Barry
05-09-2012, 23:43
Introduction
Those members that know me might be thinking: “What is AoS’s resident cable sceptic doing listening to cables?” Well the important adjective here is “digital”. I continue to remain sceptical on the acoustic differences of analogue interconnects, but in the digital domain, interconnects have a harder job to do.
Analogue interconnects only have to convey signals up to say 30kHz (to encompass overtones and harmonics), so any cable of reasonably low capacitance and low loop-resistance per unit length will suffice.
Cables conveying digital signals between a CD transport and an external DAC have to transfer signals of much higher frequency. The sampling waveform can have a 10% to 90% rise time as short as 25ns, implying frequencies as high as 14MHz. The Sony-Philips digital interface (S/PDIF) is specified to be 75Ω via an RCA phono connector. This suggests digital cables should have a characteristic impedance of 75Ω. Strictly speaking treating cables as transmission lines really only makes sense if the length of the cable is longer than λ/2π, or in this case ~ 2 metre. There is also the moot point that RCA phono connectors can never have an impedance as high as 75Ω (and those that claim they do, don’t when mated with a standard RCA socket). For this reason 75Ω impedance BNC connectors are preferred, though it must be admitted, such deviations in characteristic impedance occur over a short electrical length and tend to mutually cancel. The only other property a digital cable ought to have (in my opinion) is good shielding or screening effectiveness to external interference, and this is related to the surface transfer impedance (Zt) of the shielding of the cable.
The reason I undertook these listening tests was because I wanted to try using an external DAC, and compare the results with the internal DAC of my Sony player (model CDP XB-930). At this point I must thank two AoS members (DaveM and The Grand Wazoo) for the long-term loan of a couple of DACs. Thanks Guys. Whilst auditioning the DACs, I thought it would be interesting to compare interconnects as well.
The cables
I tried four commercially available digital cables – and since resenting paying large sums for cables (of any colour), I assembled three others. The latter were experimental variants of the commercial designs.
Cable 1
This was a cable manufactured by Chord. It was unmarked and all I can tell you is the outer insulation sheathing was a cream colour and it was about 1 metre long. It was fitted with gold-plated RCA phono plugs of good quality, but these were covered by black heat shrink sleeving, so I can’t tell you too much about them. I assume the cable has a characteristic impedance of 75Ω.
Cable 2
Manufactured by Cambridge Audio, this nicely flexible cable has its sheathing clearly marked with “Digital 75 Ohm Double Screened High Resolution Interconnect”. Fitted with very high quality gold-plated RCA phono plugs and 1metre long, this cable is surprisingly cheap, being available from eBay for £8 (!)
http://i780.photobucket.com/albums/yy88/barrydhunt/499.jpg
http://i780.photobucket.com/albums/yy88/barrydhunt/500.jpg
Cable 3
This cable assembly is manufactured by Pomona Electronics, using Belden 8263 cable the latter being identical to RG-59B/U. Clearly of 75Ω characteristic impedance, this cable came fitted with a BNC connector at one end and an RCA phono plug at the other; both of very average quality. 3 metres long, it was found in my cable junk box, so I thought it was worth trying. The resistance of the shielding is 8.5mΩ/m, suggesting a screening effectiveness of 68dB.
http://i780.photobucket.com/albums/yy88/barrydhunt/501.jpg
Cable 4
Manufactured by Trompeter Electronics, this is a triaxial cable, type TRC-75-2, where the inner coaxial cable has a characteristic impedance of 75Ω. This cable is often used for Ethernet LAN installations.
Being a triaxial cable it is terminated with triaxial BNC-style connectors: BNT.
http://i780.photobucket.com/albums/yy88/barrydhunt/496-1.jpg
and so BNT to BNC adaptors had to be used, where the outer of the two electrically insulated screens is left 'floating' and the plug conforms to the dimensions for a 75Ω BNC.
http://i780.photobucket.com/albums/yy88/barrydhunt/497.jpg
Triaxial cables, owing to the extra shield have a very low Zt. For the Trompeter TRC-75-2, used in the above configuration, the screening effectiveness is estimated to be 71dB. The cable length was 2m.
Cable 5
This was a 1m length of the Trompeter TRC-75-2 cable, where both screens were connected together at each end where they were terminated in gold-plated RCA phono plugs.
http://i780.photobucket.com/albums/yy88/barrydhunt/498.jpg
In connecting both screens together, Zt is further reduced to a calculated value of ~2mΩ/m, suggesting a screening effectiveness of 74.5dB.
