Stealth DC-1: fatigue at 44/16, goes away at 176/16

[mattgarman]

Hi,

I've been listening to my Stealth DC-1 DAC for a few months now. It sounds great and is definitely the best DAC I've ever used... Except that I realized it causes listening fatigue fairly quickly. My music collection is almost entirely Redbook CDs, i.e. 44.1 khz, 16 bit. Source is a PC using Windows 7, Foobar2000, WASAPI (event) via USB (also tried optical), into DC-1. Amp is mostly DIY based on the Sure tpa3110 amp board, into DIY desktop speakers built around Mark Audio Alpair 7.3 speakers.

I used this same setup for about a year with a Cambridge Audio DACMagic (the original) prior to the DC-1. I never had any listening fatigue.

Just for kicks, I started experimenting with upsampling of my music. I ran some tracks through Secret Rabbit Code's libsamplerate, upping the sample rate by 4x to 176.4 khz. I can't hear any audible differences in the sound of the music, but the upsampled music does not cause any fatigue!

My very cursory understanding of sigma delta DACs like the DC-1 is that they employ some kind of filter to remove the artifacts created in the actual digital to analog conversion process. I also understand that designing the filter is a game of managed compromises, particularly with how close 44.1 khz digital is to the audible range. So perhaps the fatigue is caused by the filter, and the filter's dynamic changes with the higher sample rate? Just a random idea.

Just wondering if anyone might be able to comment on this phenomenon that I've observed. The upsampling experiment was more to see if I could perceive any improvement in sound quality. Since I don't, I'd rather not have to use it as a workaround.

Thanks for any thoughts on the matter.

[dac0964]

I've had my DC-1 for more than a year now. In my experience I only have listening fatigue when I listen to rock music (I listen to various genre) for about 2 hours in somewhat loud volume (-25 dB in the DC-1). Otherwise, I have no problem.

[thorcorps]

Hard to speculate on why you're experiencing fatigue in your setup. I can listen for hours and hours and all of my music is ripped at 44.1/16 (FLAC & WMA Lossless). Rock, Metal, Slow Jazz, Blues, Southern Rock, etc. I also used an original DacMagic prior to switching to the DC-1. I run optical from a Squeezebox receiver.

Have you tried Synchronous mode by chance, instead of the default Asynchronous? When I got my first DC-1 (when they first came out), I experimented with both briefly and a couple of times hence, and left it in synchronous (I know most folks on the forum listened to Keith and Lonnie's explanation and use Asynchronous, assuming it is better). It's always possible you'll prefer synchronous like I do.

[mattgarman]

Just to add more detail, I usually listen at moderate levels, 60-75 dB depending on which free sound meter app I use on my phone. I listen to a variety of musical styles, classic rock, blues, electronic, classical, etc.

I think the answer is "no", but just to be sure, is there any kind of switchable filters on the DC-1? Some DACs allow the filter to be configured, that might alleviate the issue.

I did a bit of general Internet searching, no one else seems to be having this issue with the DC-1. So that suggests I either have a dud unit, or it's a component matching/synergy (or rather, lack of synergy) issue.

To anyone else out there with the DC-1 and time, you might try feeding your CD-sourced 44/16 material in upsampled form. This is easy to do with (for example) the SoX plugin for Foobar2000. I'm curious if someone with better ears can perceive a difference. It's weird, I don't hear any obvious difference in the music's presentation, but I was doing some listening this weekend, and the fatigue effect is definitely there at 44/16. It has me constantly wanting to the turn the music down, even at my already low levels. But I get none of that with the upsampled stuff.

[garbulky]

Ok so this may sound a bit strange but....everytime you change equipment....you want to change your speaker position to fine tune it. I think this may solve a lot of your issues. (I know that doesn't explain why 192 sounds better) but I think it will help.
As for me.....I've noticed somethings that may or may not be in my head. And I really mean that I may be imagining it. Anyway on WASAPI mode regular 16 bit 44.1 khz was fine. But I got the impression that when I changed the output to 24 bit mode (but kept it still at 44.1Khz) that it made a difference. Essentially what I gather is 24 bit mode simply pads it with zeros but...maybe it makes a difference as to the mode that the DC-1 processes the sound with. As for 192 khz mode, I tried it and didn't think it offered a benefit.

Anyway, try positioning your equipment. I reccomend posting a picture of your view of your speakers taken from the area of your couch - preferrably with your couch included. A lot of times, we can offer suggestions that cost pretty much nothing or very little to implement and the changes can be almost like buying new equipment.

