My volts, over two days, as reported by the UPS at the PC, same branch circuit as the stereo:
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
General debate thread about audio measurements
- Thread starter FrivolsListener
- Start date
Seems completely uncalled for but I shall do my best to not respond to your posts.
Have a good day, unless you've made other plans - Don
My bad.
Arnold Krueger
Active Member
- Joined
- Oct 10, 2017
- Messages
- 160
- Likes
- 83
Scientifically speaking science if properly interpreted properly would seem to agree with you. If you eliminate bogus experiments that ignore well-known effects like Temporary Threshold Shift, the threshold of hearing for even the most obnoxious noises or distortions are no more than about 70 dB down, and often less. After that, reasonable gains are about safety margin for operations that are often repeated during production and playing. And after that, it is about numbers for the sake of numbers and just because we can.
It never fails to amaze me the things you can come up with numbered reports on.
I do and you do for sure, just try out some music with high dynamic, orchestra or jazz or something peaceful but with strong notes from time to time: you will hear the background noise!
Instead, subtracting 10 dB from the background noise of the DAC will make my ears unable to feel any noise, so 100 dB seems to be very difficult to hear in normal listening conditions.
Then why paying 2000 USD on a fancy DAC if is unable to get at least 15-bits of ENOB? This is why measurements are good, because it's not right to pay a huge amount of money on a device that doesn't even performs immaculate.
This is also happening when someone purchases a digital camera or even a car, because most of the time specs do make a difference and unless lot of gold and diamonds are involved why spending more on a car that has less power and less mpg?
P.S.: I have a DAC that has a SNR of around 110dB when playing PCM files and 83dB when playing DSD files (SNR measured via RMAA tests). Difference with the same song PCM vs. DSD is huge and sometimes, when listening DSD files with it, I think that good old Dolby C was better. So, when using that DAC I prefer to transcode DSD music to PCM to get rid of the background hiss.
There is a bug inside the DAC I was speaking about, most likely related to the DSP code inside. However, I'm happy with the software DSD-PCM transcoding, I honestly can't find any difference and if unsure, I can always have a listen to native DSD decoding from my completely noiseless Burson Play combo.
Source: http://archimago.blogspot.ro/2015/09/measurements-asus-xonar-essence-one-mki.html.
P.S.: -83dB is an important amount of background noise that I don't like at all, so 14-bits ENOB quite sucks!
Source: http://archimago.blogspot.ro/2015/09/measurements-asus-xonar-essence-one-mki.html.
P.S.: -83dB is an important amount of background noise that I don't like at all, so 14-bits ENOB quite sucks!
Just before I get in to the main meat. One should measure a $2000 balanced DAC using its balanced outputs. Why measure using the unbalanced? That balanced dac is meant to ideally used with balanced gear. I use my balanced DC-1 with balanced amps. Schiit provides balanced amps and preamps to be used with it.
Having said that....
I've mentioned that I take issue with the self stated "arbitary" 0.1 db measurement that @amirm uses to state DAC resolution and now there is another person talking about the 0.1 db which is Jude who also uses an AP.
From what I gather the reason Amir uses 0.1 db is because he really wants to talk about perfect linearity with absolutely no deviation. Also that if you increase the 0.1 db criteria to something more realistic say +/- 1 bit (or a 1 bit swing) , a lot of dacs appear to measure the same. So for him this is a way of differentiation. Which I disagree. Narrowing things so visually there's a differentiation and then using that to state number of bits is a bad interpretation. It may be visually different. But it lacks meaning when you use an arbitary number. It's like saying here are all these apples. I'm going to use my definition of apples as being +/- 1 nanometer in radius from this apple over here. Every other apple that fails my nanometer limit is a MUCH WORSE APPLE, perhaps even an orange. it doesn't make sense because it's arbitary and not actually part of a documented way of interpreting measurements of apples.
Here are the issues:
The biggest one is that he is using a 0.1 db variation criteria to state dac resolution. This is an issue because this is indeed an arbitary number. I have asked for some sort of valid documentation to back up the reason he uses 0.1 db to state dac resolution and have received none.
Now here's why this is an issue In Jude's head-fi post: he posts the linearity of the balanced output which performs quite a bit better than unbalanced
This is at about 117 to 122 db of linearity depending on channel if you expand it to 1 db and perhaps more if you expand it to 2 db.
