[randytsuch]

I am starting to look at the SPDIF interface between my DAC and squeezebox, to see how I can optimize it.

My knowledge of traveling waves is from one college course, which was a few years ago now , so I started to do some internet research, to try to remember this stuff.

I just read this article from Steve of Empirical, which helped some.
http://www.positive-feedback.com/Issue14/spdif.htm

Another article here
http://www.epanorama.net/documents/wiring/cable_impedance.html

So, impedance is an AC measurement, which is equivalent to resistance in a DC circuit.

Problem is impedance is not easy to measure, with test equipment normally available.
You really need a Time Domain Reflectometer (TDR).  This type of device sends a pulse down a cable, then measures the reflections from the pulse.  If you have a perfect cable, and perfect terminations on the sender and receiver side, there will be no reflections, but of course that is not really possible.

So, I was wondering how I could find a way to measure how good the terminations and cable are, in my SPDIF interface.

I found this page, with a circuit for making your own Time Domain Reflectometer (TDR)
http://www.tkk.fi/Misc/Electronics/circuits/tdr.html

There is also this post, from head fi
http://www5.head-fi.org/forums/showpost.php?p=813396&postcount=72

I have access to a storage scope, and the DIY TDR looks pretty easy to build.  But, I also have access to a pulse generator, which I can use instead, and then I just need a dual pot, as Kevin recommended.  

So, I just have to hook the pulse generator with the pots used to terminate the cable at both ends, and use the scope to look for reflections.  

Then, I could try different cables, cable lengths, digital transformers etc to minimize the amplitude of the reflections, and maximize the time from the original to the reflection.  I would also be able to tune the pots, to find the optimal resistor values for both ends of the cable.

I would want to do this into both my SB3 (transmitter) and DAC (receiver).

As an aside, I measured the output impedance (really the resistance) on the SB3 digital output with a meter, and it measures around 103 ohms.  I was expecting it to be 75 ohms.

One more thing about this.  In the head-fi post, they were saying you should terminate all cables, not just digital ones.  Since every cable has a characteristic impedance, if you measure that impedance, and then terminate your receiver with the corresponding resistor, than differences in IC’s cables become much less apparent.

So, does this sound reasonable?

Oh, and as with all my other projects, I am probably a couple months away from actually doing anything on it   .

Randy

[Dan Banquer]

Last I knew you are going to need a network analyzer. It's the unit of choice in RF for these types of measurements
 You should not have any problems with cable if you choose a reputable vendor such as Belden.
  Are we having fun yet?
                d.b.

[Carlman]

Cool!  You're doing more than (I think) cable manufacturers do for the 'digital cables' they sell.  Most 'digital cables' I've bought were just IC's with a different color band on them.  My guess is it's part of the reason they all sound a little different from each other.

I made some IC's using Belden 8422 and DHLabs mid-level RCA's (thanks for the recipe, Dan)... and those ended up sounding great as digital IC's.... not bad for analog IC's but not 'special' either. I feel I may have 'lucked up' and gotten something that has the right impedance.

I'll watch this thread with great enthusiasm.  I applaud your efforts to get it right.

-C

[randytsuch]

Last I knew you are going to need a network analyzer. It's the unit of choice in RF for these types of measurements
 You should not have any problems with cable if you choose a reputable vendor such as Belden.
  Are we having fun yet?
                d.b.

Hi Dan
I have not seen mention of a network analyzer for this kind of stuff.  Do you know what it is used for?  Can it measure the amount, and time delays of the reflection?  We actually have one at work, at least we used to, and I am pretty sure it is around somewhere.  It was an expensive piece of gear, but nobody uses it.

Carl,
I will update this thread as I have a chance to do things, but it will take some time.  I am not going to use RCA's for digital, you can't get 75 ohm RCA's.  I will use BNC's, by DAC already has a BNC, and I am going to get one for my SB too.  If you have anything in the signal path that is not 75 ohm, then it will create reflections, which will muck up the sound.

