What kind of evidence do you consider as sufficient?

Status
Not open for further replies.
It looks like there is blame to go around everywhere, and I would agree it starts with ESS. They don't do a good job of applications support or even know how to make their own dacs sound their best.

Then there are DIYer's who think they want tubes and all kinds of options that don't actually make SQ better for a SOA dac. How about a nice NOS dac if tube sound is wanted?

There are vendors that produce stuff customers ask for but that have wires running all over the place between spread out boards that might better share one ground plane and set of power planes. Nice to get rid of the ground loops too.

Unfortunately, the situation with ESS dacs is that designers are mostly on their own, and that includes needing to do a lot of research and measurements to come up with a good design. That means having a certain amount of expensive test equipment. It matters what jitter is at 1Hz or 10Hz, I don't care if it still looks pretty good at 100Hz. It's like 1/f noise, it can get worse fast as frequency goes down.

I think Benchmark is the only company that says it offloads interpolation filtering, but all one gets from ESS is the hint that you can do it.

What about upsampling, trade-offs between high sample rate capability and sound sound quality at low sample rates? What is the trade off if any? What are the options? Nobody seems to know anything except ESS and Benchmark. ESS isn't saying much, and Benchmark seems to have done a lot of their own discovery in their own lab.

Still, DIYer's are not paying attention to these things, not asking all the right questions, and not doing their own testing to find out what they need to know.

In other words, its rather a mess. But, the fact remains that ESS dac sound is very implementation dependent, there are a lot of compromised implementations, and a carefully engineered one can have superb sound quality.

I do want to clarify that I like their product and would use it, but the fact is they don't disclose any details of their listening tests and never submitted a paper to the AES about it that I am aware of.

@Abraxalito: My answer was in my followup question. One can only assume they find it satisfactory because they did demos of the product with it.
 
Last edited:
There's no disagreement that they're content with that approach. However you earlier attributed the reason for that approach - inability. I suggest there's another possibility - choice not to. Hence my question.

Right, but if they are doing listening tests that are truly effective, how did they conclude their modulator is better and then "miss" on other parts of the design, in the opinion of some?

Anyway, that's the last I'll comment here because it's just going to go in circles forever. 🙂
 
Right, but if they are doing listening tests that are truly effective, how did they conclude their modulator is better and then "miss" on other parts of the design, in the opinion of some?

I also would like to know the answer to that. Its clear that marketing-wise, the THD and SNR numbers are crucial to the positioning of the DAC. SQ appears to be a secondary concern - after all there's so far been no update on the RMAF talk (in 2012 I believe) featuring the later generations of DAC.
 
Wrong answer, sorry. I know you could have done much better though, sometimes I see the ability in you. If you could just wait until the initial emotional responses pass, given it some time and thought about more... Why don't you consider giving it a try? Just wait until the urge to respond immediately passes, wait until the dislike passes. You and everyone else will likely think the quality of your responses has improved. All the more so if you can keep it brief, linearly organized, and to the point. </hopefully friendly lecture>
Dear mark, I waited to reply - your post still reads as highly condescending - what should I do?

avid reader, mmerrill99
 
Well, condescending definition says:"having or showing a feeling of patronizing superiority."

If it helps to clarify, I sure don't feel superior. You know a lot about hearing because you studied it, right. When talking to other people about it, would you say that you feel somehow superior to them? My guess is you will not say that you do feel superior, although I guess someone could think that you do.

It would probably be more accurate to say you might be more learned in that field. Learning something doesn't make you superior as a person, although you might be able to tell someone useful information to help them listen better based on what you learned.

What I suggested to you was something intended as very analogous, as something I know more about that you do, I'm pretty sure. Ten years of serious study in the applicable area. Something I could tell you about.

But, I also know this: You have had some time to think about it and dislike has not passed. That is to say, based on the condescending interpretation or description it sounds like the dislike has not passed. If you think you will likely feel the same way about it tomorrow, then I would say you should forget about and not try it any more. It may not be for you, is all.