How many db is represented as a bit? It's not 1 db. I'm pretty sure it's more than that! Certainly not 0.1 db.
Now keep in mind R2R DACs have issues with linearity at very low db values. But 117 db is really quite good. If you use that, you are going to get more than 16 bits of resolution. Don't quote me on this but it's going to be something like 18 or 19 bits.
Now here's where the issue comes from.
From Amir's interpretation using 0.1 db on the worse performing unbalanced outputs
You see that 0.1 db as a "resolution finder" cuts this resolution to -98 db which is a lot lower than 117-122 db of a 1 db variation linearity.
Now why is this an issue?
It's because of this
A Delta sigma dac is more linear . But the Topping which if I'm not wrong is one of the top measured DACS here gets a very nice 19 bits of resolution at .....
Coming in at 120 db using 0.1 db which is the limit of the measurement gear. Even if you relaxed that 0.1 db to 1 db that is about what you would get 120 db using the Topping.
Ther Ygdrassil DOES 117 to 122 db depending on the channel at the 1 db value which is very close to the Topping.
But there is also another method of determining bits in this case effective number of bits and that is using SNR which Amir does not perform though he has the capability.
That is this SNR = 6.02 × N + 1.76 dB,
If he did, I'm willing to bet that it wouldn't back up the "16 bits of resolution" the Yggy has. It would point closer to the 117 db or 122 db value of the Yggy getting it a much higher score -. Though to be fair the ENOB test does not take in to account linearity which is another aspect of dac resolution. But I mention it to just note that there is a way to state bit resolution that is accepted and doesn't include any 0.1 db values to come up with the values.
I personally think the linearity swing should be kept at +/- 1 bit (was that 6 db?) if one is using that to state bit depth. Because that would make sense to me. Once it gets to +/- 1 bit then you are now at a different bit resolution.
So we see not only that he presented the measurements of a balanced dac using its unbalanced outputs which isn't the most optimal way to measure a balanced DAC (becaused the balanced out measures better in this case). But also that the arbitray numbner of 0.1 db makes a BIG difference if you use it to arbitraryily state that as DAC resolution.
So here's the reason of my post.
I'll be the first person to admit that I know barely anything compared to you and your experience.
But what I would like to see is a more balanced approach with some more logical interpretations. This tremendous focus on minute differences aren't providing any kind of realistic context.
Of course this is your forum and measurements and you can run it the way you see fit. Here's my suggestions for improving them at least from my point of view.
- Please stop using 0.1 db as a way to state bits unless you have some proven documentation. It's misleading. People say why do I want this trash 16 bit dac when this Topping does 19 bits. When the reality is not quite as obvious.
You can say it varies away from perfect linearity if you want to use an arbitary 0.1 db number but also use wider definitions to state bits instead of using 0.1 db for bits. Also include the SNR ENOB measurement to help provide wider context regarding bits.
-Give the gear a chance.
If it's balanced - do balanced measurements.
If it requires a driver, use the driver! It's okay, you can do this! Use a throwaway old laptop if you want if you are very worried about system instability.
Measure and post your own ac power measurements as you said you would. This would just provide more stability to your published measurements.
- Take in to account different topologies. Provide a bit more context than focusing on specific measurements to imply large differentiators when different topologies do differently at those specific measurements.
- The hard line only measurements make little sense if you are comparing completely different topologies. It makes more sense to compare like topologies. R2R t R2R's. DS DACs to DS Dacs especially if there are things to take into account.
I noted Amir measured the Lyr and noted a higher distortion than solid state. This is true. Tubes have worse distortion. But you didn't focus on that the distortion was still quite low and the power output was quite a bit larger than the comparable solid state amp. There's a give and a take sometimes.
A way I would use to illustrate it is to compare electric to gas. Both get you where you are going. But using measurements without using context would get you bad information. This Tesla model S accelerates very very fast compared to this more expensive gasoline car. But we ignore that the gasoline car is not built to deliver all its torque right at the start like an electric engine. Or that at higher revs a gas engine can do better. And that a gas engine's fuel can be refilled faster than an electric battery.
Having said that....
