In Pursuit of true 24 bit resolution

[Matthew J Poes]

ok so I’ve had this nagging thought the last few months. The kind of thought that I rationally know is probably overkill and probably a path to unhappiness, yet I can’t shake it. I have to know, if Dacs that cost less than $100 are now finally able to reproduce near 24 bit resolution (note to date I’m unaware of any dac whose measured performance has proven to equal 24 bits, 144.5 db of dynamic range), then what about the rest of the system. Aren’t I holding back my ability to hear what a new 24 bit file can resolve if the rest of my system has a dynamic range limited to say just 100db’s and with a distortion figure far above the theoretical lower limit of these files.

I know, such a pursuit is fool hardy. My room is among the quietest of any I’ve known, it’s noise floor so low that my typical inexpensive measurement equipment is incapable of fully detecting the actual lower limit, but which most lies well below 25dbs. Yet even this is well above the theoretical noise floor of these digital files and often that of the recorded performance. I can’t shake it.

I’ve found in listening to these high resolution files that often sudtle differences are detectable but only under careful scrutinized conditions, and thus far only with headphones.

After reading reviews and measurements of the Benchmark AHB2 I found myself drawn to the idea that maybe the whole system should be considered and that maybe every component up to the speaker needs to offer at least 20bits of resolution.

This has lead to a desire to find equipment which is both inexpensive and technically perfect. It may not even sound good, but I want to know if having such a setup could somehow change my ability to resolve the finest of details buried in those high resolution files. So far this has lead me to search out amplifiers and preamplifiers that near perfect performance. So far that is largely pointing to high quality professional monitor controllers. Models from companies like SPL and Coleman Audio are currently on my short list. For amplifiers I have found models from Apogee Audio and Anthrm Audio with measures dynamic range values up around 120db.

I plan to acquire this equipment for testing over the course of the next year and to share my perceptions. My hope is that this leads to some level of enlightenment around how resolving a speaker system can be.

What do others think? As Digital technology becomes capable of storing more of the information in the original performance, low level details that were in the original performance and dynamic swings as big as life, does the reproduction system itself need to keep pace?

 

[Tony V.]

In reality can you actually hear a difference or are you just convincing yourself that you can because it says 24bit?
My issue with any digital format is that how do we know that the source and the output are truly able to produce the full range needed. Just because it says it's 24bit does not mean that it really is.
A 24bit output could simply have been up sampled or converted.
A really good recording at 16bit could sound just as good as a 24bit if it's done correctly. I would likely find it almost impossible to hear a difference between the two.

 

[Matthew J Poes]

In reality can you actually hear a difference or are you just convincing yourself that you can because it says 24bit?
My issue with any digital format is that how do we know that the source and the output are truly able to produce the full range needed. Just because it says it's 24bit does not mean that it really is.
A 24bit output could simply have been up sampled or converted.
A really good recording at 16bit could sound just as good as a 24bit if it's done correctly. I would likely find it almost impossible to hear a difference between the two.

You are absolutely right. The source material will dictate the true resolution of the musical file. If it isn’t high resolution the. Upsampling adds little to nothing.

I think that is different. It isn’t fair to say that because some material is not true 24 but then all 24 bit is not high resolution 24 bit. It just means that a 24 bit files provenance matters. If the file is truly high resolution then I want to be able to reproduce it.

As for can I hear it, I’ve actually participated in abx tests of this looking at audibility of bit rates and separately the audibility of sampling frequency. I’ve successfully passed these tests with headphones and with bit rage actually did so with 100% of my choices in one such test (a recording of a Bell was the source).

There is also really strong scientific evidence to support this view. Not only did a recently completed meta-analysis show that he current evidence supports the notion that we can in fact hear the difference in higher sampling rates. For those not familiar with a meta-analysis it’s a method of creating an experimental study based on the results of many high quality studies. This approach is the best way to look across studies in an objective manner and draw conclusions. Rather than looking at all studies equally and saying something like “out of ten studies 5 showed a difference and 5 did not” which is a false comparison to make, it allows you to create a single overall effect size across studies and evaluate the significance of the evidence.

I’ve also been studying Bayesian statistics as of late and used these same studies to do a Bayesian meta-analysis as a means of practicing he approach. My own findings added uncertainty to the results and still provided a similar final conclusion.

Bit rate studies aren’t quite there so I can only rely on my own tests results and experience. I find that cd quality 16 bit files are audiophile. They sound great and provide amazing sound. However i have proven to myself that given the right source material of sufficiently high quality, I can in fact discern them at a high rate.

