• 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!

A new Dac to improve sound quality from Khadas Tone Board 1

raindance

Major Contributor
Joined
Sep 25, 2019
Messages
1,037
Likes
968
Not what the chips are able to decode. But what line outputs are able to produce, their bandwith could be limited, to some 20-20k, 10-30k... where 24/96 is up to 48k (or less, depending what is in the PCM data).
It's pretty challenging to design a line stage that is bandwidth limited unless you get it very wrong. Using decent op amps 50K bandwidth is probably about normal. It's not the limiting factor unless you intentionally screw it up. An example: years ago a friend was a mobile DJ. He came to me for help because his system sounded like a big, loud telephone with no bass or high frequencies. It turned out that the cheap Chinese mixer he was using used something like 741 op amps, designed as comparators and with insufficient gain-bandwidth product for audio. I changed them to TL081 series from Texas Instruments and voila, full hifi sound.

His mixer only had a frequency response of around 5KHz before the change! An extreme example, and a long time ago when technology wasn't quite what it is today.
 

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
His mixer only had a frequency response of around 5KHz before the change! An extreme example, and a long time ago when technology wasn't quite what it is today.

That is the point, technologies and standards have changed. And it seems we will move away from lossy MP3 standards. Acheivable with internet. Some how pushed away also from the 80's CD quality standard and the 741 era.

Several equipments are now available with hi-res certification. A definition, from the internet:

Hi-Res Audio (HRA) is lossless audio capable of reproducing the full range of sound from recordings that have been mastered from better-than-CD quality music sources, a sound that closely replicates the quality that the musicians and engineers were working with in the studio at the time of recording.

As far as I know, since years, artists and studios don't exactly work at the limit of 20k/16 bits, microphones can catch more, and mixes are made after upscalling, at at least 96/24, or using 96/24 records. Or the records and mixes are made with DSD formats. Positive point is that we now can read DACs are able to reproduce up to 20 bits for 24.

Then if you want to reproduce that, you need the adequat analog line output stages, not only the DAC/converter chipsets.
 

PierreV

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
1,437
Likes
4,686
Hi-Res Audio (HRA) is lossless audio capable of reproducing the full range of sound from recordings that have been mastered from better-than-CD quality music sources, a sound that closely replicates the quality that the musicians and engineers were working with in the studio at the time of recording.

Oh, we are well beyond that these days. The lossy codec that shall not be named can even recover the intent of dead musicians and engineers.
:cool:
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Not what the chips are able to decode. But what line outputs are able to produce, their bandwith could be limited, to some 20-20k, 10-30k... where 24/96 is up to 48k (or less, depending what is in the PCM data).

BW is determined by the sample frequency. DAC's are usually specified at 44.1kHz that's why you see those numbers.
The post filter is a lot higher, around 70 to 100kHz or so (depending on the steepness and max samplerate)

To be more clear and basic. Topping D50s is 20Hz - 20kHz (+/-0.1dB). Where is it at 30kHz? At -6dB?

That depends on the sampling frequency.
At 30kHz it is down -90B but only for 44 and 48kHz.

Then if you want to reproduce that, you need the adequat analog line output stages, not only the DAC/converter chipsets.

It is very cheap and easy to make a line output stage with very low noise, a few ohm output R, very low distortion and 1MHz BW.
There are but very few DAC chips that have an onboard I/V converter and or post filter output stage.
The post filter is usually set around 70kHz b.t.w.
 
Last edited:

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
The post filter is a lot higher, around 70 to 100kHz or so (depending on the steepness and max samplerate)

Open the box you have at home. I'm speaking about the analog output stages, those close to your RCA connectors. There are usually some op amps, and discrete components, resistors and capacitors. The characteristics there are fixed, don't change according to PCM or DSD rate.

If you think post filter detailed characteristics may have an incidence of the output analog signals, I would appreciate detailed specs for those also.

It is very cheap and easy to make a line output stage with very low noise, a few ohm output R, very low distortion and 1MHz BW.

Topping (just an example, other do the same) only communicate: 20-20kHz (-0,1dB).

