Intro & Measurement Articles

[nwavguy]

It has been suggested by a fellow head-fier (nick_charles) that I say hello in this forum and might find a more appreciative audience here for some of the articles on my new blog. I'm an electrical engineer by education, and profession, and have worked in the audio industry in various capacities. I just, for example, finished a detailed review of the NuForce uDAC-2 that included some interesting drama and results I'm sure some NuForce owners may not be thrilled to know.
 
I have the equipment to conduct audio measurements that go far beyond the usual RMAA results found on the web. And I wrote an article about the limitations of RMAA and why most of the RMAA results on the web are difficult to compare with other results.
 
And, no promises, but I hope to also be posting various sorts of sound files that can be downloaded and compared using the free Foobar ABX comparator so others can get in on the more subjective/psychological side of the fun. One idea is to post some files generated in very different ways with generic names and have a sort of "brown paper bag wine tasting" to see what sort of differences are detected by various listeners.
 
I know on Hydrogen Audio there's been some interesting differences in what some can readily hear and others can't. It might be a lot like how the "rainbow artifacts" of single chip DLP video projection drive some people crazy and others can't see them even when they're told what to look for. Some of us, genetically or otherwise, might be more sensitive to certain audible differences. So just because I can't hear a difference, for example, I don't automatically assume nobody else will either. I think you get more valid results when you share the sound clips with a wider audience.
 
I've also used Audio DIffMaker in some interesting ways and might be posting some results from that in the future as well. As many may already know, it's a really good tool for some things, and a not-so-good tool for others.
 
And, finally, I welcome input on the blog as it is now, directions people might like to see it go, corrections to mistakes I've made, article suggestions, etc?
 


 

 

[nick_charles]

Welcome to the back of the bus !
 
I am quite interested in the measurable and audible differences between analog RCA audio cables. I have done my own crude tests of the frequency response differences between cables using musical signals rather than single tones, 
 
I found them to be very small. 
 
My measurments were a bit hamstrung by the relatively ordinary quality or the ADCs (Edirol UA-1EX and Behringer UCA202) I used to capture the analog signals, since apparently a night and day difference between silver and copper cables does not somehow survive a 16-bit digitization process,
 
 
Could your kit do a better job on this kind of test and in any case to what extent will the quantization error of an ADC really mask underlying differences ?

 

[holland]

Welcome aboard.  I will need to read your blog when I have more time, it looks very interesting at a cursory glance.

 

[nwavguy]

Quote:

Could your kit do a better job on this kind of test and in any case to what extent will the quantization error of an ADC really mask underlying differences ?

 
On the issue of quantization error, I've met and talked with John Siau--the designer (and company founder) of the Benchmark DAC1 and ADC1. John's a really interesting guy in that he designs products lots of very esoteric sonic-brick-loving audiophiles buy and love, but he himself is much more an objectivist and analytical engineer. His products that are so loved by esoteric audiophiles are also, interestingly, some of the best measuring examples of their kind I'm aware of. And he provides many pages of detailed Audio Precision results in the owner's manuals.
 
There are not many products that fit the description of "best measurements" AND "esoteric audiophile approved". Usually it's one or the other as many high-end companies compromise measured performance in their pursuit of better sound--some to an extreme degree like single ended triode amps. So Benchmark is on rare ground to have a foot solidly in both camps.
 
John told me about incorporating a Benchmark ADC1 and DAC1 pair into systems for various listening tests including blind tests and nobody being able to tell they were there. Of course the founder of Benchmark can be expected to say such things.
 
So I've done my own blind tests with my Benchmark ADC/DAC pair with my ears and hardware and that of others. I've even been so evil as to toss the pair into a vinyl lover's high-end pure analog system behind the equipment rack. And he didn't even notice it was there. And he says he really dislikes anything digital (and I might add, even after that sly experiment, still dislikes digital). I personally believe the pair is very transparent as long as the digital levels are reasonably well chosen.
 
The only statistically valid tests I've run are with Foobar's ABX comparator. But it, obviously, won't let you compare analog to digital. And the analog vs digital tests I have run (like with the vinyl lover above) wouldn't stand up to serious scrutiny. But I think they at least provide some clue the proper ADC and DAC can be transparent (i.e. not mask anything) even in a very high-end system.
 
And how many people really still listen to mostly pure analog mastered vinyl or tape over an entirely analog signal path? Once you insert anything digital into the chain, there's quantization error already. So wouldn't that already mask some or all of what the cable fans are worried about?
 
I think a great way to test cables is with Audio DiffMaker. Bill Waslo is a really smart guy, and he's written an engineering paper on digital audio differencing some may have read:
 
http://www.libinst.com/AES%20Audio%20Differencing%20Paper.pdf
 
He talks about the various sources of masking and error in the process and makes a rather good argument the test is still extremely revealing of even the smallest differences well below the threshold of audibility. This is mainly because it's a relative test not an absolute one. In my experience, if anything, his test is too revealing. It's too sensitive to all sorts of things you often don't care about. So when you get a really deep null (virtually nothing in the difference result) you can be reasonably certain there's certainly nothing audibly different between the two sources.
 
