Recently an AK member (DKak) sent me a few digital audio cables to measure. First one I grabbed out of the box is a four ft. Canare L5CFB cable with BNC/RCA terminations. I have no idea about the cost of this cable, or the claims made by the manufacturer. Attached is a picture of the measurements. The cable came supplied with a directional arrow on it, and I respected this for the measurement.

RESULTS
I arbitrarily placed the measurement markers at 2ns and 7ns down the length of the wire. I also opted to normalize all of the measuremts for a risetime of 500ps. I'm not sure what would be a typical risetime in the SPDIF world, so I chose this one. This cable measures to be right at exactly 75ohms. I cannot explain the small rise at 7ns, the cable looked fine from the outside with no physical damage. The rise at the very end of the line looks like a compensation network is included (like a series inductor) for peaking purposes, but without reading the manufacturer's claims, I don't really know. This is also the end that had the RCA connector. This also appears to be perhaps a true 75ohm RCA connector by the looks of it. The best part is the risetime at the end of the cable, measured at 677ps. Of the four cables I've measured so far, this is the fastest.

SOUND
I only measure the cable parameters, I'll leave it to DKak to offer any comments on the sound of this cable.

A Canare cable that actually specs as advertised? I am not shocked at all. Canare is excellent across the board.

Thanks,
Ron-C

That cable would be a very cheap cable to build, if you had the crimper!

That is a high quality coax, TC braid with full foil under-shield, and a solid OFC center conductor. Their RCAP terminations, which that surely is, if its a Canare branded connector, are well rated as well.

The cable is less than 25c a foot, and the connectors are about 2.50$ each. Unfortunately, the crimping tool is 70$, and the crimping die is about 60$. But if you are making a bunch, that is the way to go.

I was going to send you a cable to test, House de Kris, but the cable you just tested is the same as the one I was gonna send you! How convenient for me. :D

I've always respected Canare gear, they deal with the pros who aren't easily hoodwinked, and so their specs are usually very close to what they claim.

peace,
sam

Canare Tech support. :thmbsp: :thmbsp:

House De Kris asked me to comment as he tested the cables I sent. First, I'm very interested in the results he gets -- there's quite a cable variety, 10 in all.

Second, no, this cable has no networks or anything, just L5CFB (an RG-6 type), about 4 feet in length with a Canare 75-ohm crimp BNC on one end and the 75-ohm crimp RCA (RCAP) on the other. I have no idea what caused the small rise or end peaking. Detailed tech info on Canare's stuff is easy to find at www.canare.com

All Canare cables Kris will test for us were assembled by HAVE Inc. or Markertek, both in New York. HAVE put their brand on some heat shrink on the cables they made, Markertek did not. Maybe Kris can note which are which, for whatever reason.

As far as sound quality, I'll make a few comments, but this isn't the place to argue whether cables matter. There are plenty of threads here where that argument can continue. However, I believe I hear cable differences. With that, all comments are mine only, and within my system only. My curiosity in this stems from Kris being able (and willing) to measure cables and me wanting to know if the measurements reflect at all what I hear. Not scientific, not ABX, not this, not that. I know. This is just a "thought experiment" for me. Kris and I thought the measurement results would be interesting to the AK community. Read/interpret/deal with them as you'd like.

Of the digital wires I've tried, the L5CFB is very nice. Open, detailed, etc. I've never noticed any different sound with different lengths of Canare, and I have cables made from L5CFB, L7CFB and LV-61S. I've never used one over 6 feet long, though. Incredible for video, too. Use three of 'em for component connections.

Of the ten I sent Kris, my favorite so far for sound is the Belden 1694A with Canare BNC or RCA (all 75-ohm) connectors. The cables of this type that I have were all made up by www.bluejeanscable.com There will be plenty of other cables I send to Kris that are far more expensive, but these just sound "right" in my system. The Canare aren't far behind for me, but I think the PS Audio Digital XStream is my second favorite. Truly will be interesting to see how they all measure. This should be quite an informative set of posts from Kris.

I really appreciate his technical explanations of the measured results, they shed some light on this topic for all of us. Thanks a bunch, House De Kris!

DKak

Thanks to DKak for adding a comment, and thanks to him again for sending me a box of many cables. Just to reiterate, since the effects of a digital cable are entirely dependent on the match to the other equipment used, DKak's comments on sound are valid for his particular system. But, I did think it would be interesting to hear his opinion on various cables in his system for comparison purposes. This particular cable is marked as being made by Have Inc.