I've mentioned that I take issue with the self stated "arbitary" 0.1 db measurement that @amirm uses to state DAC resolution and now there is another person talking about the 0.1 db which is Jude who also uses an AP.
From what I gather the reason Amir uses 0.1 db is because he really wants to talk about perfect linearity with absolutely no deviation. Also that if you increase the 0.1 db criteria to something more realistic say +/- 1 bit (or a 1 bit swing) , a lot of dacs appear to measure the same. So for him this is a way of differentiation. Which I disagree. Narrowing things so visually there's a differentiation and then using that to state number of bits is a bad interpretation. It may be visually different. But it lacks meaning when you use an arbitary number. It's like saying here are all these apples. I'm going to use my definition of apples as being +/- 1 nanometer in radius from this apple over here. Every other apple that fails my nanometer limit is a MUCH WORSE APPLE, perhaps even an orange. it doesn't make sense because it's arbitary and not actually part of a documented way of interpreting measurements of apples.
Here are the issues:
The biggest one is that he is using a 0.1 db variation criteria to state dac resolution. This is an issue because this is indeed an arbitary number. I have asked for some sort of valid documentation to back up the reason he uses 0.1 db to state dac resolution and have received none.
Now here's why this is an issue In Jude's head-fi post: he posts the linearity of the balanced output which performs quite a bit better than unbalanced
This is at about 117 to 122 db of linearity depending on channel if you expand it to 1 db and perhaps more if you expand it to 2 db.
How many db is represented as a bit? It's not 1 db. I'm pretty sure it's more than that! Certainly not 0.1 db.
Now keep in mind R2R DACs have issues with linearity at very low db values. But 117 db is really quite good. If you use that, you are going to get more than 16 bits of resolution. Don't quote me on this but it's going to be something like 18 or 19 bits.
Now here's where the issue comes from.
From Amir's interpretation using 0.1 db on the worse performing unbalanced outputs
You see that 0.1 db as a "resolution finder" cuts this resolution to -98 db which is a lot lower than 117-122 db of a 1 db variation linearity.
Now why is this an issue?
It's because of this
A Delta sigma dac is more linear . But the Topping which if I'm not wrong is one of the top measured DACS here gets a very nice 19 bits of resolution at .....
Coming in at 120 db using 0.1 db which is the limit of the measurement gear. Even if you relaxed that 0.1 db to 1 db that is about what you would get 120 db using the Topping.
Ther Ygdrassil DOES 117 to 122 db depending on the channel at the 1 db value which is very close to the Topping.
But there is also another method of determining bits in this case effective number of bits and that is using SNR which Amir does not perform though he has the capability.
That is this SNR = 6.02 × N + 1.76 dB,
If he did, I'm willing to bet that it wouldn't back up the "16 bits of resolution" the Yggy has. It would point closer to the 117 db or 122 db value of the Yggy getting it a much higher score -. Though to be fair the ENOB test does not take in to account linearity which is another aspect of dac resolution. But I mention it to just note that there is a way to state bit resolution that is accepted and doesn't include any 0.1 db values to come up with the values.
I personally think the linearity swing should be kept at +/- 1 bit (was that 6 db?) if one is using that to state bit depth. Because that would make sense to me. Once it gets to +/- 1 bit then you are now at a different bit resolution.
So we see not only that he presented the measurements of a balanced dac using its unbalanced outputs which isn't the most optimal way to measure a balanced DAC (becaused the balanced out measures better in this case). But also that the arbitray numbner of 0.1 db makes a BIG difference if you use it to arbitraryily state that as DAC resolution.
So here's the reason of my post.
I'll be the first person to admit that I know barely anything compared to you and your experience.
But what I would like to see is a more balanced approach with some more logical interpretations. This tremendous focus on minute differences aren't providing any kind of realistic context.
Of course this is your forum and measurements and you can run it the way you see fit. Here's my suggestions for improving them at least from my point of view.
- Please stop using 0.1 db as a way to state bits unless you have some proven documentation. It's misleading. People say why do I want this trash 16 bit dac when this Topping does 19 bits. When the reality is not quite as obvious.
You can say it varies away from perfect linearity if you want to use an arbitary 0.1 db number but also use wider definitions to state bits instead of using 0.1 db for bits. Also include the SNR ENOB measurement to help provide wider context regarding bits.