You are using 8422?  According to Beldens sight, that is 90ohm cable, I would use something like 1694 or 1695, which is 75ohm impedence controlled stuff.  I may buy some from these guys
http://www.bluejeanscable.com/store/digital-audio/index.htm
Their prices are cheap, and it is hard to find Belden is small quantiites.  Takefive sells 1695, but I think it will be about the same price from take5, and that is raw cable.

I wonder if the 8422 sounds good with RCAs because the RCA's are not 75ohm either.

Randy

[Dan Banquer]

The network analyzer is generally used for insertion loss or gain and reflected power, also known as return loss or VSWR. I don't think it has time domain capability, or at least I never have seen one with time domain capabillity.
  If you are doing 44.1 Khz sampling rate then the time domain issues should be at best minimal.  For cable I would recomend a double shielded 75 ohm coax, of which Belden has a number of.  An RG6 of this type should be total and utter overkill.
                    d.b.

[randytsuch]

The network analyzer is generally used for insertion loss or gain and reflected power, also known as return loss or VSWR. I don't think it has time domain capability, or at least I never have seen one with time domain capabillity.
  If you are doing 44.1 Khz sampling rate then the time domain issues should be at best minimal.  For cable I would recomend a double shielded 75 ohm coax, of which Belden has a number of.  An RG6 of this type should be total and utter overkill.
                    d.b.

Hi Dan
OK, so you can use the network analyzer to measure the reflected power, which is what I really want to know, how much of the signal is getting reflected back.

I was just at the Agilent sight, and it looks like they have a time domain option for their new, fancy network analyzers, but we have an older model.

I normally go for utter overkill, and the Belden 1694 and 95 are rg6 cables, the difference is the 1695 is teflon core, so it is more expensive.  I'll probably try the 1694 with bnc connectors, and see how they look and measure.

Randy

[Dan Banquer]

If you use 75 ohm BNC connectors and terminate the cables correctly than your cables should measure well.  For below 100 Mhz your return loss should be at least -20db or better.
                   d.b.

[randytsuch]

OK, have not had a chance to make any progress, can only really do this at work, and I have been too busy at work to do anything, but I did find this link

http://www.diyhifi.org/forums/viewtopic.php?t=429&postdays=0&postorder=asc&start=0&sid=10598578e769bd159e4cec40796aa9e3

It is kind of long, but there is some very usefull information in it.  

One other thread you guys might find interesting, this one on terminating IC's
http://www.diyaudio.com/forums/showthread.php?s=&threadid=9354&highlight=

Randy

[Jon L]

I've been struggling with this issue for a long time.  Basically, I have come to the conclusion that I need to bypass spdif completely out of the loop, which is not easy or cheap.

[slwiser]

What do you think of the theory behind the LessLoss cables?  

www.lessloss.com/cable.html
www.lessloss.com/cable_new.html

[randytsuch]

I've been struggling with this issue for a long time.  Basically, I have come to the conclusion that I need to bypass spdif completely out of the loop, which is not easy or cheap.

Jon,
I think you can implement SPDIF correctly if you can DIY.  There are no true 75 ohm RCA', unless you can change both the RCA jacks and plugs.  There are companies that make 75 ohm rca jacks, but they are not 75 ohm anymore once you plug them into a normal jack.  And, if you are going to change the RCA plugs, then you may as well install 75 ohm BNC connectors.
And, from what I have been reading, even if you use BNC's and 75 ohm coax RG6 or RG59 cable, companies don't always implement SPDIF outputs and inputs correctly, so you may still need to mod the units to really do it right.  Even the board traces should be impedance controlled for 75 ohms.  An easier option is Guido Tent makes a module that reclocks SPDIF, and outputs at 75 ohms.  I have not tried, but heard good things about it.

