Why digital seems to be affected by power and cables

[CraigS]

Neutrik also do a somewhat smaller 20A version which truly is a bargain http://www.canford.co.uk/Products/42-021_NEUTRIK-POWERCON-NAC3FCA-Mains-input-cable-connector-20-Amp .  They have global markets, so all this stuff is going to be available in Aus.

 

The regular IEC60320 C13/C14 IEC connector pair are only rated at 10A 250V by the way.  The high current version, with the flats horizontal (C19/C20 pair) is 16A 250V.  However esorteric the manufacturer, those are the ratings.  

 

Best I can find is that a C13/C14 pair has a contact resistance of 10 milliohms.  Which means that at 10A the live and neutral pins are dissipating 1W each and given the contact size will get noticeably warm.

 

Now you could argue that the 32A Neutrik has <=3 milliohms, and so worst case each contact will dissipate 3W - but the connector is physically huge as compared with even the most esoteric IEC and is designed to handle that power.  In any event, only under exceptional conditions would any amplifier take anything like the current carrying capacity of any IEC or Neutrik (other than a rock concert sound reinforcement rig)

[warweary]

Hey guys, just a little question.

 

Is one of the general opinions about power for our hifi, that if a person has a noisy fridge/washer/kettle whatever, that putting a line conditioner eg, a Thor board, on the noisy fridge/whatever works better than putting the thor board on the hifi powerline.

[CraigS]

Would it be silly to ask if a Power Cable from the wall to my PS Audio P10 would be a waste as the power is being cleaned from the P10 not the mains power?

 

Well that is a bit of a beast!  I know someone with a power conditioner of either this make or similar, and using high capacitance mains cables (Specifically the Russ Andrews ones I designed) between the output and connected gear freaked it out a real treat. We only noticed this because his streamer took forever to lock and then sounded like a bag of hammers.  Because the output comes directly from a high voltage power amplifier it seems prone to instability under certain conditions.  PS will probably deny that is a possibility of course - and in fairness it might not have been their product we were using.

 

So I'd be careful what cables you use, and suspect because the conditioner (a) isolates the incoming mains power and (b) regenerates AC mains with low distortion and very low output impedance I'd tend to go with really simple vanilla mains cables.

[SdP]

Well that is a bit of a beast!  I know someone with a power conditioner of either this make or similar, and using high capacitance mains cables (Specifically the Russ Andrews ones I designed) between the output and connected gear freaked it out a real treat. We only noticed this because his streamer took forever to lock and then sounded like a bag of hammers.  Because the output comes directly from a high voltage power amplifier it seems prone to instability under certain conditions.  PS will probably deny that is a possibility of course - and in fairness it might not have been their product we were using.

 

So I'd be careful what cables you use, and suspect because the conditioner (a) isolates the incoming mains power and (b) regenerates AC mains with low distortion and very low output impedance I'd tend to go with really simple vanilla mains cables.

Thank you for your advice. cheers Simon

Edited February 26, 2016 by SdP

[Red MacKay]

So why the high capacitance if it is so undesirable in a mains lead?

[CraigS]

Hey guys, just a little question.

 

Is one of the general opinions about power for our hifi, that if a person has a noisy fridge/washer/kettle whatever, that putting a line conditioner eg, a Thor board, on the noisy fridge/whatever works better than putting the thor board on the hifi powerline.

 

If you have a noisy contact - ie one that goes splat and puts a spike down the mains each time it closes, they are best cured at source using something like this http://www.russandrews.com/vdr-contact-suppressor/ .  Such kits will doubtlessly be available worldwide from whatever supplier you can find.

 

So a surge suppressing distribution board like you suggest would stay on your audio system.

 

But bear in mind that any of these surge suppressing technologies has a finite life.  They usually make use of voltage dependent resistors to clamp the mains during a spike, and each time that happens the performance degrades.  It is dependent on the total spike energy it sees during its life, so it is impossible to know if it is working or not.  It is all in the Thor technical specification - but since they offer a 6 year replacement guarantee it looks like quite a bargain.

[CraigS]

So why the high capacitance if it is so undesirable in a mains lead?

 

It isn't if you use it between the mains and the component.  Using it between the power conditioner and the component gave unexpected results, and has been reported to Russ Andrews.  But they offer a no quibble 60 day return, so if there is a problem you just send it back for a 100% refund.

[Red MacKay]

OK, so back to digital cables:

 

Is one material better than another for conducting/ transmitting digital signals?

 

ie, pure silver is put forward as the best option by a seller here whom I respect, for all digital transmissions.  He does not recommend silver for analogue at all though.

 

Another audio buddy tried a diy gold/silver Mundorf wire with very poor results.

 

So what works best?  Are Os and Is any different to analogue signals?