-Give the gear a chance.
If it's balanced - do balanced measurements.
If it requires a driver, use the driver! It's okay, you can do this! Use a throwaway old laptop if you want if you are very worried about system instability.
Measure and post your own ac power measurements as you said you would. This would just provide more stability to your published measurements.
- Take in to account different topologies. Provide a bit more context than focusing on specific measurements to imply large differentiators when different topologies do differently at those specific measurements.
- The hard line only measurements make little sense if you are comparing completely different topologies. It makes more sense to compare like topologies. R2R t R2R's. DS DACs to DS Dacs especially if there are things to take into account.
I noted Amir measured the Lyr and noted a higher distortion than solid state. This is true. Tubes have worse distortion. But you didn't focus on that the distortion was still quite low and the power output was quite a bit larger than the comparable solid state amp. There's a give and a take sometimes.
A way I would use to illustrate it is to compare electric to gas. Both get you where you are going. But using measurements without using context would get you bad information. This Tesla model S accelerates very very fast compared to this more expensive gasoline car. But we ignore that the gasoline car is not built to deliver all its torque right at the start like an electric engine. Or that at higher revs a gas engine can do better. And that a gas engine's fuel can be refilled faster than an electric battery.
Last edited:
For reference regarding my linearity post.
Here's what Jude has to say about it.
Linearity
Getting back to the linearity measurements, I want to ask again where the +/- 0.1 dB threshold on the linearity error test was standardized. While others may be using it, you're the only one I've seen using it. I saw that you recently cited this article by Robert Harley inStereophile in discussing the topic. I may have missed where Harley defined a +/- 0.1 dB threshold (I didn't read every page of that article). I did, however, search Stereophile for Harley's digital component reviews, and found this example:
View attachment 2506250
Of the above plot, Harley said:
In a review by Kalman Rubinson in Stereophile, I found this example:
View attachment 2506253
Of the above plot, Rubinson said:
In both cases (and I'm quite sure there are more), their descriptions of when the devices under test remain linear seem to far exceed your 0.1 dB tolerance. Looking at those plots, Harley seems to be describing ~1 dB deviation and Rubinson ~2 dB. I decided, then, to go with 1 dB deviation in linearity tests expressed in both a linearity error measurement and an RMS measurement (where 0 dBFS output at 1 kHz is 0 dBrA in each RMS example, to make the corresponding change in RMS (versus dBFS) easier to read).
Here is my linearity error measurement of the Yggdrasil2 from its unbalanced analog outputs (281 steps, 0.500 dBFS per step, from -140 dBFS to 0 dBFS):
View attachment 2506276
The lowest step on the X-axis at which the plots in the above linearity error measurement remain within 1.00 dB of 0.0 is -95.000 dBFS (left) and -99.000 dBFS (right).
Here is my RMS measurement (versus dBFS) of the Yggdrasil2 from its unbalanced analog outputs (281 steps, 0.500 dBFS per step, from -140 dBFS to 0 dBFS):
View attachment 2506281
The lowest step on the X-axis at which the plots in the above RMS linearity measurement have dBFS and dBrA levels within 1.00 dB of each other is -95.000 dBFS (left) and -98.500 dBFS (right). (Again, they're in 281 total 0.500 dBFS steps, which is why they end in either .000 or .500.)
As I've now said a few times, the Yggdrasil2 definitely performs better from its balanced outputs (which you did not post, but did seem to run measurements from).
Here is my linearity error measurement of the Yggdrasil2 from its balanced analog outputs (281 steps, 0.500 dBFS per step, from -140 dBFS to 0 dBFS):
View attachment 2506273
The lowest step on the X-axis at which the plots in the above linearity error measurement remain within 1.00 dB of 0.0 is -122.000 dBFS (left) and -117.500 dBFS (right).
Here is my RMS measurement (versus dBFS) of the Yggdrasil2 from its balanced analog outputs (281 steps, 0.500 dBFS per step, from -140 dBFS to 0 dBFS):
View attachment 2506277
The lowest step on the X-axis at which the plots in the above RMS linearity measurement have dBFS and dBrA levels within 1.00 dB of each other is -122.000 dBFS (left) and -117.500 dBFS (right).