With poor positioning, certain DACs that have a different sound siognature may indeed sound less 'harsh' etc. But correct the positioning and that can change. Also something I found to be a real benefit with the Dc-1 is one of these behind your couch at head level. In terms of value for money this was pretty much right at the top.
www.atsacoustics.com/item--ATS-Acoustic-Panel-24-x-48-x-2--1001.html
Now...more will benefit you but just that one made a very sweet difference

[moovtune]

I prefer a more straight path from CD to DAC rather than ripping, converting, using computer drives, software etc. There are lots of variables that could effect the playback quality. My preference is to hook a CD player through coaxial to the DC-1 and then analog outs to my preamp or headphone amp. No software involved, no conversion software (FLAC or whatever) no software players (itunes, Foobar, Media Monkey) etc. - and I don't get listener fatigue myself but really don't know if the hookup has anything to do with it.

[mattgarman]

I could create a schematic of how things are arranged, but this is a simple desktop/nearfield setup, so I think I can just verbally describe it. Desk dimensions are typical: 60" wide by 30" deep. Speaker dimensions are: 10.5" deep by 7" wide by 14.5" tall. I have them sitting on yoga blocks (poor man's speaker stands) to raise the driver up to be approximately at my ear's level. The speakers are toed-in a bit. From a bird's eye perspective, it creates a triangle, with my head at one point, and the speakers at the other two points. This makes the speakers roughly 3 feet from my head.

As for direct-connecting a CD player to the DAC... What's a CD player? I'm semi-kidding, but I can't remember the last time I owned an actual CD player. I don't think I could ever give up the convenience of having all my music available with a few mouse clicks.

[garbulky]

Thank you for the description but an actual photo helps us talk about toe in, distance from walls, obstructions. I reccomend a photo, it's easier to give better advice on room treatments.

[moovtune]

Have you tried toeing them out more, so they're not directly playing into your head - especially if they're only three feet away?

[celing]

The only time I experience listening fatigue with the DC-1 is when playing poorly mastered music... Crap in-Crap out

[mattgarman]

I spent some time with different positioning, and (1) it doesn't affect the fatigue effect, and (2) anyway, I believe my original position sounds best.

I really believe this is an electronic issue, as the effect goes away when feeding the DAC higher samplerate content. I was just hoping someone might have some deeper insight into this.

[garbulky]

Well it could just be bad match with your setup. Bad synergy. However it is possible it's that TPA amp you got. They aren't really high end stuff... Try a nice emotiva mini-x, that may help ?
It may also be possible that your near field listening position isn't helping with the sound. When I listened really near field stuff was more in your face and the depth element was lost. The DC-1 does a better job than most with depth. "Flattening" things out and not allowing things to "breathe" may also cause a too focused sound that could lead to fatigue. Of course this is all hypotehticals. It may just be that the DC-1 doesn't produce the sound you like.

[mattgarman]

Apr 13, 2015 21:13:30 GMT -5 garbulky said:

However it is possible it's that TPA amp you got. They aren't really high end stuff... Try a nice emotiva mini-x, that may help?

I've personally been pretty impressed with these tpa311x amps. FWIW, I'm not alone, they have a big following over on DIYAudio and a couple other forums.

At any rate, I realized just now that every amplifier I own now is based on a TPA311x chip, so I can't completely eliminate that as the culprit. But I have tried several implementations: the Sure tpa3110; and two group buys from DIYAudio: dual DUG mono tpa3116 boards, and the new GMARSH "Wiener" tpa3118 board. All have different output filters. I've tried different power supplies to the tpa311x amp boards. The GMARSH Wiener board allows for the tpa3118's switching frequency to be changed easily; changing frequencies doesn't help with the fatigue.

Also, I've tried a decent number of DACs with these tpa311x boards, and never experienced this before. All the other DACs were lower-end than the DC-1.

Apr 13, 2015 21:13:30 GMT -5 garbulky said:

It may also be possible that your near field listening position isn't helping with the sound. When I listened really near field stuff was more in your face and the depth element was lost. The DC-1 does a better job than most with depth. "Flattening" things out and not allowing things to "breathe" may also cause a too focused sound that could lead to fatigue.

I eliminated that variable, as I also tried my DC-1 in the living room, so not nearfield, and also different speakers (Salk Songtowers). Note the normal DAC here is an Emotiva XDA-1, which I've used happily for many years. And for the last few months, the XDA-1 has paired with the DUG tpa3116 mono amps I mentioned above. I'm not sure how similar the implementations of the XDA-1 and DC-1 are; I know both are based on the AD1955 chip, so they at least have that in common.

I guess at this point, I'm wondering if this particular unit has an issue. I guess I'll take it up with Emotiva support.

[novisnick]

I've owned the DC-1 since Emotiva has offered it. I really do love the sound.   I've played so many different sources through it and have never,,,,ever,,,ever been fatigued  by it.

ive played two separate computers through it as well as a half dozen CD players and my Rega P6, still haven't tired of it    
The only time that I grew annoyed by it was MP3 s and streaming, neither of which was the DACs fault.

