 to pin2 is a bit of a pain, try it when connecting the shield to pin2 as well!

Huge posted:

In respect of a BNC - BNC digital coax, you could try using RG6 - it's considerably better than RG59 in all ways and will preserve clean digital edges better.

I'm using WF100. Nominal DC resistance is significantly better & attenuation a touch better than RG6.   The main concern was if it needed  to be flexible & would go for a Canare,  but not needed,   

A last observation:

Using the pseudo-balanced configuration into an unbalanced differential input - as per the theoretical models - the RFI rejection does indeed seem better than the simple unbalanced (twinax) configuration.  So much so that I've been able to drop the clip on ferrite without the level of 'background clutter' increasing.

With a twinax cable ('Lavender' or W2549) removing the ferrite from the amp end of the cable results in a loss of apparent resolution in the music.  Without the ferrite in place, I believe low level details are being masked by the audible effects of the RFI.

You must have a seriously high RFI background Huge,  you've mentioned it before but do you know where it's coming from?

Mike, sadly I don't, there's no visible signs of industrial activity in the area, other than a BT exchange, a shop with a cold store and a suite of offices, all about 400m away.

I live in a 300+yr old terraced house that was re-fitted by the council in the early 60's and re-wired in the late 80's.  All the houses are listed, and in all (both this and the other terrace here) there's a dearth of power sockets and just one telephone socket; but the whole area is becoming more gentrified, hence high levels of technology are in use.  As a result I think a lot of people round here use extension cords AND PLAs (and in some cases may be multiple PLA channels in one house!) and that's a bad combination as the extension cords just act as additional antennae.  They also heavily rely on WiFi, so there's now a lot of channel contention here as well (sometimes I can pick up 12 WAPs).