Maybe like not believing ABX is perfect would not be for someone else. That is to say, if someone's mind is already firmly made up, it is unlikely they will change. It sounds you have firmly made up your mind my suggestions are not right for you. So, that's it. It didn't help. Sorry. I tried.
 
On the other hand there are people building or putting together ESS dacs on this very forum who don't have any idea of how to make one sound it's best. This is actually I think the first time I have said anything about it, and I think it would probably be best to stop here and leave it at that.

It has been brought to my attention that I may have spoken so strongly on this matter that people could easily misinterpret my meaning.

I would therefore like to clarify on one point. With regard to these ESS dacs, making one sound truly its best is something almost nobody anywhere knows how to do, including me. There is only one best sounding ESS implementation I know of and it is that of Benchmark DAC-3. There may be others too, but if so I don't know of them. Some of how they do it appears to remain as trade secret.

While I feel I spoke accurately to the effect there are some people who seem to have no idea of how to make ESS dacs sound their best, no doubt there are some who probably have some pretty good ideas of what is likely left that can be improved towards reaching that goal. Hopefully, we will all keep learning and making progress to get there. I know I will keep thinking on it. 😀
 
Last edited:
Well, condescending definition says:"having or showing a feeling of patronizing superiority."

If it helps to clarify, I sure don't feel superior. You know a lot about hearing because you studied it, right. When talking to other people about it, would you say that you feel somehow superior to them? My guess is you will not say that you do feel superior, although I guess someone could think that you do.

It would probably be more accurate to say you might be more learned in that field. Learning something doesn't make you superior as a person, although you might be able to tell someone useful information to help them listen better based on what you learned.

What I suggested to you was something intended as very analogous, as something I know more about that you do, I'm pretty sure. Ten years of serious study in the applicable area. Something I could tell you about.

But, I also know this: You have had some time to think about it and dislike has not passed. That is to say, based on the condescending interpretation or description it sounds like the dislike has not passed. If you think you will likely feel the same way about it tomorrow, then I would say you should forget about and not try it any more. It may not be for you, is all.

Maybe like not believing ABX is perfect would not be for someone else. That is to say, if someone's mind is already firmly made up, it is unlikely they will change. It sounds you have firmly made up your mind my suggestions are not right for you. So, that's it. It didn't help. Sorry. I tried.

Don't worry, I don't dislike you - I might dislike that particular post but I don't carry it forward & it doesn't color my interactions with you.

People may dislike what I say but I would request that they also try not to let it become personal & color every interaction .

Just a point of correction for everyone - I treat posting on a forum like a conversation & when I have something to say, I post it - if my replies come quickly after a post, it doesn't mean I'm emotional or angry - please understand this & don't project onto me emotional charge that isn't there

Remember this form of communication lacks so much of the context we would have if involved in a face to face conversation, things can be easily misinterpreted
 
<snip>
One issue with the Harmon tests was that the electronics used to drive the speakers were not particularly impressive so this becomes as much an amp/speaker combination test - it's actually the full playback chain that's under test. <snip>

Toole did this before he joined Harman, but i´ve to check back if he gave some informations about the whole chain.
But the main point was the suprisingly wide spread in ratings given by listeners, so i´d argue that it doesn´t matter (at this point) if it was just based on the loudspeakers or on the combination of amplifier and loudspeaker.

<snip>
I think what might be confusing is that we can judge the same reproduction anomaly in different ways as usually there is not just one flaw so we may have different weightings between these flaws - maybe that's what you mean by "multidimensional evaluation" If that's what you meant then I agree but If we could just isolate to hearing one flaw at a time, I'm pretty sure we would all coalesce to the same correct sound.

Multidimensionality in sensory testing means that the stimulus might have been changed in several parameters and the evaluation by test subjects is not restricted in any direction, so they have to judge/consider the performance overall.