Again, the performance from the balanced analog outputs is significantly better than from the unbalanced analog outputs (and, on these forums, as best I can recall, Schiit has always recommended using the Yggdrasil balanced). I'm not sure why you only included a balanced measurement for frequency response, but nothing else (e.g. linearity, etc.).
As for determining a DAC's resolution, when I find the time, I want to examine how John Atkinson from Stereophiledetermine's a DAC's resolution in bits, which is certainly different than your +/- 0.1 dB linearity error standard. When I do that, I'll show the corresponding measurements for this DAC (and perhaps others).
Here's what Jude has to say about it.
Linearity
Getting back to the linearity measurements, I want to ask again where the +/- 0.1 dB threshold on the linearity error test was standardized. While others may be using it, you're the only one I've seen using it. I saw that you recently cited this article by Robert Harley inStereophile in discussing the topic. I may have missed where Harley defined a +/- 0.1 dB threshold (I didn't read every page of that article). I did, however, search Stereophile for Harley's digital component reviews, and found this example:
View attachment 2506250
Of the above plot, Harley said:
In a review by Kalman Rubinson in Stereophile, I found this example:
View attachment 2506253
Of the above plot, Rubinson said:
In both cases (and I'm quite sure there are more), their descriptions of when the devices under test remain linear seem to far exceed your 0.1 dB tolerance. Looking at those plots, Harley seems to be describing ~1 dB deviation and Rubinson ~2 dB. I decided, then, to go with 1 dB deviation in linearity tests expressed in both a linearity error measurement and an RMS measurement (where 0 dBFS output at 1 kHz is 0 dBrA in each RMS example, to make the corresponding change in RMS (versus dBFS) easier to read).
Here is my linearity error measurement of the Yggdrasil2 from its unbalanced analog outputs (281 steps, 0.500 dBFS per step, from -140 dBFS to 0 dBFS):
View attachment 2506276
The lowest step on the X-axis at which the plots in the above linearity error measurement remain within 1.00 dB of 0.0 is -95.000 dBFS (left) and -99.000 dBFS (right).
Here is my RMS measurement (versus dBFS) of the Yggdrasil2 from its unbalanced analog outputs (281 steps, 0.500 dBFS per step, from -140 dBFS to 0 dBFS):
View attachment 2506281
The lowest step on the X-axis at which the plots in the above RMS linearity measurement have dBFS and dBrA levels within 1.00 dB of each other is -95.000 dBFS (left) and -98.500 dBFS (right). (Again, they're in 281 total 0.500 dBFS steps, which is why they end in either .000 or .500.)
As I've now said a few times, the Yggdrasil2 definitely performs better from its balanced outputs (which you did not post, but did seem to run measurements from).
Here is my linearity error measurement of the Yggdrasil2 from its balanced analog outputs (281 steps, 0.500 dBFS per step, from -140 dBFS to 0 dBFS):
View attachment 2506273
The lowest step on the X-axis at which the plots in the above linearity error measurement remain within 1.00 dB of 0.0 is -122.000 dBFS (left) and -117.500 dBFS (right).
Here is my RMS measurement (versus dBFS) of the Yggdrasil2 from its balanced analog outputs (281 steps, 0.500 dBFS per step, from -140 dBFS to 0 dBFS):
View attachment 2506277
The lowest step on the X-axis at which the plots in the above RMS linearity measurement have dBFS and dBrA levels within 1.00 dB of each other is -122.000 dBFS (left) and -117.500 dBFS (right).
Again, the performance from the balanced analog outputs is significantly better than from the unbalanced analog outputs (and, on these forums, as best I can recall, Schiit has always recommended using the Yggdrasil balanced). I'm not sure why you only included a balanced measurement for frequency response, but nothing else (e.g. linearity, etc.).
As for determining a DAC's resolution, when I find the time, I want to examine how John Atkinson from Stereophiledetermine's a DAC's resolution in bits, which is certainly different than your +/- 0.1 dB linearity error standard. When I do that, I'll show the corresponding measurements for this DAC (and perhaps others).
Sorry, no. I provide the full measurements for anyone to analyze with any criteria which is exactly what you did. If I were just providing the number of bits as the only piece of data, you could complain but not when the entire measurement spectrum is provided in a graph. You can use any criteria you want, 1 db, 10 db, etc.