Cable 6
This was a 1m length of triaxial cable, with the inner coaxial cable being identical to RG-59/B/U. It is available as Belden 88232 (?). Only the centre coaxial section was terminated. 75Ω impedance BNC connectors were used, having a silver-plated finish on the outer contact and bayonet-locking ring. The outer screen was brought out and terminated in a spade connector so as to connect to the chassis of the DACs.
http://i780.photobucket.com/albums/yy88/barrydhunt/495.jpg
Cable 7
This was a 1m length of UniRadio URM70 coaxial cable, having a characteristic impedance of 75Ω. It is a general-purpose video cable, with a single braided screen, it is more flexible than RG-59B/U. It was terminated in Neutrix REAN RCA phono plugs.
http://i780.photobucket.com/albums/yy88/barrydhunt/494.jpg
(shown on right)
Methodology
Nothing scientific - I tried them out in random order, for a few days each, over a period of about three months. All seven were tried with two DACs: an Arcam Black Box 50 and a Beresford TC-7520. A variety of music was used. No notes were taken apart from some very general impressions.
The results
In general all the cables sounded very similar. There were some very, very small differences, but all ‘preserved’ the differences displayed by the two DACs. There was a theme or trend to the results, which might be of interest to members.
I found the shorter cables (1 metre) (cables 1,2, 5, 6 and 7) to be slightly better regarding focus and transient attack. This might be due to possible aliasing effects caused by the longer cables. I don’t know – I’m not particularly knowledgeable about matters digital.
I also thought I preferred those cables with additional screening (cables 4, 5 and 6), that is the triaxial designs – but this could be an engineer’s whim.
Any improvement wrought using BNC over RCA phono connectors cannot be ascertained, since I had to use BNC to phono adapters on those leads fitted with BNCs, as the both CD player and DACs were fitted with such. Again, it appeals to my engineering background to prefer BNC connectors.
It was gratifying to note that the results using good quality 75 Ω cable were as good as using (understandably) more expensive commercial digital interconnects.
As an aside (and as a result of a member’s question), I briefly tried an analogue cable (of modest quality: one of a set of leads that came with a Nakamichi cassette deck). I have to say the results were not encouraging. The sound stage was ‘smeared’ and lacked focus, and at times the treble was ‘splashy’. Sorry – it would seem you have to use cables having an impedance of 75 Ω.
Conclusions
If you can solder, make up your own interconnects using RG-59 cable or URM 70, 75Ω cable and good quality RCA phono connectors. Both cable types are available from Maplin for about £1 a metre. Cable used for connection to TV satellite dishes may also prove to be usable. I would suggest the digital interconnect need only be 0.5m long.
Of the triaxial cables I tried, I want to do some further tests with the Belden 88232 and the Trompeter TRC-75-2, and experiment with the earthing arrangements.
Those members that know me might be thinking: “What is AoS’s resident cable sceptic doing listening to cables?” Well the important adjective here is “digital”. I continue to remain sceptical on the acoustic differences of analogue interconnects, but in the digital domain, interconnects have a harder job to do.
Analogue interconnects only have to convey signals up to say 30kHz (to encompass overtones and harmonics), so any cable of reasonably low capacitance and low loop-resistance per unit length will suffice.
Cables conveying digital signals between a CD transport and an external DAC have to transfer signals of much higher frequency. The sampling waveform can have a 10% to 90% rise time as short as 25ns, implying frequencies as high as 14MHz. The Sony-Philips digital interface (S/PDIF) is specified to be 75Ω via an RCA phono connector. This suggests digital cables should have a characteristic impedance of 75Ω. Strictly speaking treating cables as transmission lines really only makes sense if the length of the cable is longer than λ/2π, or in this case ~ 2 metre. There is also the moot point that RCA phono connectors can never have an impedance as high as 75Ω (and those that claim they do, don’t when mated with a standard RCA socket). For this reason 75Ω impedance BNC connectors are preferred, though it must be admitted, such deviations in characteristic impedance occur over a short electrical length and tend to mutually cancel. The only other property a digital cable ought to have (in my opinion) is good shielding or screening effectiveness to external interference, and this is related to the surface transfer impedance (Zt) of the shielding of the cable.
The reason I undertook these listening tests was because I wanted to try using an external DAC, and compare the results with the internal DAC of my Sony player (model CDP XB-930). At this point I must thank two AoS members (DaveM and The Grand Wazoo) for the long-term loan of a couple of DACs. Thanks Guys. Whilst auditioning the DACs, I thought it would be interesting to compare interconnects as well.
The cables
I tried four commercially available digital cables – and since resenting paying large sums for cables (of any colour), I assembled three others. The latter were experimental variants of the commercial designs.
Cable 1
This was a cable manufactured by Chord. It was unmarked and all I can tell you is the outer insulation sheathing was a cream colour and it was about 1 metre long. It was fitted with gold-plated RCA phono plugs of good quality, but these were covered by black heat shrink sleeving, so I can’t tell you too much about them. I assume the cable has a characteristic impedance of 75Ω.