From food test(s) a typical example for an onedimensional test would be a formulation with more salt compared to the original one with less salt.
Asking the test subject(s) wich one tastes more saltier means they only consider the saltiness and in this case even the direction is prededined so it´s an onedimensional, directional test.

Sometimes as an experimenter you are fooling yourself in thinking to do a onedimensional test while in reality there is some crosscoupling so that the impression overall changed in several directions.

Therefore i´m a bit reluctant wrt your hypothesis about the converging judgements although it might be true. 🙂
 
Toole did this before he joined Harman, but i´ve to check back if he gave some informations about the whole chain.
But the main point was the suprisingly wide spread in ratings given by listeners, so i´d argue that it doesn´t matter (at this point) if it was just based on the loudspeakers or on the combination of amplifier and loudspeaker.
Maybe I'm confusing tests then - the speaker blind tests I'm talking are these from Sean Olive at Harmon
Audio Musings by Sean Olive: The Dishonesty of Sighted Listening Tests


Multidimensionality in sensory testing means that the stimulus might have been changed in several parameters and the evaluation by test subjects is not restricted in any direction, so they have to judge/consider the performance overall.
OK, but all that's changed in the above test is the speaker however the speaker interaction with the amplifier is a hidden secondary parameter which I was referring to as a possible parameter in relation to Quads (also the difference between box & panel speakers with regard to room interactions & placement)

From food test(s) a typical example for an onedimensional test would be a formulation with more salt compared to the original one with less salt.
Asking the test subject(s) wich one tastes more saltier means they only consider the saltiness and in this case even the direction is prededined so it´s an onedimensional, directional test.

Sometimes as an experimenter you are fooling yourself in thinking to do a onedimensional test while in reality there is some crosscoupling so that the impression overall changed in several directions.

Therefore i´m a bit reluctant wrt your hypothesis about the converging judgements although it might be true. 🙂

OK, I get you & indeed playback systems are a complex interconnected system with all sorts of current flows & signal flows happening concurrently - they are non-linear systems & difficult to predict.

I agree that we are judging the efficacy of an illusion which complicates matters significantly but I don't see why this is categorically different from judging TV or film illusions of reality - in those instances we evaluate a moving image based on similar criteria - how realistic the illusion is perceived to be
 
I would therefore like to clarify on one point. With regard to these ESS dacs, making one sound truly its best is something almost nobody anywhere knows how to do, including me. There is only one best sounding ESS implementation I know of and it is that of Benchmark DAC-3. There may be others too, but if so I don't know of them. Some of how they do it appears to remain as trade secret.

Thanks for this post. The discussion IMO has become narrowed to audibility tests for one particular implementation of a DAC. There is still a strong DIY community around traditional multi-bit and even NOS DAC's. The artifacts between the two are quite different.

I might add if anyone thinks there have never been any "chip" disasters they just never had a chance to see them.
 
When I perform mods, the nature of these mods is such that I can readily/instantly swap between 'A' and 'B' conditions.

Usually these changes are beneficial (but not always) and I allow systems to run for extended period in the 'B' condition before reverting to 'A' condition as check/confirmation.
By learning the 'two sounds' fully, I find it is easy to distinguish the two sounds reliably.

Consider the hi-fi shop showroom speaker comparator.....the salesman flicks through the choices and IME all speakers in such environment sound different, and with preferences.
What's so hard about identifying and describing differences between only two speakers....or amps....or preamps....or DAC's in a quiet familiar home environment ?.

Regarding the Foil/MF resistor test, I downloaded, topped and tailed the files to same alignment/length
I have yet to listen on my big system and run Foobar ABX test, however the differences are readily apparent on laptop speakers and I venture this should be relatively easy and repeatable.


Dan.
 
When I perform mods, the nature of these mods is such that I can readily/instantly swap between 'A' and 'B' conditions. What's so hard about identifying and describing differences between only two speakers....or amps....or preamps....or DAC's in a quiet familiar home environment?