What you can't do is to have me adopt your standards for excellence in engineering. My Audio Precision analyzer is nearly 20 years old. Yet its DAC has no problem almost getting a perfect response to 120 dB. It needs no relaxing of standards to get there. If you are going to sell me a luxury DAC in this day and age, for thousands of dollars, you better believe I am going to demand that it at least match the DAC in my 20 year old instrument.
And it is not like other products fail that. We have DACs as cheap as $200 that ace that test. And it is no accident because they were designed with care and importantly, performance verified using instrumentation prior to product release.
There is no way, no how I am going to reward companies that charge large sums of money in this day and age for DACs yet we need to make excuses for them left and right why they did not wash their dishes before serving us food on them.
But again, nothing is foreclosed from you as readers. Read the full graph, apply whatever criteria you want. But please don't tell me that as a design engineer and manager who strives for excellence, I have to reward those who lack the same sense of responsibility.
As I noted elsewhere, this is an incorrect argument. Just about any DAC silicon has balanced output. Yet many manufacturers put unbalanced outputs on them and using those, I am able to get superbly better results than Schiit multibit. You are being fed and PR response line and taken advantage of the fact that you are not a design engineer and don't know what I just explained about other DACs.
- Joined
- Feb 23, 2016
- Messages
- 20,691
- Likes
- 37,419
I agree the .1 db deviation limit is too tight in my opinion for linearity. However, as long as the whole measurement data is presented you can interpret measures from two different sources in whichever criteria you like.
What limit would you use? We could use +/- 3db. If the output is off more than that then it is presenting an output that belongs to another bit level. For that reason I think no one should use more than that. For instance a whole 1 bit deviation is too much. However, as so many can do better than that at very low signal levels I do agree with the idea we shouldn't accept such a wide limit. So my choice would be +/- 1 db which is itself arbitrary. I think .1 db is too tight partly because we are usually measuring at very low signal levels and some odd noise could interfere enough to throw off your results that much. Maybe that is too forgiving because I've done this multiple times with the same gear and find the results surprisingly consistent. Yet some lesser gear may lack quite that level of consistency without differences in time growing large enough to be any concern.
The other approach would be to determine how much linearity error is audible. That would be a very sticky wicket however.
So if Amir's graph shows a little wrinkle at 12 bits, but it stays pretty good for several lower bits I wouldn't be bothered. If there is a wrinkle at 12 bits and the next several bits go wildly off the proper line, then yes I would say that DAC has a linearity issue.
What limit would you use? We could use +/- 3db. If the output is off more than that then it is presenting an output that belongs to another bit level. For that reason I think no one should use more than that. For instance a whole 1 bit deviation is too much. However, as so many can do better than that at very low signal levels I do agree with the idea we shouldn't accept such a wide limit. So my choice would be +/- 1 db which is itself arbitrary. I think .1 db is too tight partly because we are usually measuring at very low signal levels and some odd noise could interfere enough to throw off your results that much. Maybe that is too forgiving because I've done this multiple times with the same gear and find the results surprisingly consistent. Yet some lesser gear may lack quite that level of consistency without differences in time growing large enough to be any concern.
The other approach would be to determine how much linearity error is audible. That would be a very sticky wicket however.
So if Amir's graph shows a little wrinkle at 12 bits, but it stays pretty good for several lower bits I wouldn't be bothered. If there is a wrinkle at 12 bits and the next several bits go wildly off the proper line, then yes I would say that DAC has a linearity issue.
Careful! Have you seen documentation how that linearity test is performed on APx555? This is very important because out of the box, APx555 measurements for linearity are completely unsuitable for this use. The analyzer itself heavily influences the measured results making the outcome useless.
To demonstrate that, I provide the stock linearity measurements on APx555 together with my customization to get me to where my older analyzer is at:
This is a loopback test meaning the Audio Precision APx555 is measuring its own performance. The connection from input to output is direct (so not even a wire).
Now look at at the gray/blue curve labeled unoptimized. This is the stock measurement of Linearity that comes from Audio Precision. As we see as soon as you go to levels below 100 dB, the error starts and it quickly climbs way high (well above 10 dB which I have picked to match Jude's graph).