Although I would have liked to have measured more of his collection by now, I ran into a small snag. Seems I've destroyed four TDR plugins so far and I only have one left. Since I cannot afford to risk losing my last one, I've put TDR measurements on hold until the others get repaired. I've also decided to measure these on a network analyser to get frequency domain measurements like the s-parameters and dielectric constant. But, that piece of gear is being hogged by another engineer right now.

More later.

To flesh out my comment that I believe I hear cable differences...

I think I understand why I can hear them in analog cables -- I estimate that different capacitance, resistance and shielding are the main causes. Not sure how each relates to what I hear, though.

However, in digital cables it seems to me that if you get the right bits from one end to the other intact and at the right time then pretty much all should be equal and cables would sound virtually (if not completely) identical.

So why do I hear differences when all of these are advertised as real 75 ohm digital cables? Maybe not all are really 75 ohm? Maybe the risetime is poor? Maybe the bandwidth isn't sufficient? Maybe the connectors screw up the transmission line? Maybe the precision of the connectors' connection attained during assembly will make a difference? There are probably more things to look at than these.

I'm anxious to see some measurements to these specific cables (all considered 'digital') where I hear differences but I'm not sure I should. All are claimed to be very good at what they do by the respective manufacturers. And so far, my listening tells me the inexpensive Belden or Canare offerings are right up there at or near the best in my system.

I hear only very slight differences between the different Canare and Belden wires. All are sweet, smooth and great sounding to me. My system is highly revealing, and while I'm glad it lets me hear digital cable differences it troubles me on a whole 'nother level that I do.

I was able to test this cable on the network analyzer. This plot has both the amplitude and phase response plotted. Ch3 is the phase, and I'll say right now just go ahead and ignore it. I thought I would be fancy and show the phase response, but I didn't get it calibrated properly and thus is useless (other than showing general shape, the tilt is completely wrong).

The important thing to note is that the -3dB point is 824MHz. This is sorta interesting when compared to other flavors of this particular Canare cable that will be reported on in a few posts later in this very thread. I'll try to tie this all together in a few posts.

DKak also loaned be another cable made up of Canare L5CFB. This time a 6ft length terminated with RCA connectors on both ends. The RCAs appear to be 75ohm. This cable was assembled by Have Inc. I only have the network analyzer view of this cable. Turns out I blew up yet another TDR module trying to test this cable. This frequency response plot also has an attempt at getting a phase plot. Like the cable above, just ignore the phase trace. It is not showing anything relative at all, not even the shape is correct. The results of this measurement is a -3dB point of about 510MHz.

Luckily, this time I figured out the problem I've been having with measuring digital audio cables on my TDR. Turns out it's those darn RCA connectos. I'll summarize all this after introducing the next cable for consideration.

The final flavor of digital cable made with Canare L5CFB on loan from DKak. This cable is 3ft long and terminated with BNCs on both ends. Attached are the TDR and network analyzer pictures. The TDR trace (yellow) is increadibly flat at 74.34ohms throughout the entire length. This curve is 25ohms/div with 50ohms in the center. As with all of my recent TDR pictures, the graticle value is very low, thus hard to see in the picture.

Like other network analyzer pictures in this series, ignore the phase trace on Ch3 - curve is correct, tilt is wrong. Noteworthy from this picture is the 3dB point of 8.73GHz. Yowzer!

Having three lengths of the same cable turned out to be rather interesting to me when comparing data. The frequency response as measured by the network analyzer for all three is:

3ft - 8.73GHz
4ft - 824MHz
6ft - 510MHz

Why, oh why, one may ask, do we lose 90% of the bandwidth when the cable gets only 33% longer? Plus, the bandwidth hit is less than an octave when increasing the length by 50% from 4ft to 6ft. If we include termination data in that little chart above we get:

3ft - 8.73GHz - BNC/BNC
4ft - 824MHz - BNC/RCA
6ft - 510MHz - RCA/RCA

Therein lies the true contributors to erroneous measurements. In another TDR measurement thread, I mentioned that my BNC-to-RCA adapter is 28ohms. In the time domain, this isn't too bad in that you can still see the cable imedance past such a discontinuity created by the adapter. But, due to the bandwidth limitation imposed by these adapters, the ability to accurately see the cable at a futher distance rapidly degrades. TDT risetime measurements would also be highly suspect if they have to travel through one of these adapters as well.