Having just completed a conference based on Bayesian methods and with constant warnings over the misinterpretation of statistics and p values I want to be careful in my language. Just try to remember, those choices do NOT reflect guesses nor am I comparing against chance alone. In other words, my choices are not comparing against chance alone differences (and removing the word “alone” doesn’t make that statement any more true). I hope to test some ideas using Bayesian methods to see if the inclusion of prior information will yield more accurate results. This isn’t my field of study but it’s such a great use case that I may even seek to publish a methods case study paper on it.

Having said all that, my subjective opinion is that higher resolution music is only slightly different sounding at best. That it is very hard to discern differences. That when I do hear differences it is because I use listening tricks to do so. With the jingling keys and bells, I started with known files and studied them for hours. I listened to them sited to their provenance to look for differences. I often listened to difference files to identify what I’m looking for. Then I switch to the ABX files and do the test. These are the conditions by which I am able to get a 10 out of 10 forced choice test correct.

 

[Wayne A. Pflughaupt]

Here’s a little tidbit not widely known: No A/D or D/A converter actually performs at its “named” bit depth. For instance, 16-bit converters are doing good to have an actual 14- or 15-bit performance; the other data bits are lost to noise and distortion for various reasons that don’t bear discussion here. And distressingly, this loss increases disproportionally as bit depth increases: Twenty-bit converters might get only 16- or 17-bit performance, and 24-bit is doing good to deliver a mere 18-bit performance.

Other functional realities force limitations on both the analog and digital side of performance specifications as well. For instance, using the example of 24-bit converters because it’s easy to find information on them: If they were really able to deliver the 147 dB dynamic range they theoretically are capable of, they would have to be able to resolve signals as small as one billionth of a volt. Naturally, they can’t do that.

On the analog side of things, down in this range transistors and resistors produce noise just by having electrons moving around due to heat. So even if A/D or D/A converters could be designed to resolve such low levels, the low-noise requirements of the other circuitry in the component – power supplies, balancing ICs or transformers etc. – would be so stringent that they would either be impossible to build, or too expensive.

What is the result of these real-world limitations? Dig through the manuals of even higher-end 24-bit digital audio processors and you’ll find that the best dynamic-range spec they can muster is between ~105-115 dB, which is no better than the best analog gear. I’d certainly like to see the spec sheets on that gear from SPL, Coleman, Apogee, and Anthem you mentioned.

In reality, there are no DACs that are capable of discerning greater than 20-bit resolution, so any company that claims their DAC does better is simply whistlin’ Dixie, as Sonnie might say. Oh, they can decode 24-bits, because 24-bits does exist in software, but the output from their DAC has less than 20-bits of resolution and dynamic range. Using the lowest noise power supplies, the most sophisticated grounding, and the most sophisticated resonance control currently available in a digital-to-analog converter, current technology cannot resolve the least significant bit on even a 20-bit recording.

Of course that doesn’t even account for the distortion added by signal cables, amplification, and speakers, all of which would not resolve even an 18-bit recording.

When people claim to hear differences between 16-bit, 20-bit, and 24-bit recordings, it is not the difference between the bit depths that they are hearing, but rather the difference in the quality of the digital mastering. The fact is that even most so-called 24-bit recordings are mastered with less than 16-bit dynamic range.

Regards,
Wayne

 

[Matthew J Poes]

Thanks for your response Wayne. All the gear I mentioned was gear I had seen 3rd party measurements of. None of them resolve 24 bits nor is that really my intent. Earlier I mentioned recognizing that nothing is truly 24 bit right now, but rather wanting to come closer. The amps I mentioned all measured with between 120 and 135db dynamic range. The benchmark amplifier was tested In stereophile and is so quiet and of such low distortion that his test cable was adding distortion.

As for hearing differences. As mentioned I did this in ABX testing. Same file, 24 bit source material, down converted to 16 bit. Run theough a fully blinded ab comparator. In one such test based on a recording of a ringing bell, I chose perfectly 10 out of 10 times on one trial and 9 out of 10 on the other. I have to be able to hear the difference to do that and there is no difference in mastering.

 

[Matthew J Poes]

I should mention the real impetus for this is that when I have listened to these test files or when I have tried ABX testing I have been successful with headphones and never ever my speakers. I have very good speakers. I guess it bothers me that I know there are things getting between me and the music that are obscuring those details.