So sorry you have this annoyance, do you have treatments in your room?
if this has been addressed already then I'm at a loss.

Good Luck

[soundmaster]

I also have never experienced listeners fatigue, I've owned the XDA-1, and with that DAC I found it to be to bright, which is why I sold it and purchased the DC-1, which sounds focused with a lot of detail but never bright!

[KeithL]

First off, I'm a bit surprised to hear anyone suggest that there's anything "fatiguing" about the sound of the DC-1 (especially when compared to a DacMagic).

However, there are a few things that I really need to put into perspective here.

1) The actual process of converting digital audio signals into analog audio produces "extra energy" which must then be removed by a filter. This is true for both Delta-Sigma and other types of DACs. If you choose to avoid oversampling, then the design of the reconstruction filter becomes very tricky, and involves several design compromises. Oversampling DACs, which include the one in the DacMagic, the one in the DC-1, and almost all other modern DACs, avoid this by converting the audio to a higher sample rate internally, which then allows the use of a simpler filter which avoids those compromises. However, the oversampling process itself also requires some filtering to remove "extra energy". You should also understand that similar filtering is required if you upsample the digital audio using software (or let Windows do it); it is simply a consequence of the math involved. (The extra "artifacts" are part of the process itself, and are not a consequence of how well it's done.)

2) The TPA3110 amp that you're using is an 8 watt/channel CLASS D AMPLIFIER. This means that, just like a DAC, its "raw" output contains high levels of "digital artifacts", and so it too has a filter on its output to remove this energy. However, since it is delivering power into relatively low impedance loads, its output filters are rather more "prone to compromises" than the filters in a DAC (or in a software digital filter). They are also known to be sensitive to the load impedance you connect to them. Therefore, honestly, although I'm a fan of Tripath amplifiers, I would be much more willing to believe that the amp is "producing listener fatigue" than the DAC.

Generally "listener fatigue" is simply an obvious symptom of somewhat high levels of distortion - which is not something we've ever observed from the DC-1.

Note that ANY sample rate conversion (software or hardware) involves some filtering and so may slightly alter the sound - so it's quite possible that the difference you're experiencing is due to some slight change being introduced by the upsampling process itself.

Apr 4, 2015 11:20:59 GMT -5 mattgarman said:

Hi,

I've been listening to my Stealth DC-1 DAC for a few months now. It sounds great and is definitely the best DAC I've ever used... Except that I realized it causes listening fatigue fairly quickly. My music collection is almost entirely Redbook CDs, i.e. 44.1 khz, 16 bit. Source is a PC using Windows 7, Foobar2000, WASAPI (event) via USB (also tried optical), into DC-1. Amp is mostly DIY based on the Sure tpa3110 amp board, into DIY desktop speakers built around Mark Audio Alpair 7.3 speakers.

I used this same setup for about a year with a Cambridge Audio DACMagic (the original) prior to the DC-1. I never had any listening fatigue.

Just for kicks, I started experimenting with upsampling of my music. I ran some tracks through Secret Rabbit Code's libsamplerate, upping the sample rate by 4x to 176.4 khz. I can't hear any audible differences in the sound of the music, but the upsampled music does not cause any fatigue!

My very cursory understanding of sigma delta DACs like the DC-1 is that they employ some kind of filter to remove the artifacts created in the actual digital to analog conversion process. I also understand that designing the filter is a game of managed compromises, particularly with how close 44.1 khz digital is to the audible range. So perhaps the fatigue is caused by the filter, and the filter's dynamic changes with the higher sample rate? Just a random idea.

Just wondering if anyone might be able to comment on this phenomenon that I've observed. The upsampling experiment was more to see if I could perceive any improvement in sound quality. Since I don't, I'd rather not have to use it as a workaround.

Thanks for any thoughts on the matter.

[audiobill]

Agree, Keith, but important to add that third order distortion is more irritating than second order, as has been extensively documented.

And that's why many avoid pushing audio through sand, preferring gentle vacuums.

[ads]

Having owned DACs that seem to impart a glare or edge to high frequencies, as well as other DACs that seem to soften high frequency detail, I would characterize the DC-1 as more “neutral, detailed, and without edginess” vs. the other DACs I’ve owned…so much so, that I’m considering another DC-1 for a secondary system.

[KeithL]

True, most people agree that third HARMONIC distortion is more unpleasant than second harmonic.
(However, since the TOTAL harmonic distortion on the DC-1 is well below 0.001%, I can't honestly believe that anybody can hear any of any of the individual harmonics.)

And this is especially true since he's using a Tripath amp. Personally, I've found Tripath amps to generally sound very good, but they are known to produce quite a bit of spread spectrum noise, and most of them aren't especially low in THD either, so I would expect the THD from the amp to far exceed any coming from the DC-1.