Now that I know that my adapters probably have a bandwidth of less than 1GHz, pretty much all network analyzer pictures of digital audio cables terminated in at least one RCA connector done by me don't represent the true cable. Sorry about that folks. I'll search for some wide bandwidth controlled impedance RCA adapters, and redo some of these tests if I can find better adapters.

My disdain for these RCA adapters doesn't stop there. As mentioned above, it was this whole RCA connector thing that lead to blowing up a total of five TDR modules. The equipment I'm using for these measurements very fast fairly hgh quality gear, and these TDR modules double as regular verticle plugin modules in a scope with a bandwith of 24GHz. Fast scopes like this typically don't accept very much voltage, and the max input voltage of my scope is 2V - above that it breaks. Coming off the front panel of the TDR modules are 3.5mm jacks which are a high speed variant of SMA. To this I attach an SMA-to-BNC controlled impedance adapter, and onto that I attach the garbage BNC-to-RCA adapters. With RCAs now sticking out the front of the scope I plug in the digital cables. ZAP! That's the problem. Whoever the clown was who designed RCA jacks to begin with had the brilliant idea to make it so that the center conductor mates before the shield does. BNC, SMA, SMB, SMC, 3.5mm, 2.5mm, SMP, N-type all make the shield connection first, then the center conductor. This allows any static charge on the cable to be drained to ground first. With RCA, any static charge on the teflon dielectric of the cable gets dumped directly into the scope input. Well, live and learn. In the future, before I plug in a cable I will have to put a shorting adapter on one end, plug in the open end, remove the short and plug in that end.

I've attached a couple pictures of these adapters. First one shows the BNC(M)-to-RCA(F). As you can see looking down the barrel, the center conductor sleave is only about a millimeter or so past the entry way hole. Also note that the shield shroud comes very close to the center conductor. This is where the very low impedance of this adapter comes from. The other picture shows this same adapter with the opposite gender partner right next to it (BNC(F)-to-RCA(M)). Looking at the male RCA, it is obvious that this one too will have a very low impedance with the shield metal approaching the center conductor very closely. I didn't use this adapter for any of these measurements, just thought I'd share it.

Now that I know that my adapters probably have a bandwidth of less than 1GHz, pretty much all network analyzer pictures of digital audio cables terminated in at least one RCA connector done by me don't represent the true cable. Sorry about that folks. I'll search for some wide bandwidth controlled impedance RCA adapters, and redo some of these tests if I can find better adapters.

Shortly after posting this, I received a private email with a link to some Monster RCA-BNC adapters that claimed to be 75ohms. The picture in the link appeared to be different than my cheapie adapters, so it would seem possible that they would be better than what I already have. So, I set out at lunch last Friday to find better adapters. Since the Monsters were sold in sets of three, this got me to thinking I should be able to find this stuff at any video shop.

Three stores later I was able to locate some Audioquest RCA-BNC adapters. The box didn't make any mention of it being 75ohms or not (unlike the F-adapter right next to it), so I asked the salesman. The conversation was comical enough to warrent inclusion here:

HdK: Is this adapter 75ohms?
SalesDroid(SD): Yeah, it'll pass a 75ohm signal.
HdK: Are you saying, then, that this adapter is 75ohms?
SD: We use them here on the demo floor, looks good.
HdK: All that aside, what's the impedance of this adapter?
SD: What are you trying to do?
HdK: I'm trying to determine if this is a true 75ohm path through this adapter.
SD: It'll pass a 75ohm signal.
HdK: Does the manufacterer claim this is a 75ohm adapter?
SD: I dunno, but we use them here and they look good.

I kept pushing him for infomation he didn't have, so finally he offers to consult the AQ website. No additional info there. So he calls up the AQ hot line. There is little need to go into a blow-by-blow of the phone call with AQ, but it became evident that the salesdroid learned most of his witty responses from AQ. The AQ rep is being just as evasive as the salesdroid, finally he offers to make a cable for me and terminate it however I want. I had to reiterate that I didn't want a cable, all I wanted was to know what AQ claimed for the characteristic impedance of this adapter. I finally just bought the thing, I could have measured it in the amount of time I wasted there at that store.

Opened the box and was pretty sure I shouldn't have wasted my time. It looks exactly like a gold colored version of the adapter in the previous post. Measured pretty much the same as well. I've included pictures for fun. In AudioQuest_BNCRCA.GIF, the yellow trace is the Audioquest adapter. The two blue traces are in memory and show the adapter path for measurement (explained later). The middle of the screen is 50ohms, and it is pretty evident that this is NOT a 75ohm path inside that adapter. Just for additional fun, I put a comparison picture (AQ-CheapCompare.GIF) between this and the cheapie adapter I had used. In this case, the AQ is blue and the cheapie is yellow. A little different, but not significantly.