 

[Wayne A. Pflughaupt]

The amps I mentioned all measured with between 120 and 135db dynamic range.

Can you provide links to specs for these amps? The only one I was able to find wasn’t anywhere near that figure.

Regards,
Wayne

 

[Matthew J Poes]

https://benchmarkmedia.com/products/benchmark-ahb2-power-amplifier
133db unweighted mono.

https://medias.audiofanzine.com/files/da-800-specs-477790.pdf
Hum and noise: 117db and distortion at 4ohms were .007%. Thd is .07% which reflects a worse case figure. I’ll try to dig up the AP charts I have for these but that amplifier measured under .005% over almost the entire frequency band into 8ohms and noise was better than 120db accept for a slight rise from the power supply to 117db and a general rise at very high frequencies.

https://www.anthemav.com/products-current/type=amplifier/model=p2/page=specs
Signal to noise is 125db. Distortion is .0007% at 1khz. It rises at high frequencies as do most amps to around .005%.

Which did you find with much worse figures from what I claimed?

 

[Wayne A. Pflughaupt]

Don’t recall for sure, I think it was an Anthem, but not the one you linked.

https://medias.audiofanzine.com/files/da-800-specs-477790.pdf
Hum and noise: 117db and distortion at 4ohms were .007%.

https://www.anthemav.com/products-current/type=amplifier/model=p2/page=specs
Signal to noise is 125db. Distortion is .0007% at 1khz. It rises at high frequencies as do most amps to around .005%.

Both those figures are dBA, according to the links.

An A-weighted noise spec is a less-rigorous standard than an unweighted spec. As you can see in the graph below, an A-weighted curve rolls out the upper and lower frequencies. By comparison, an unweighted spec is a flat-response reference with no roll-out of the bottom or top end.

The problem with using an A-weighted curve to obtain a signal-to-noise spec, obviously, is that it “ignores” upper- or lower-frequency noise that may be present. So if a component happens to generate a bit more hum than it should, and this problematic component can only muster an unweighted spec of 88 dB, A-weighting would allow the manufacturer to “honestly” bump the figure up to a more respectable 94 dB. This is merely a hypothetical example of a component that’s only a bit worse than it should be; in reality A-weighting can “improve” a noise spec as much as 10 dB.

All that aside, I’m sure both those amps are excellent performers. That Benchmark amp looks especially stellar, assuming the noise specs are legit. However, unless everything in front of it is at least that good or better, it’s all for naught. The audio chain is as strong as its weakest link: If your amp has a noise floor of -130 dB, and your pre-amp or source is only -110 dB, then your system noise floor is -110 dB.

Regards,
Wayne

 

[Matthew J Poes]

The AHB2 is a ridiculously pure amplifier. Its distortion and noise are at the limits of test gear at this point.
https://www.stereophile.com/content/benchmark-media-systems-ahb2-power-amplifier-measurements
There is some backup but he had issues with measuring noise. I especially like that his test load and lead was adding distortion. That says something about how clean the thx AAA technology is.

I know the A weighting scale but appreciate the detailed response. I stated and posted numbers I could backup. I can't speak to the anthem as I don't have test data for it. I talked to a former engineer that helped design and test the Apogee amps and is who turned me onto it. He sent me a bunch of the test data and I more recently found some Audio Precision tests. That nosie spec, while A weighted is both worst case and relative to an arbitrary nonmax power value to allow a single figure for all. Since dynamic range is relative to max power the dynamic range of the Ca-800 is 3db higher and about 4-5 more than that bridged into 8 ohms (from what I recall of this conversation). The engineer pointed out that it's a fully differential amp input to output in bridged monk mode.

I think it may be an under-appreciated pro amp for home audiophile purposes.

Also sorry for typos, using a terrible amazon tablet

 

[Matthew J Poes]

Ok so the real impetus for this posting and this adventure is, as mentioned previously, the realization of two things:

1. There exist today digital systems whose measured performance is state of the art and at a price that anyone could easily afford
2. That my receiver and other components may not offer this state of the art performance and as a result resolution is being lost in translation

I came to this conclusion because of a blog I enjoy reading. I am not endorsing this blog or what it says, but I'm a sucker for measurements in a review and it was one of the few places measuring stuff I cared about. In the blog he measured an S.M.S.L. IDEA DAC which had state of the art performance equal or close to equal to that of the TEAC reference dacs and the Oppo BDP-105. This DAC costs less than $100 and can be connected to your phone or laptop for portable use, while still offering state of the art performance for home audio. I found this shocking, as other small pocket DACs that he measured on the blog or in stereophile had very excellent performance, but with issues here or there. Now mind you, he is measuring through a Focusrite USB interface whose A/D converter is likely a limiting factor. It is possible that -110db is the noise limit of his system. This may not reflect 24 bit resolution but it far exceeds 16 bit.

http://archimago.blogspot.ca/2017/06/measurements-oppo-sonica-dac-ess-sabre.html?m=1
See this for a set of measurements which include he idea dac.