[CraigS]

Fist off there are no 0's and 1's on a digital signal (such as SPDIF) - it is a very wide bandwidth analogue signal.  Generally from low tens of kHz to more than 100MHz (ideally).

 

Provided that the gear designers have done it right (but most haven't) and the send and receive ends are 75 ohms over that wide bandwidth, then all you need is decent quality 75 ohm video cable.  Someone I know with half million quid's worth of gear uses one core of a component video lead - after frigging around with all sorts of esoterica.

 

But if the 75 ohm proviso above is not the case (>90% of the gear out there), then all bets are off.  But there is no reason that metallurgy of the cable should matter two jots

[Duke40]

@@CraigS, I am currently using one core of a component video lead (though just on a humble headphone rig, not on a mega dollar rig).

 

The 75 Ohm video cable is a Blue Jeans Cable (from early 2000's, it is made using Canare cable and Canare RCA plugs both ends).

 

Source has 2 SPDIF outputs :             Digital RCA and Digital BNC sockets.

Destination only has 1 SPDIF input :   Digital RCA socket.

 

Currently connecting using above mentioned BJC digital cable with plugs on RCA for both source and destination.

 

Was thinking that I could try the following options;

1) get a BNC-BNC digital cable (probably an Oyaide DB-510), add an adaptor to change from BNC to RCA (for the destination)

2) get a BNC-RCA digital cable (probably from Blue Jeans Cable)

3) get another brand of RCA-RCA digital cable to try (probably an Oyaide DR-510). 

4) do nothing ... just keep using the old Blue Jeans Cable 75 ohm video cable which has RCA on both ends.

 

I am interested in your thoughts regarding Option 1 above, do you think adding a BNC/RCA adaptor to a BNC-BNC cable would cause issues,

or would Option 2 be better, as it would not require an adapter.

 

Also, if I go Option 2 (BNC-RCA cable), I only need a short length and could get by with 30cm (Blue Jeans Cable can make cables this short), would this be too short for a digital cable ?

Look forward to your thoughts.

Edited February 26, 2016 by Duke40

[CraigS]

I'm not a great fan of adaptors - it is just yet another potential impedance discontinuity.  That said I've pressed BNC/RCA adaptors into use from time to time - but just for quick trials and not permanent use.

 

Have a check what the BNC output looks like - 75 ohm and 50 ohm BNC's look quite different in detail.  This shows the difference quite well http://www.audioaficionado.org/mbl-dcs-goldmund-gryphon-etc/27643-clock-experiments-13.html .  Let me know what you find - because that might influence the way you go.

[Duke40]

Thanks Craig. I checked the link you provided ... it looks like the adapter that I was looking at getting ("High Quality BNC to RCA Adapter" from futureshop)

was 50 ohm (too much plastic inside), not 75 ohm.

 

I am also not a fan of adaptors ... so I may just forget about Option 1.

 

Thank you for your thoughts. I appreciate it.

[Red MacKay]

So this 75 ohm thing is mega important.  That seems to be the issue that is always in these conversations.

 

Does this mean that I am better to replace the unknown original rca (or BNC) plugs on my dac and transport with known 75 ohm units like those WBTs that you have mentioned previous?  I am not precious about originality and I hang onto my gear for quite a while, so that's not an issue for me.

 

And which do you feel serves the purpose best here for 75 ohm- RCAs or BNC?

 

Man, we have been hammering you the last couple of days Craig.  Thanks again for your time and thougts.

[CraigS]

I was really really surprised when those darned WBT RCA's outperformed (by measurement) every 75 ohm BNC I've tried.  But it is icing on the cake if the send and receive is not well implemented.  The basic problem is that the pulse transformers are 1:1 turns ratio, so what designers do is plonk a 75 ohm resistor either across the input, or more commonly across the output.  Problem is that that the transformers do not really like looking at 75 ohms.  With the Newava transformers I'm using on my gear, the receive end works with about 300 ohms on the primary and 100 ohms across the secondary (in parallel these equal 75 ohms - but that is not what the transformer sees).  The transmit end goes from the output logic gate via a 169 ohm resistor and 0.1uF capacitor (to block the average DC) to the primary.  The secondary has a 43.2 ohm resistor in parallel, and then a 39 ohm resistor from there to the output socket.  The combination of that lot looks like 75 ohms looking back into the output, and the output voltage is bang on the SPDIF requirement.

 

Of course all this needs very careful surgery.  This is what the mod it looks like at the SPDIF output from a Logitech Transporter.  The original transformer (surface mount) is removed, and the new one upside down in dead bug mode, with the resistors and capacitors wired in with absolute minimum length leads.  That produces a clean rise and fall time of 1.3ns, as per earlier photos.