Today I searched more for adapters that claim to be 75ohms. Finally found a set of five Monsters. At least they didn't appear (physically) to be photocopies of my cheapie adapters. The shield ring didn't get as close to the center as the AQ or cheapie. Plus, the BNC end looked more like a 75ohm connector than the AQ. This has to do with the amount of teflon in the area between center and shield. Put them on the TDR too and have included a picture of it here called MonsterBNCRCA.GIF. In spite of the claims printed on the packaging "...featuring precision 75-ohm impedance..." this adapter is nowhere close to right. I also picked up some precision 75ohm Monster male RCA to female BNC and added that to the stack and took another picture, MonsterBNCRCABNC.GIF (which is three adapters in a row).


Here's a quick guide for reading these TDR pictures. The first wave I put into memory (blue trace) is with nothing attached to the front of the scope whatsoever. At two divisions left of center, the transmission line (50ohm) ends and the impedance rises to the impedance of air (377ohm). With the addition of an SMA-BNC adapter, I put the next waveform into memory. The pathway is now about 250ps longer and will rise up to air impedance closer to the middle of the screen. Notice how this adapter is able to maintain a nice 50ohm environment from the front of the scope to the end of the adapter, this is a quality adapter. Finally, the live trace (yellow) dips way down, then finally drunkenly heads towards the impedance of air. At least with the Monster adapter, you can see that the impedance rises slightly before it takes a huge dive. This is because they actually attempted to make a 75ohm environment in the BNC section of the adapter. But, the majority of the signal path is in the RCA section and suffers accordingly.

My quest for quality BNC-RCA adapters continues. Not only that, but I think I'll drop Monster a line and see what they say.

I've received a number of emails about how to interpret the TDR pictures. I'll try to do a little step-by-step explanation about the adapters mentioned above, and hopefully this will shed some light on the subject. I've attached three pictures that should help.

http://audiokarma.org/forums/attachment.php?attachmentid=14396&d=1146183875
The first picture is just the TDR instrument itself. The internal pulse generator makes an edge, and this edge must travel through a pathway to the front panel. In the schematic, the pathway is represented by the transmission line component. Sorry, but I just realized I forgot to put the series terminator after the generator before the transmission line. Oh well, assume the generator has a 50ohm output impedance. The scope trace starts out two divisions below center, then is at center for about 400ps, then rises to two division above the center line. In this case, the center line represents 50ohms with area above it being greater than 50ohms and the area below center being less than 50ohms. Going back the left side of the scope display, the trace (two divisions below center) is showing 0ohms and is the source impedance of the pulse generator. This would represent being right at the output transistors of the pulse generator. When the trace shoots up to the center, that is when the signal goes through the series terminator and travels down the physical path to the front panel connector. If we were to know the construction (semi-rigid), then we would know the propagation time in this line (~140ps/in), then we could calculate the distance traveled. 450ps / 140ps/in = 3.2inches. At the end of the line, since it is open, the trace then rises to the impedance of air. Note, in this case, the resistance is displayed non-linearly.

http://audiokarma.org/forums/attachment.php?attachmentid=14397&d=1146183875
Next, I've added a 50ohm SMA-BNC adapter. All this does is make the pathway slightly longer. Schematically, I added another section of transmission line. Now, the scope trace stays in the center of the screen for a little longer time before heading up to the impedance of air. I kept the trace of nothing attached to the front panel in memory (blue trace) for reference. I also offset its trace for clarity. The difference in time from where the blue trace rises from the 50ohm line to where the yellow traces rises from the 50ohm line is the amount of time required to travel through this adapter. Looks like a little more than 250ps.

http://audiokarma.org/forums/attachment.php?attachmentid=14398&d=1146183875
Finally, a Monster "precision 75ohm" BNC-RCA adapter is placed onto the previous adapter. Again, schematically I've represented this with another section of transmission line. Likewise, I've kept the previous two traces in scope memory and offset them. You can see the trace at the very beginning of this final adapter starts to rise up since the BNC section of this adapter does appear to be made as a 75ohm environment. But, it shoots down quickly to a very low impedance for the length of the path in the RCA section. Well, I don't need to rail on this BNC-RCA adapter any futher, the intent of all this is just to explain the measurement method.

Hope this was clear and/or helpful.