Look at that chart. Those are impressive numbers.

http://archimago.blogspot.com/2013/10/measurements-onkyo-tx-nr1009-as-hdmi.html?m=1
Now look at this link. It’s a receiver similar to mine in age and componentry. A bit higher end model even.


Now look at this. Similar dynamic range but much worse noise floor and much higher distortion.

That rise in distortion is actually largely an increase in both noise floor and odd order distortion. What that doesn’t show is the increase in higher odd order distortion which is still low (but much much more audible at lower levels than lower order distortion).

So all this made me think that a better preamp with lower noise and a better dac with lower noise and distortion might be worthwhile. Is it audible, I can’t say. I’d like to find out.

 

[Matthew J Poes]

Here is a distortion example for comparison. First I’ll show a graphic that gives the threshold of hearing by harmonic accounting for masking.



Now look at a harmonic distortion plot for the onkyo

Those spikes reflect the harmonics. You will see most of the low order harmonics are low enough in level to be inaudible. The high order distortion is much higher in the onkyo than the other gear but still likely inaudible.

 

[AudioThesis]

By simply focusing on the bitrate and noise floor of your room, you will always limit yourself. Each component in your system, cables included, will have a noise floor. Some are higher while some are almost inaudible. This is usually one of the big factors that separates high end from mid-fi. A great example is the Marantz components. As you move up the line, you see better design regarding separating of digital, analog, and power sections. You also begin to see some big improvements on the chassis and additional separation between different components within the product.

Your room is much quieter than mine. Don't get stuck in one arena and not address the others. If you do, this will be a pursuit with no gains.

 

[Matthew J Poes]

Thanks Skip! Good to see you hear.

I’ve been really hung up on this idea lately but think you are right. It’s hard to figure out what those things would be to address.

Maranta gear is nice. One issue I take with their high end gear is that they don’t appear to be true differential designs. Their amps show more noise and distortion with a balanced connection than an unbalanced connection. That typically implies use of a balanaced to unbalanced conversion using parts noisier than the unbalanced input connection. With modern differential conversion chips like those of THAT it shouldn’t be the case anymore (and to me, if it’s being converted it’s no longer valuable).

 

[AudioThesis]

I use unbalanced for most of my connections, as do most audiophiles. I find that the cable itself is more important than unbalanced/XLR. I get more gain with the XLR cables than the unbalanced, but the SQ is identical (both sets are Wireworld Silver Eclipse).

While technical issues like that can be important, my approach differs vastly from yours. I listen first and then try to figure out why it sounds the way it does instead of the other way around. I think so many in this hobby can get stuck on those technical details that the music itself gets lost.

If I didn't have North Star Design as a product line, I'd gladly still use the Marantz gear as my reference sources. I've grown to appreciate the NSD products quite a bit more due to their sound, but I still like the Marantz products and recommend them based on the client's desired sound. Marantz still has a digital sound - slight bite to the top end, but not in a negative way. NSD products are smooth and detailed with a great holographic presentation. Two different approaches that work on a per system basis.

Anyways, my challenge to you is to use specs as a guideline and not a rule. I've heard amps and speakers that shouldn't sound good leave me shaking my head in disbelief. It happens.

 

[Matthew J Poes]

I use unbalanced for most of my connections, as do most audiophiles. I find that the cable itself is more important than unbalanced/XLR. I get more gain with the XLR cables than the unbalanced, but the SQ is identical (both sets are Wireworld Silver Eclipse).

While technical issues like that can be important, my approach differs vastly from yours. I listen first and then try to figure out why it sounds the way it does instead of the other way around. I think so many in this hobby can get stuck on those technical details that the music itself gets lost.

If I didn't have North Star Design as a product line, I'd gladly still use the Marantz gear as my reference sources. I've grown to appreciate the NSD products quite a bit more due to their sound, but I still like the Marantz products and recommend them based on the client's desired sound. Marantz still has a digital sound - slight bite to the top end, but not in a negative way. NSD products are smooth and detailed with a great holographic presentation. Two different approaches that work on a per system basis.