 

The original circuit had a 330 ohm resistor from the logic gate output (a 74HC74 latch) to the primary and a 91 ohm resistor in parallel with the primary.  And nothing on the secondary at all.  The designer's logic was that 330//91 = 71.3 ohms, plus a few ohms output resistance of the gate gets it up to close to 75.  There are two problems - first is the subtle one of the transformer not seeing the impedances it should be seeing, and the second is glaring - no DC block.  So the transformer core has a whopping DC bias (the average level of the SPDIF signal).  I didn't photograph the signal; should have done, but it was truly depressing.

 

Note that the Transporter uses a 75 ohm BNC; that will be changed to a WBT next time I'm in there.  At the moment I'm just enjoying music!

 

Edited February 26, 2016 by CraigS

[Slartibartfast]

If you do the math, using RCA instead of BNC results in a return loss of approx 12dB at 150MHz.

 

Below 50MHz it's >20dB. 

 

 

So yes, the characteristic impedance of the cable is very important, but the impedance of the termination connectors are not important at all.   

[CraigS]

Don't agree.  You can clearly see the connector difference using fast TDR.  And since the rise time of the modified output is 1.3ns, that implies a 3dB bandwidth of 270MHz.

 

Besides which, if you can get it bang on right so easily, why get it wrong?

[Slartibartfast]

Don't agree.  You can clearly see the connector difference using fast TDR.  And since the rise time of the modified output is 1.3ns, that implies a 3dB bandwidth of 270MHz.

 

Besides which, if you can get it bang on right so easily, why get it wrong?

 

 

For SPDIF digital audio transmission bandwidths, it's all inconsequential.

[CraigS]

Well, lets just leave it just as it is then.

[Red MacKay]

The basic problem is that the pulse transformers are 1:1 turns ratio, so what designers do is plonk a 75 ohm resistor either across the input, or more commonly across the output.  Problem is that that the transformers do not really like looking at 75 ohms.  With the Newava transformers I'm using on my gear, the receive end works with about 300 ohms on the primary and 100 ohms across the secondary (in parallel these equal 75 ohms - but that is not what the transformer sees).  The transmit end goes from the output logic gate via a 169 ohm resistor and 0.1uF capacitor (to block the average DC) to the primary.  The secondary has a 43.2 ohm resistor in parallel, and then a 39 ohm resistor from there to the output socket.  The combination of that lot looks like 75 ohms looking back into the output, and the output voltage is bang on the SPDIF requirement.

 

Of course all this needs very careful surgery.  This is what the mod it looks like at the SPDIF output from a Logitech Transporter.  The original transformer (surface mount) is removed, and the new one upside down in dead bug mode, with the resistors and capacitors wired in with absolute minimum length leads.  That produces a clean rise and fall time of 1.3ns, as per earlier photos.

 

The original circuit had a 330 ohm resistor from the logic gate output (a 74HC74 latch) to the primary and a 91 ohm resistor in parallel with the primary.  And nothing on the secondary at all.  The designer's logic was that 330//91 = 71.3 ohms, plus a few ohms output resistance of the gate gets it up to close to 75.  There are two problems - first is the subtle one of the transformer not seeing the impedances it should be seeing, and the second is glaring - no DC block.  So the transformer core has a whopping DC bias (the average level of the SPDIF signal).  I didn't photograph the signal; should have done, but it was truly depressing.

 

 

Ahh, so that is what the transformer does in that position!

 

I performed the Lampizator Goldmundizer mod to my old Pioneer DVD player pulling the SPDIF signal straight from the chip to the RCA out - it was a backward move.  I lost all the lower midrange/upper bass presence.  So if I have this correct, what I did was screw up the impedance?  Lukas makes it all look so good too, but it just did not work in the real world for me.

 

I certainly don't have the skill to work out what is needed for my much loved old CEC 51Z, but this info is all good to know.

[CraigS]

Although I don't agree with *everything* that Scientific Conversion say (they are after all manufacturers of pulse transformers, so there is some spin), this is an AES paper on why pulse transformers are a good thing for SPDIF, and why wide bandwidth and clean rise and fall times are things to be striven for http://www.scientificonversion.com/AES1998.pdf 

Edited February 27, 2016 by CraigS

[Red MacKay]

That link does not work for me Craig.

[rocky500]

That link does not work for me Craig.

Goto this page then scroll down and click on the link.

http://www.scientificonversion.com/technical_paper_1.html

[Red MacKay]

Yep, that one is good, thanks.

[CraigS]

Will be away for a week, so unable to reply to anything until back.

[davewantsmoore]

So this 75 ohm thing is mega important.

 

Yes, when dealing with such fast electrical signals, the impedances terminating each end of the cable affect the signal greatly.

 

 

It was mentioned many many pages ago, but this is a reason against DIYing I2S connections, without the ability to correctly (investigate) and control the terminating impedances.