Slwiser,
I quickly read the lessloss page.  What they seems to make some sense, shielding is important.  What they fail to show there is some companies use foil shielding.  Belden uses foil shielding on some, but not all of their 75 ohm cables.  Foil will be less flexible, but it will provide better shielding than braid.  
I also have no idea what they are talking about "floating shields"?  Seems like marketing BS to me.
The 75 ohm charateristic impedence is correct, but they fail to mention that the RCA's are not 75 ohm, so even if the cable is, you will get reflections from the RCA's.
I would not spend $300 on thier cables, but it may sound nice.m
My current theory on commercial digital cables it they have to fight the RCA jack problem, and the cables differences you hear are based on what the cables do to handle the reflections from the RCA jacks.

One other thing I have seen is the 75 ohm cable is not necessarily 75 ohms.  There is some tolerance on the 75 ohms.  The better Belden stuff I have found is +/- 1.5 ohms, they also have some at +/- 3 ohms.  They don't always specify it, and I am not checking all the cable they make, just the ones I can buy.

Randy

[Dan Banquer]

Hi Randy;
     I would like to make a couple of comments here. The foil shield may well offer better shielding but the copper braid will give you a lower ground impedance and shielding somewhere in the 90% plus area, so I would go with a double shielded cable with a dual copper braid shield.
   Your most important observation however is about the output driver, and the input receiver. Consumer audio from what I have seen does a lousy job on output drivers. The designs seem really more of an after       thought than something reasonably well thought out. Input receivers should be easy, but you never know until you do a thorough examination.
    My general observation is that too much emphasis has been put on the cable and connectors when the output drivers are really the culprit. I suspect that they do the most damage. If you need a schematic for a an input receiver, let me know. If you need something for an output driver,  
give me some time and I will cook one up.
                     d.b.

[randytsuch]

Hi Randy;
     I would like to make a couple of comments here. The foil shield may well offer better shielding but the copper braid will give you a lower ground impedance and shielding somewhere in the 90% plus area, so I would go with a double shielded cable with a dual copper braid shield.
   Your most important observation however is about the output driver, and the input receiver. Consumer audio from what I have seen does a lousy job on output drivers. The designs seem really more of an after       t ...

Hi Dan,
I was thinking about buying some Belden 1694A, which has two shields, a foil one, and a copper braid:
1  Duofoil®  Tape  Aluminum Foil-Polyester Tape-Aluminum Foil  100%
2    Braid  TC - Tinned Copper  95%

As for an output driver, I am trying to figure one out for my Squeezebox.  The SB digital comes out of a FPGA, to a HCU04.  The HCU04 output goes through a cap, then a resistor, and out of an RCA.  

I have been doing some reseach at the diyaudio places.  I am planning to take the HCU04 output, use a cap and couple resistors to set the impedance to 75 ohms.  Then use a 1:1 transformer, and a h pad to set the voltage out to between 0.5 and 1.0 volts.

I need to figure out how to calculate the h pad values, so I can attenuate the output correctly.

Randy

[Jocko Homo]

Such a deal I have for you:

http://www.diyhifi.org/forums/viewtopic.php?t=429&highlight=

As for network analysers:

Some have time domain transfer functions built in, but no one here can afford one. TDR's are fast an easy for this sort of thing.

BNCs: Have to be 75 ohm.

RCAs: Don't waste your time. Yes, I have heard stories that magic RCAs exist that are 75 ohms. But as one person has pointed out, you would have to use both a male and female together. So if you have to go to all that trouble, why not just rip out the stinking RCAs and be done with them. The only thing worse is XLRs.

Return loss: >20 dB. When you start getting down to around -30 dB, things don't seem to makae as much difference.

Yes, most cable companies don't measure this crap. I know, because some send their cables to me for blessing, only to have me call them on the phone and say "Who the 'eff told you this cable is 75 ohms? They must be as stupid as you are. $200 for a 3' hunk of RG-58 with expensive RCAs on the end. Sell them as interconnects."

Yeah, I make friends easily.

As for output stages: Most are designed to minimise EMI, not performance. As a result, they sound like crap. Most of the ones on the 'Net are not much better. Jitter problems.

Anything else?

Jocko

[konut]

Yes. What commercially available brand(s) of 75ohm coaxial can you recommend? Of course with BNC termination.