Bravo…

There's nothing significantly difficult about doing pseudo-ABX testing at home, so long as you've set up the listening conditions and 'kit' to do the ABX testing reasonably competently!!!

I purchased some time back a number of small, quiet, gold-contact 3PDT ⅜" toggle switches. I employ them 'in circuit' for all sorts of A-vs-B (-vs-C-vs-D…) reconfiguration of circuits "on the fly". Sometimes (changing bias methods) the switching is potentially damaging enough that one (unfortunately) has to power down at least the HV part of an amplifier pair (hence why I always have HV "cutout" DPDT switches hard wired in…), but in general a lot of reconfigurations only take a second or two of switch flipping.

In one amp, I actually had a push-button normally-closed 3PST "fatty" switch on each monoblock which when held down would break the output HV, the 'downwind' HV, and the output transformer's to-speaker secondary common to cut the astounding POP that some circuit configuration changes could make. Quiet as a mouse. Hold down the big red button, then flip the "A-B" switches, then let the big button go.

It was an old WW2 surplus device … Sprague? I can't remember … which had the unique function of having a particular order to the 3 poles make-break sequence. Therefore, one pole would cut the speakers first, then the output HV, then the prior-stages HV. When released, it'd do that in reverse. Cool… I wish they still existed.

Having such a setup allows for a lot of learning rather quickly. One can identify virtually instantly that capacitor bypass of cathode-bias resistors “changes” the output substantially. Gain, of course is the first thing impacted, but also the "quality" of audio passing thru the amp. Better without the cap? Well… that's like stating that Cabernet is better than Zinfandel. Up to the listener, I'm afraid.

But moreover, in so doing this, I was able to also test out 'constant current sources' as front-end loads vs. appropriate resistors; I was able to (with a trio of 3PDT switches) try out interstage coupling transformers vs. appropriately configured anode load resistor → coupling cap → next grid paths. Lundahl interstage, too. Remarkable acoustic differences when very carefully "padded" for equal output volume. (Having old fashioned bakelite 'pointer knobs' and little triangles of sticky tape to identify A and B pad positions makes it all quite easy to do competently.)

When cost is not a significant material object, I can say with some modest assurance in my own opinion, that I prefer interstage coupling transformers in general, but if in only "one position" then in a single one in service as the phase-inverter function driving a pair of push-pull output finals. Interstage XFMRs do great service in that position. Eliminates one valve, too.

Well… clearly I've rambled overlong. Apologies.

GoatGuy
 
Please don't get this thread shut down too

Why do you think it was my doing that the thread was closed - do you know something we don't know? - I see nothing untoward with that thread, maybe you - do you?

As far as the question posed in this thread title - I'm with RNMarsh - sufficient momentum of people indicates to me that something is worth investigating - my own auditory perception are then sufficient evidence based on all the factors I mentioned before - naturalness of sound
 
Last edited:
Bravo…There's nothing significantly difficult about doing pseudo-ABX testing at home, so long as you've set up the listening conditions and 'kit' to do the ABX testing reasonably competently!!!
Well… clearly I've rambled overlong. Apologies.
Thank you, and good to see that you have done your homework, in a thorough way....my method is quite different.

.....As far as the question posed in this thread title - I'm with RNMarsh - sufficient momentum of people indicates to me that something is worth investigating - my own auditory perception are then sufficient evidence based on all the factors I mentioned before - naturalness of sound
Yes, naturalness of sound is what it's all about. In my living space I am able to close all doors and windows and listen in an essentially silent room from near field through far field.

I am also able to open large doors to the back verandah and allow natural ambient sounds in, and also sit on the verandah surrounded by natural ambient sounds whilst listening to my hi-fi in the far field, or take a listen from the back fence in the far, far field. Having the ability to realtime reference to natural sounds in this manner is particularly useful IMHO, and probably quite unusual.

Dan.
 
Status
Not open for further replies.