I would appreciate do see more details specs about those outputs stages, (the analog section close to your RCA or XLR connectors):
- bandwidth at -3dB also (if it is a 1MHz BW, all vendors will be very proud to present the full specs)
- output impedance curve over that bandwith
- phase shift curve over that bandwith

Edit... they could also add:
- distortion over that bandwidth
- SNR over that bandwitdh
- maybe slew rate (shall be pico seconds order for 1MHz bandwith and low distortion there)
 
Last edited:

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Open the box you have at home. I'm speaking about the analog output stages, those close to your RCA connectors. There are usually some op amps, and discrete components, resistors and capacitors. The characteristics there are fixed, don't change according to PCM or DSD rate.

If you think post filter detailed characteristics may have an incidence of the output analog signals, I would appreciate detailed specs for those also.

Yes, you are talking about the post filter which also functions as the line driver.
As said they are usually set to around 70 to 100kHz depending on the max sample rate of the device.
They are fixed but the cut off point may vary from product to product depending on the max sample rate and used DAC chip and its HF content.

You can look at schematics of post filters yourself. I don't need to provide them for you. Just look at datasheets for known DAC chips.

Topping (just an example, other do the same) only communicate: 20-20kHz (-0,1dB).

specified for 44kHz sample rates only. I don't give a crap about what manufacturers specify. It's advertising material.
They could specify frequency range per sampling frequency but they don't.

I would appreciate do see more details specs about those outputs lines, (the analog section close to your RCA or XLR connectors):
- bandwidth at -3dB also (if it is a 1MHz BW, all vendors will be very proud to present the full specs)
- output impedance curve over that bandwith
- phase shift curve over that bandwith

Edit... they could also add:
- distortion over that bandwidth
- SNR over that bandwitdh
- maybe slew rate (shall be pico seconds order for 1MHz bandwith and low distortion there)

I merely mentioned it is easy to design a line output stage that can reach 1MHz. I never said this is present in a DAC.
Usually the output stage is the post filter which is BW limited because of the filtering it does.

Then you should ask manufacturers to specify this. Maybe just for you and a handful of others they are willing to change their marketting materials and get lots of other questions in return.

The info you want is product dependent. The phase behaviour is determined by the digital filter and the post filter so also filter setting and sample rate dependent.

When you want Amir to measure all those aspects for every sample rate he would have to spend days of testing on every DAC and publishing, creating plots etc. Getting schematics and/or disecting each device. Not going to happen.

All of this has nothing to do with the OP and his subjective found differences.
This is with a high degree of certainty caused by the 1.2dB (15%) higher output level which was not compensated for.
 
Last edited:

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
You can look at schematics of post filters yourself. I don't need to provide them for you. Just look at datasheets for known DAC chips.

I'm not interested in generic information chipset providers published, but in what vendors implemented. And what the specs really are, because this wil have an impact on audio. And this will vary between DACs/vendors/implementations for same or similar chipsets.

Fact is most communicate nothing more than 20-20kHz.

Nagra and Hegel where up to 40k and 50kHz for the output buffers (shall have an impact on phase shifts). But this wasn't priced the same.

When you want Amir to measure all those aspects for every sample rate he would have to spend days of testing on every DAC and publishing, creating plots etc. Getting schematics and/or disecting each device. Not going to happen.

I didn't say Amir must measure or get any schematics from vendors. Topic being here: "A new Dac to improve sound quality from Khadas Tone Board 1" (or what might be technical and audible differences between available DACs).

Now if Amir wants to add some plots, would be his choice. And that could be limited, to representative sample rates (like 44.1, a reference, plus 88.2, for an idea of what results could be).
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
And what the specs really are, because this wil have an impact on audio.

Consider audibility levels.

Fact is most communicate nothing more than 20-20kHz.

Again this is for 44.1kHz and NOT of the post filter.
It is also not valid for every other sample frequency.

Nagra and Hegel where up to 40k and 50kHz

96kHz most likely, and it is NOT the spec. for the postfilter.
I have never seen any DAC product folder specifying the post filter characteristics.
 

PierreV

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
1,437
Likes
4,686
Consider audibility levels.

Yeah, while I understand one might be tempted to speculate on the possible consequences of technical choices that, in extreme scenarios, might explain "audible" differences, I remain amazed by the fact that people aren't ever motivated to strongly prove that "audible" differences exist in the first place...
 

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
Consider audibility levels.