Cable 2
Manufactured by Cambridge Audio, this nicely flexible cable has its sheathing clearly marked with “Digital 75 Ohm Double Screened High Resolution Interconnect”. Fitted with very high quality gold-plated RCA phono plugs and 1metre long, this cable is surprisingly cheap, being available from eBay for £8 (!)
http://i780.photobucket.com/albums/yy88/barrydhunt/499.jpg
http://i780.photobucket.com/albums/yy88/barrydhunt/500.jpg
Cable 3
This cable assembly is manufactured by Pomona Electronics, using Belden 8263 cable the latter being identical to RG-59B/U. Clearly of 75Ω characteristic impedance, this cable came fitted with a BNC connector at one end and an RCA phono plug at the other; both of very average quality. 3 metres long, it was found in my cable junk box, so I thought it was worth trying. The resistance of the shielding is 8.5mΩ/m, suggesting a screening effectiveness of 68dB.
http://i780.photobucket.com/albums/yy88/barrydhunt/501.jpg
Cable 4
Manufactured by Trompeter Electronics, this is a triaxial cable, type TRC-75-2, where the inner coaxial cable has a characteristic impedance of 75Ω. This cable is often used for Ethernet LAN installations.
Being a triaxial cable it is terminated with triaxial BNC-style connectors: BNT.
http://i780.photobucket.com/albums/yy88/barrydhunt/496-1.jpg
and so BNT to BNC adaptors had to be used, where the outer of the two electrically insulated screens is left 'floating' and the plug conforms to the dimensions for a 75Ω BNC.
http://i780.photobucket.com/albums/yy88/barrydhunt/497.jpg
Triaxial cables, owing to the extra shield have a very low Zt. For the Trompeter TRC-75-2, used in the above configuration, the screening effectiveness is estimated to be 71dB. The cable length was 2m.
Cable 5
This was a 1m length of the Trompeter TRC-75-2 cable, where both screens were connected together at each end where they were terminated in gold-plated RCA phono plugs.
http://i780.photobucket.com/albums/yy88/barrydhunt/498.jpg
In connecting both screens together, Zt is further reduced to a calculated value of ~2mΩ/m, suggesting a screening effectiveness of 74.5dB.
Cable 6
This was a 1m length of triaxial cable, with the inner coaxial cable being identical to RG-59/B/U. It is available as Belden 88232 (?). Only the centre coaxial section was terminated. 75Ω impedance BNC connectors were used, having a silver-plated finish on the outer contact and bayonet-locking ring. The outer screen was brought out and terminated in a spade connector so as to connect to the chassis of the DACs.
http://i780.photobucket.com/albums/yy88/barrydhunt/495.jpg
Cable 7
This was a 1m length of UniRadio URM70 coaxial cable, having a characteristic impedance of 75Ω. It is a general-purpose video cable, with a single braided screen, it is more flexible than RG-59B/U. It was terminated in Neutrix REAN RCA phono plugs.
http://i780.photobucket.com/albums/yy88/barrydhunt/494.jpg
(shown on right)
Methodology
Nothing scientific - I tried them out in random order, for a few days each, over a period of about three months. All seven were tried with two DACs: an Arcam Black Box 50 and a Beresford TC-7520. A variety of music was used. No notes were taken apart from some very general impressions.
The results
In general all the cables sounded very similar. There were some very, very small differences, but all ‘preserved’ the differences displayed by the two DACs. There was a theme or trend to the results, which might be of interest to members.
I found the shorter cables (1 metre) (cables 1,2, 5, 6 and 7) to be slightly better regarding focus and transient attack. This might be due to possible aliasing effects caused by the longer cables. I don’t know – I’m not particularly knowledgeable about matters digital.
I also thought I preferred those cables with additional screening (cables 4, 5 and 6), that is the triaxial designs – but this could be an engineer’s whim.
Any improvement wrought using BNC over RCA phono connectors cannot be ascertained, since I had to use BNC to phono adapters on those leads fitted with BNCs, as the both CD player and DACs were fitted with such. Again, it appeals to my engineering background to prefer BNC connectors.
It was gratifying to note that the results using good quality 75 Ω cable were as good as using (understandably) more expensive commercial digital interconnects.
As an aside (and as a result of a member’s question), I briefly tried an analogue cable (of modest quality: one of a set of leads that came with a Nakamichi cassette deck). I have to say the results were not encouraging. The sound stage was ‘smeared’ and lacked focus, and at times the treble was ‘splashy’. Sorry – it would seem you have to use cables having an impedance of 75 Ω.
Conclusions
If you can solder, make up your own interconnects using RG-59 cable or URM 70, 75Ω cable and good quality RCA phono connectors. Both cable types are available from Maplin for about £1 a metre. Cable used for connection to TV satellite dishes may also prove to be usable. I would suggest the digital interconnect need only be 0.5m long.
Of the triaxial cables I tried, I want to do some further tests with the Belden 88232 and the Trompeter TRC-75-2, and experiment with the earthing arrangements.