Since the DacMagic isn't known for being especially neutral either, I'm more inclined to suspect that the DacMagic was "smoothing something over somewhere" that the DC-1 is not.

May 6, 2015 16:27:26 GMT -5 audiobill said:

Agree, Keith, but important to add that third order distortion is more irritating than second order, as has been extensively documented.

And that's why many avoid pushing audio through sand, preferring gentle vacuums.

[mattgarman]

May 6, 2015 15:44:22 GMT -5 KeithL said:

The TPA3110 amp that you're using is an 8 watt/channel CLASS D AMPLIFIER. This means that, just like a DAC, its "raw" output contains high levels of "digital artifacts", and so it too has a filter on its output to remove this energy. However, since it is delivering power into relatively low impedance loads, its output filters are rather more "prone to compromises" than the filters in a DAC (or in a software digital filter). They are also known to be sensitive to the load impedance you connect to them. Therefore, honestly, although I'm a fan of Tripath amplifiers, I would be much more willing to believe that the amp is "producing listener fatigue" than the DAC.
[...]
And this is especially true since he's using a Tripath amp. Personally, I've found Tripath amps to generally sound very good, but they are known to produce quite a bit of spread spectrum noise, and most of them aren't especially low in THD either, so I would expect the THD from the amp to far exceed any coming from the DC-1.

At the risk of being pedantic: the TPA311x series is made by Texas Instruments, and is not a Tripath amp (unless you are using "Tripath" loosely to refer to class D in general, like people use the word "Kleenex" as a generic term for facial tissue). They are indeed class-D, but as far as I'm aware, were developed independently of the Tripath amps. If I'm not mistaken, Tripath actually went bankrupt a while ago, so those designs haven't been updated in years (whereas the TI chips are comparatively new designs). I do have a true Tripath tk2050-based amp (one of the generally better-regarded HiFiMeDIY ones). It was replaced by the $10 Sure tpa3110 amp I bought on a whim a while ago. Subjectively, my experience moving from Tripath to the TI tda311x amps is consistent with what a lot of people say: better in every way.

I'll defer to your expertise on the output filtering, as my knowledge is somewhat superficial (it's only my hobby). I'm not sure what the switching frequency on the tpa3110 is, but the datasheet says it's permissible to use only ferrite beads on the output, and even that's only for avoiding AM interference. Same goes for the tpa3116 and 3118, though all the implementations of those chips I've seen and used do have LC output filters. But I do know that the switching frequency of the tpa3116 and tpa3118 is adjustable; the lowest setting is 400 khz, which is an order of magnitude beyond what anyone can hear. If not filtered by the LC filter, it will definitely be filtered by the speaker.

So, I've tried not just the 8wpc tpa3110 amp with the DC-1, but also several 3116 and 3118 amps. The 3118 is the most notable one, as the designer included a DIP switch for changing the amps switching frequency. Unfortunately, changing that did not have any impact on the DC-1 fatigue I am experiencing.

The THD figures for the tpa311x amps get ugly when they are pushed near their output limit, but are quite respectable well before that. And keep in mind, this is primarily for a nearfield setup at a typical 60-70 dB. I estimate I'm putting less than 1 watt into each speaker, so well below any kind of limits.

May 7, 2015 11:39:31 GMT -5 KeithL said:

Since the DacMagic isn't known for being especially neutral either, I'm more inclined to suspect that the DacMagic was "smoothing something over somewhere" that the DC-1 is not.

Also, to be clear, I've used several other DACs with these amps. The (original) Cambridge Audio DACMagic, the Emotiva XDA-1, a NOS TDA1387 DAC, a cheaper USB ES9023-based DAC. Indeed, the DC-1 is the most expensive one, for whatever that's worth. However, if the explanation is that lesser DACs are smoothing something over, that would have to apply to all the others as well, including the XDA-1.

As soon as I get some more parts, I'm going to build a simple class AB amp. That will be interesting to see if the fatigue effect goes away with a non-tpa311x amp.

I also have another DAC on order, an es9018-based one. Keith, I'll take you up on your excellent suggestion to do a blind ABX with the DC-1, as you have a valid point, it could now be a matter of self-fulfilling prophecy.

I don't believe that the DC-1 itself has a design flaw or is a bad DAC in any way. I hope that implication doesn't come across; it's certainly not my intent. I have two theories right now on the problem: (1) it's simply a bad match with the tpa311x amps, for whatever reason, or (2) this particular unit has an issue, e.g. some component operating ever-so-slightly out of spec. If, after comparing to yet another DAC; paring with a class AB amp; and doing a blind ABX, then I think we can eliminate theory (1) and see about the other one.