Anyways, my challenge to you is to use specs as a guideline and not a rule. I've heard amps and speakers that shouldn't sound good leave me shaking my head in disbelief. It happens.

Again, all fair points. Well, come AXPONA you are more than welcome to bring by some gear for comparison. As you know I am VERY interested in adding an aurilac piece to my system for Tidal and in need of a preamp.

I too have heard gear that while technically poor, did something special or interesting that left me wanting more. I’ve heard some really great full range single driver speakers with SET tube amps that I’m sure were technically poor, but were very musical and enjoyable to listen to. As I mentioned to you before, selling off my tube amplifiers was the worst mistake I ever made.

 

[AudioThesis]

Given the issues you had, I'm not sure I agree. Maybe not REPLACING them with another variant was a mistake, but selling yours appears to be a sound move. Those SET amps with big power tubes do something righteous to the music.

If you can find an AURALiC Aries Mini, I'd strongly suggest jumping on one now. There won't be another 'budget' streamer to touch it anytime soon and the production has stopped for good on them. I sold the last few that came stateside. The Bluesound units are an option, but their sound quality is well below what you get with the Mini. I've had several clients and seen people on other forums confirm this.

 

[Matthew J Poes]

Given the issues you had, I'm not sure I agree. Maybe not REPLACING them with another variant was a mistake, but selling yours appears to be a sound move. Those SET amps with big power tubes do something righteous to the music.

If you can find an AURALiC Aries Mini, I'd strongly suggest jumping on one now. There won't be another 'budget' streamer to touch it anytime soon and the production has stopped for good on them. I sold the last few that came stateside. The Bluesound units are an option, but their sound quality is well below what you get with the Mini. I've had several clients and seen people on other forums confirm this.

Does Aurlic have no plans to replace it with something similar in price? I recall reading they lost money on them early on by eating the cost of the Tidal subscription but no longer offer that.

I really hope they come out with a replacement. That was a great unit. It even had an ESS Sabre dac inside, something the Airies doesn’t even have.

Maybe I should just save my pennies for an Altair. Maybe if these side gigs pick up I’ll finally earn enough to justify. The wife says I can spend that money however I like, just no going nuts with our primary income.

 

[AudioThesis]

A critical component that helped them hit the price point was discontinued, causing them to halt production. I think they tried a few workarounds but couldn't get over that hump. There are no plans to pick up production again and I haven't been informed of any replacement coming down the chute.

To be fair, the Altair was built specifically to be a high quality streamer only and the DAC implementation in the Mini is OK, but not great. It actually did fair well against stand-alone $500 DAC's so that is saying something. Compared to the Altair, well, there is no comparison. Most client's I've had preferred to go with the Mini + NSD Intenso over the Altair, but that isn't an option anymore. The Altair is the best bang for the buck product AURALiC sells and you can't go wrong with one.

I built my system prior to dealing strictly by flipping audio gear. That experience taught me a great deal about most common products and began my branching into more boutique brands. Now I spend the money from my business on gear because I HAVE to, not because I WANT to lol. I'd much rather draw for the family than continue to buy more gear, but such is the life.

 

[Matthew J Poes]

A critical component that helped them hit the price point was discontinued, causing them to halt production. I think they tried a few workarounds but couldn't get over that hump. There are no plans to pick up production again and I haven't been informed of any replacement coming down the chute.

To be fair, the Altair was built specifically to be a high quality streamer only and the DAC implementation in the Mini is OK, but not great. It actually did fair well against stand-alone $500 DAC's so that is saying something. Compared to the Altair, well, there is no comparison. Most client's I've had preferred to go with the Mini + NSD Intenso over the Altair, but that isn't an option anymore. The Altair is the best bang for the buck product AURALiC sells and you can't go wrong with one.

I built my system prior to dealing strictly by flipping audio gear. That experience taught me a great deal about most common products and began my branching into more boutique brands. Now I spend the money from my business on gear because I HAVE to, not because I WANT to lol. I'd much rather draw for the family than continue to buy more gear, but such is the life.

Fair points.

By the way you should introduce yourself over in the new members section.

https://www.avnirvana.com/forums/introduce-yourself.13/

I know you! Not everyone else does. I think you will be a welcome voice here and look forward to more great talks.

 

[AudioThesis]

Do what forumites do best and get your read on in that subforum. You're being redundant.