I shouldn't have taken time to read reviews. And made an error buying an expensive Topping DAC. Any $5 USB stick would perform exact the same as long as I adapt the levels.
 

PierreV

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
1,437
Likes
4,686
I shouldn't have taken time to read reviews. And made an error buying an expensive Topping DAC. Any $5 USB stick would perform exact the same as long as I adapt the levels.

1) Not any.
2) Connectivity may be an issue with a USB stick, and not everyone loves a mini jack to jack cable hanging from their main integrated amplifier/preamp. But yes, it works quite well.
3) as I understand it, the levels in question were the audibility levels of frequencies above 20kHz. Assuming your hearing works at all above that threshold, you won't hear a thing until you push the volume up to insane levels for the rest of the spectrum. Which is why there is no point in worrying about some minor hypothetical impact at the output stage.
 

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
You seem to be out of ammo and are firing blank and silly statements.

That is your opinion. I've mine about your own inputs here :)

96kHz most likely, and it is NOT the spec. for the postfilter.

Why are you so focusing on chipset filters? Nagra and Hegel where annoucing up to 40k and 50kHz for DACs able to decode 192k PCM streams (think about the output lines buffers). Like solid state amplifiers are announced to have a bandwith of example 0 to 100k or limited at 20 to 20k.

1) Not any.
2) Connectivity may be an issue with a USB stick, and not everyone loves a mini jack to jack cable hanging from their main integrated amplifier/preamp. But yes, it works quite well.
3) as I understand it, the levels in question were the audibility levels of frequencies above 20kHz. Assuming your hearing works at all above that threshold, you won't hear a thing until you push the volume up to insane levels for the rest of the spectrum. Which is why there is no point in worrying about some minor hypothetical impact at the output stage.

House that USB stick in a 43cm box of your choice, it will then perfectly integrate with your amp and preamp.

Younger people shall be able to hear up to 20k and more. "The Mosquito or Mosquito alarm is a machine used to deter loitering by emitting sound at high frequency. In some versions, it is intentionally tuned to be heard primarily by younger people. The devices have attracted controversy on the basis of human rights and discrimination concerns." https://en.wikipedia.org/wiki/The_Mosquito

So if you cut off PCM 192k at 20kHz or close to, part of sound and hamonics are gone.

I'm wondering also if the output buffer (not the sharp filter of the conversion chipset) could impact the audio, by introducing phase shifting. With amplifiers or preamps specified for up to 40k to 100k, there shall be not much shifting in the audible range.

Anyway, I would personnaly appreciate to read more specs in the commercial DAC presentations, like:
- 20-20k (-0,1) for 44k (the sharp filter)
- 20-2...k (-0,1) for 48k (the sharp filter)
- 20-30k or 40k for PCM 88.2 to DSD rates (the buffers limit)
 
Last edited:

PierreV

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
1,437
Likes
4,686
Younger people shall be able to hear up to 20k and more. "The Mosquito or Mosquito alarm is a machine used to deter loitering by emitting sound at high frequency. In some versions, it is intentionally tuned to be heard primarily by younger people. The devices have attracted controversy on the basis of human rights and discrimination concerns." https://en.wikipedia.org/wiki/The_Mosquito

So if you cut off PCM 192k at 20kHz or close to, part of sound and hamonics are gone.

The newest version of the device, launched late in 2008, has two frequency settings, one of approximately 17.4 kHz[4] that can generally be heard only by young people, and another at 8 kHz that can be heard by most people. The maximum potential output sound pressure level is stated by the manufacturer to be 108 decibels (dB),

We have two choices now: either

1) you are a person who genuinely has trouble understanding information when it comes to you, has trouble reading more than one sentence in a paragraph, etc... In that case, I'd say: keep trying.

2) you are a poor troll. In that case, I'd say: stop trying.

Edit: tip 17.4kHz < 20 kHz
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Why are you so focusing on chipset filters?

I'll refer you to threads like these.
Have a look at specs from RME (one of the few correct and elaborate specs out there)

Does this address your concerns ?

Why are you so hung up on not so steep post filters at 70-100kHz and worry about noise, phase shifts and distortion at these frequencies
 
Last edited:

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
Have a look at specs from RME (one of the few correct and elaborate specs out there)

What ever the values are, those are more detailed and acceptable specs, aren't they?