What this means is that the AP has a positive error in "linearity" (It is more than that but let's go with it) at levels less than 100 dB. As such, you can not, let me repeat, NOT measure any DACs with it as you will be showing the sum total error of both the DAC and ADC measurement errors in the Audio Precision.
The graphs in my measurements of Schiit came from my much older Audio Precision analyzer which in this care, provides much better results. Reason is that my older analyzer is a combination of an analog analyzer and digital one. For the purposes of linearity tests, both are combined to produce spectacularity clean analysis of linearity down to -120 dB as you see in my graphs. The contributions from the analyzer are in the order of half a dB or less at the extreme.
Fortunately, if you know signal processing, and understand how linearity is measured, you can correct for most of the error in APx555 as I am showing in red graph. Notice how the error is less than 0.5 dB or so extending to -140 dB. I am still fine tuning this measurement to make it even more accurate. But for now, I don't accept linearity measurements from APx555 without documentation on how they are performed. And you should not either.
To wit, if I feed the stock measurement a DAC that as negative linearity error, it can cancel the error in the APx555 and incorrectly show it to have good linearity!!!
Summary
Some of you scrutinize my measurements left and right every which way.
But you put your guards down completely when you saw other measurements to your liking. Alas, in this case, we are talking about a very delicate measurement that digs deep, way deep into what the DAC is doing and with it, lots of signal processing and advanced toppings like that come into play. So while the measurements may be correct, you cannot rely on them without further documentation that demonstrates any error is from DAC and not ADC+DAC.
Another point, how is it that on one hand we want relaxed set of criteria for measurements and yet we are provided with linearity tests down to whopping -140 dB??? I stop at -120 dB because that is just beyond reasonable levels of our hearing range. Going farther out to -140 dB that brings with it lots more noise relative to signal with no better understanding of how the device may sound.
Rod
Addicted to Fun and Learning
- Joined
- Mar 24, 2018
- Messages
- 744
- Likes
- 332
You know what I did? I didn't buy shiit or topping. I looked at what Amirm was using and figured that if was using that for his personal use dac then that is probably what I should buy. The same time the RME adi-2 dac came out so that's what I did. Worked for me, sounds dam good.
Rod
Addicted to Fun and Learning
- Joined
- Mar 24, 2018
- Messages
- 744
- Likes
- 332
A lot off this stuff discussed here goes over my head anyway.
"From what I gather the reason Amir uses 0.1 db is because he really wants to talk about perfect linearity with absolutely no deviation. Also that if you increase the 0.1 db criteria to something more realistic say +/- 1 bit (or a 1 bit swing) , a lot of dacs appear to measure the same. So for him this is a way of differentiation. Which I disagree. Narrowing things so visually there's a differentiation and then using that to state number of bits is a bad interpretation. It may be visually different. But it lacks meaning when you use an arbitary number. It's like saying here are all these apples. I'm going to use my definition of apples as being +/- 1 nanometer in radius from this apple over here. "
What, what. Hu?
"From what I gather the reason Amir uses 0.1 db is because he really wants to talk about perfect linearity with absolutely no deviation. Also that if you increase the 0.1 db criteria to something more realistic say +/- 1 bit (or a 1 bit swing) , a lot of dacs appear to measure the same. So for him this is a way of differentiation. Which I disagree. Narrowing things so visually there's a differentiation and then using that to state number of bits is a bad interpretation. It may be visually different. But it lacks meaning when you use an arbitary number. It's like saying here are all these apples. I'm going to use my definition of apples as being +/- 1 nanometer in radius from this apple over here. "
What, what. Hu?
It's not arbitrary, the reasoning and logic behind it have been explained many times. We get that you don't think these measurements are reasonable, you have said so a number of times, but you also state your understanding is limited. Its not misleading.
At the end of the day you just seem to want to find a reason to defend poorer technically performing kit. Schitt apologist.
The information is presented. You can choose your own interpretation, but don't expect Amir to defend kit whose technical performance is obviously deficient.
What really bemused me about this is that these dac measurements show that some cheap or modestly priced dacs perform technically immaculatly. Why is anyone therefore interested in very expensive dacs that don't?
You may be better off reading a newspaper that coincides with you your political views.
Last edited:
Similar threads