[skrivis]

Hi Dan
I have not seen mention of a network analyzer for this kind of stuff.  Do you know what it is used for?  Can it measure the amount, and time delays of the reflection?  We actually have one at work, at least we used to, and I am pretty sure it is around somewhere.  It was an expensive piece of gear, but nobody uses it.

Carl,
I will update this thread as I have a chance to do things, but it will take some time.  I am not going to use RCA's for digital, you can't get 75 ohm RCA's.  I will use BNC's ...

How about the Canare RCAP? It's specified for 75 ohms.

Personally, I'd just buy a cable from Blue Jeans and be done with it.

http://www.bluejeanscable.com/store/digital-audio/index.htm

[Dan Banquer]

Hi Stuart;
    Many times we build something because we wish to learn from it. In this case it may not have much to do with the cable, but some of us will learn more about how a transmission line is supposed to work. The journey of DIY is not just the end result, it is also the learning process that goes with it.
              d.b.

[randytsuch]

Hi Stuart;
    Many times we build something because we wish to learn from it. In this case it may not have much to do with the cable, but some of us will learn more about how a transmission line is supposed to work. The journey of DIY is not just the end result, it is also the learning process that goes with it.
              d.b.

Hi Dan,
Yes, this is not just about finding the best digital cable, that is only part of it.  It was also to work out how to optimize the tx and rx ends of the circuit.  
The SPDIF is a system, and it will only be as good as it's weakest link, so I want to make sure that I make the tx, rx, connectors and cables correct to reduce reflections and jitter.

Jocko,
Hi, I did not expect to have you respond to this thread, didn’t think you came here very often.

I already linked to your thread on a DIY TDR.  I found, and read it last night.  At this point, the only thing I need is some good 75 ohm coax, and then I will give it a shot.  

Randy

[Jocko Homo]

Randy:

You are right, I don't come here often. When I do, it usually is because something on one forum spills over onto another one, and we would like to know in advance who is being a PITA.

OK, where was I........?

Cable:

I have found that lower loss ones tend to sound better. There is no hard and fast rule, because as you have discovered, it is a system. You have to take the whole system into account. And even then.......there still are no hard rules.

I once tried a $1000 microwave Gore-Tex cable from one of my network analysers. It had no bass, but was the cleanest, most  spacious sounding cable that I have tried. Others that I trust have given identical reports. Which leads me to believe that if I made a 50 ohm system for my personal use, it would wipe everything else away.

I have told this story before, but it is a good one, so here goes:

A famous audio designer who was (and may still be) clueless as how to make a 75 ohm output found that a 93 ohm cable sounded best. It only made sense as the output impedance was 95 ohms or so. So........he takes his player and cable over to a certain god-like reviewer, and much to their delight, they found that no other cable sounded good on that set-up. The reviewer concluded that the cable must be some present from the gods, and reported it as such in his publication.

Too bad it sounded like crap on every other system, but why let little thing like that get in the way of the pursuit of truth?

Yet another reason not to trust reviewers.

For your TDR......it does not have to be good, just the one that you want to use.

"And which one is that?"

Well.......one that does not have a copperweld center conductor. Foil shielding may not be the best choice. Foil is designed to be an effective screen at high frequencies, as the braid tends to act like a sieve at the higher frequencies. Not to say it won't work, but ones with  a good copper braid may have better manufacturing tolerances.

The cables I used to like have either been dropped, or they make you buy a 1000' roll. No thanks.

As for Canare:

Do you guys believe everything that you read in some ad? Have they ever provided some independent measurements to verify their claims? They make the claims: it is their duty to show it is right, not my job to show they are full of it.

Of course.........if they ever sent me any......they would be measured, and the results displayed everywhere.

Jocko

[AphileEarlyAdopter]

Folks,
This is a very good thread. I think differences between exorbitantly priced digital cables comes only from the varying impedence of the cables. If it so happens that the cable's impedence matches your output/input impendences, then that is your high-end cable. I have tried various digital cables with my Philips 963sa and Panasonic XR55 digital amp receiver. Right now, the $30 glass toslink sounds the best to me.