Audio differencing can also be done (under some circumstances) in the analog domain. This creates a different set of issues to be aware of compared to the digital version, but still has its place and can be very useful.
 
For those who want to dispute the validity of audio differencing, I simply ask: Where's the peer reviewed engineering or scientific research that supports their view? And so far, not one person has provided any credible reference. Many on Hydrogen Audio are really good about either ignoring those who can't prove their arguments, or running them off to more subjective pastures. It's a shame there's not more of that on other forums--including this one.

 
 

 

[Prog Rock Man]

Subscribed.

 

[inarc]

Very impressive! Subscribed.

 

[Rdr. Seraphim]

Nothing like diving into the deep end! Welcome aboard... 

 

[Roller]

Great read all around. Factual evidence holds its own without usual opinion coloring. Looking forward to read future articles.

 

[Uncle Erik]

Welcome and I look forward to reading more!

So... is it OK for an objectivist to also love 2A3 tubes and old records?

 

[khaos974]

Since the measurements of the uDac 2 are talked about in this thread, I'd like to add my contribution with those
http://www.head-fi.org/forum/thread/536324/nuforce-udac-2-rmaa-measurements
 
I am well aware of the limitations of RMAA but my own measures seem to be confirmed by the OP's more precise ones.
 
And to go with the problems noticed by the OP.
 
- Channel imbalance issue, yes I have a 8-10 dB imbalance just before 9 o'clock (on 2 different units)
- 0 dBFS clipping, didn't notice, the high gain makes me reduce the volume inside within Windows
- To high an output impedance, probably explains why some instruments are clearer with the uDac 2
- Too much volume gain, as said previously, it make me reduce the volume in Windows to avoid the channel imbalance zone.
 
By the way, the first unit shows a ~0.4 dB imbalance between the 2 channels, the second is matched within ~.1 dB  (with the exception of the 9 o'clock setting).

 

[nwavguy]

Quote:

I am well aware of the limitations of RMAA but my own measures seem to be confirmed by the OP's more precise ones.
 
And to go with the problems noticed by the OP.
 
- Channel imbalance issue, yes I have a 8-10 dB imbalance just before 9 o'clock (on 2 different units)
- 0 dBFS clipping, didn't notice, the high gain makes me reduce the volume inside within Windows
- To high an output impedance, probably explains why some instruments are clearer with the uDac 2
- Too much volume gain, as said previously, it make me reduce the volume in Windows to avoid the channel imbalance zone.

Thanks for joining in. I was hoping NuForce had solved the balance problems when I bought my uDAC-2, but obviously, that was just wishful thinking. They've certainly had enough feedback dating back to the introduction of the first uDAC. But, honestly, my impression is they think the product is just fine for 95% of their customers and anyone who complains about massive channel imbalance is some sort of anomaly, some who's being too picky, or someone expecting too much from a $129 product (i.e. see their "recording engineer" quote in my review). I just can't agree with any of that when a $29 Behringer can be drastically better and the problem is so plainly audible.
 
If the volume potentiometer they're using is not individually tested, what you get will be very random and some are likely much better than others. I know some volume pots are literally graded (by machine) when they're made--much like Intel and memory makers grade CPU's and chips for their speed capability. And they charge a lot more for the ones with good tracking and a lot less for the ones that measure worst. If NuForce is using a tested, but low grade, pot then they will ALL be poor but still vary.
 
And yeah, I have to wonder if the high output impedance was on purpose to make it sound "different"--much like HiFiMan tweaks the frequency response of their portable players to make them sound obviously different. I'm more OK with intentionally tweaked frequency response as at least it will be what the manufacture intended with any headphone if the device has a sufficiently low output impedance.
 
But if NuForce is raising the output impedance to tweak the sound, I think that's a bad approach as it will be very different with different headphones. I can say my SuperFi's, sound worse on the NuForce with the resulting boosted upper midrange and "notched" treble.
 
I can imagine someone psychologically predisposed to liking the NuForce might think the boosted midrange gives it more "presence" or, as you said, makes certain instruments clearer.
 
The harder I look at Leckerton the more I think that's what I should have bought. For only $40 more you get a lot more overall including a (claimed) 0.1 dB tracking at ALL volume levels. My recent post here has a couple of links including one to some dScope measurements:
 
http://www.head-fi.org/forum/thread/542281/nuforce-udac-2-drama-detailed-measurements#post_7314605
 

 

[Prog Rock Man]

Great blog. This article on jitter http://nwavguy.blogspot.com/2011/02/jitter-does-it-matter.html has a description of a AES listening test, but you make no mention of the actual levels of jitter were audible. What was the audible level of jitter in that test?

 

[xnor]

Interesting, will take a look at your blog.

 

[kite7]

Very interested blog, bookmarked.
 
I would like to see you debunk cables. nick_charles already has done quite a bit but more measurements wouldn't hurt.

 

[xnor]

Better spend the money on some yet unmeasured, reasonable priced audio interfaces / dac+amps.