When I bought my Aune S16, it was of the rare or only DACs on the market with more details about his capabilities. Aune had published some plots like those we see in ASR review (SNR, IMD...).

Why are you so hung up on not so steep post filters at 70-100kHz and worry about noise, phase shifts and distortion at these frequencies

I remain amazed by the fact that people aren't ever motivated to strongly prove that "audible" differences exist in the first place...

Before buying that Aune, I bougt two Fiio portables DACs. The entry range model for my daughter had much less bass in output (in headset, I suppose this is a very basic buffer/amplifier issue or design then price difference).

My second Fiio features "line out". Shall work as great as any USB stick. It is much better in bass using headset (better amp). But with either headset or line out, it sounds very bad for some precise tracks compared to the Aune S16 or now to the D50s.

I do still some time use that Fiio. It is upgraded to the latest software release. Recently, I plugged it to an amp, in a show room, to compare loud speakers with my own and known tracks. For many and most, that DAC plays very nice, it is usable. For some tracks and at precise times/signals, it sounds bad, distorded, tracks must be skipped.

That Fiio X5II FW2.1 S/N 60030123261009 is here on my desk. Since I own it, I do no more trust in basic specs and marketing materials, neither in all of what people say on the internet. Problem, Ariana Grande, is one of the tracks I have to skip with.
 
Last edited:

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
What ever the values are, those are more detailed and acceptable specs, aren't they?

I used the RME as an example of how the bandwidth changes per sampling rate. This is very similar for all other DACs that do more than 48kHz.
The fact that specs always say 20-20kHz for DACs is not because they are limited to 20kHz but because they are usually only specified at 44kHz.

You can also see that FR is not affected by the post filter (well above 80kHz) you can see some gentle roll-off that may be caused by the post filter.
Most manufacturers simply follow (copy) the recommended post filter settings from the chip manufacturer b.t.w.
Only some boutique DACs tend to wing it, use tubes or other exotic circuits just for the sake of being different (marketing)

I have a FiiO X5-II on my desk here as well. Never heard distortion and when I do it is in the file itself (only using line-out). I sometimes cannot believe how poorly some popular music is mastered. Not the player's fault. Also have an E30 and some other DACs and the same recordings sound equally crappy or good on all of them.
 

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
Only some boutique DACs tend to wing it, use tubes or other exotic circuits just for the sake of being different (marketing)

The market of Nagra: 50kHz with SNR of around -140dB.

By the way, RME seems some how limited to 100kHz (but we are here far out of audible range):
  • Frequency response @ 44.1 kHz, -0.1 dB: 0 Hz – 20.2 kHz
  • Frequency response @ 96 kHz, -0.5 dB: 0 Hz – 44.9 kHz
  • Frequency response @ 192 kHz, -1 dB: 0 Hz – 88 kHz
  • Frequency response @ 384 kHz, -1 dB: 0 Hz – 115 kHz
  • Frequency response @ 768 kHz, -3 dB: 0 Hz – 109 kHz


I have a FiiO X5-II on my desk here as well. Never heard distortion and when I do it is in the file itself (only using line-out). I sometimes cannot believe how poorly some popular music is mastered. Not the player's fault. Also have an E30 and some other DACs and the same recordings sound equally crappy or good on all of them.

I know how to distinguish bad tracks. I posted two examples in #71: https://www.audiosciencereview.com/...-khadas-tone-board-1.24596/page-4#post-830720

You say using line out my Fiio X5II unit sounds exact the same than my Aune and my Topping. It isn't the case for some tracks.

The track of Ariana Grande I mentionned is to be skipped with that/my Fiio, but is audible and acceptable with my Aune or Topping.
 
Last edited:

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
By the way, RME seems some how limited to 100kHz (but we are here far out of audible range):

That's the post filter causing this. As I mentioned this is usually set somewhere around 100kHz for DACs that reach higher bitrates.
How many studio microphones used for recording instruments do you know reach 100kHz -3dB ?
Do you think it will be of any audible consequence when the average audiophile can't hear above 14 to 16kHz (aside from a few very young ones)

You say using line out my Fiio X5II unit sounds exact the same than my Aune and my Topping. It isn't the case for some tracks.

Can you record the same song (the part that sounds bad) from both DACs so the rest of us can analyze and listen for themselves ?
 
Top Bottom