Best budget DAC for measuring?

[keantoken]

Hi everyone. I was looking at the Khadas Toneboard for doing measurements as I am designing circuits in the <0.001% THD range. The noise floor can be beaten into submission by averaging and longer measurement times, so I'm not as worried about THD+N.

There were some comments about a comparable Topping DAC measuring better.

Galvanic isolation is crucial for making good measurements. Does the toneboard have any kind of isolation on the USB port or the SPDIF input?

I saw some distortion vs frequency charts which showed higher THD at high frequencies with the Toneboard, but as far as I could tell the measurement bandwidth was a confounder and I didn't see that I could compare that chart with any of the other available ones.

What do you think would be the best measurement DAC in the <$100 price bracket?

 

[MZKM]

saw some distortion vs frequency charts which showed higher THD at high frequencies with the Toneboard

Disregarding the 90kHz bandwidth, it’s still <0.01%. Most all human trials show that audibility thresholds for THD+N when listening to music needs to be around 1% in the upper treble usually.

Now, THD+N does stack across components in your chain. Let’s say you had this DAC, a preamp, an amp, and a DSP device in-between. For 4 components to have 1% THD, they would individually need to have an average of ~0.25% THD+N if I’m not mistaken. Adding the ToneBoard into the equation (using 0.01%, even though it’s lower), the other 3 components would need an average of ~0.315% THD+N to add up to 1%, and again, this is for the upper treble.

 

[daftcombo]

Disregarding the 90kHz bandwidth, it’s still <0.01%. Most all human trials show that audibility thresholds for THD+N when listening to music needs to be around 1% in the upper treble usually.

Now, THD+N does stack across components in your chain. Let’s say you had this DAC, a preamp, an amp, and a DSP device in-between. For 4 components to have 1% THD, they would individually need to have an average of ~0.25% THD+N if I’m not mistaken. Adding the ToneBoard into the equation (using 0.01%, even though it’s lower), the other 3 components would need an average of ~0.315% THD+N to add up to 1%, and again, this is for the upper treble.

There's a lot of simplification here, in my opinion. First, it would depend on where the THD+N peaks of those components are. Second, if you have 100% THD+N on all components, you won't have 400% in the chain. It cannot be a simple addition.
Finally, I don't get your addition wih the Khadas. Why 0.315% and not 0.33% ?

 

[somebodyelse]

Galvanic isolation is crucial for making good measurements. Does the toneboard have any kind of isolation on the USB port or the SPDIF input?

Khadas publish hardware docs including the schematic, so you can check this and any other circuit questions. In short, no isolation on either.

 

[MZKM]

There's a lot of simplification here, in my opinion. First, it would depend on where the THD+N peaks of those components are. Second, if you have 100% THD+N on all components, you won't have 400% in the chain. It cannot be a simple addition.
Finally, I don't get your addition wih the Khadas. Why 0.315% and not 0.33% ?

It does indeed matter where the distortion is happening, but the majority of the time it is in the treble (except for speakers).

The formula is 20*log10(10^(-x/20) + 10^(-x/20) + ...), where "x" is the dBFS of distortion/Signal-to-Noise/crosstalk/etc..

So, if you had 4 components with -40dB (1%) at the same frequency, it would be:

20*log10(4*10^(-40/20)) = ~-30dB (~3.16%).

 

[keantoken]

Thank you for the replies, looks like I didn't get any notifications or they all went in my spam folder.

There are two mathematical derivations of THD, one can result in THD over 100% and the other only approaches 100%. Most of the time in audio we don't care which is used as our system never approaches 100% THD.

I'm not really interested in the THD+N figure as I can remove the N with averaging and high FFT size. Likewise if the treble distortion is not caused by harmonics but by digital noise, I can filter that out too. So ultimately it is a question of whether the Khadas is worse in the treble than a similarly priced Topping or other DAC, and whether I can get acceptable performance in a model that also has galvanic isolation.

 

[MZKM]

So ultimately it is a question of whether the Khadas is worse in the treble than a similarly priced Topping or other DAC

<0.005% THD+N below ~18kHz is amazing for $100.

As a reference, the $250 SMSL SU-8 & Topping D50:

Don’t know how the D10 or Schiit Modi 3 performs, but you can see that if you Anna spend $250, then you do get an improvement.

 

[JohnYang1997]

I think I'm experienced to answer this question.

Since thd is what you need and noise is not a concern (as long as you can see harmonics)

Focusrite forte is awesome. (Both adc and dac) Use output of -18db. Usually you can measure 0.00007% thd loop. It cost around 250 dollars. (Second hand)

SMSL M3 works too.(dac only) Use -12db to -15db will give you less than 0.0001% thd. 100 dollars. New

Asus u7 works too(both adc and dac) if you use headphone out low gain and very low output. You can get 0.0002% loop thd. This is also around 100 dollars. New

You can easily design a voltage amplifier and attenuator to get this to work.

 

[keantoken]

Thanks, that is very useful information.

 

[andreasmaaan]

@keantoken wouldn't it be best to get an AD/DA rather than just a DA? Kill two birds with one stone? Or do you already have an excellent-measuring AD and are now just looking for a DA to produce the test signals?

Focusrite forte is awesome. (Both adc and dac) Use output of -18db. Usually you can measure 0.00007% thd loop. It cost around 250 dollars.

Do you have a link please? List price is US$750 AFAIK.

From the Focusrite line, the Clarett 2Pre has slightly better specs than the Forte and (unless I'm mistaken about the Forte's price!) costs a little less.

In general, I'd say your best bet would be with Focusrite, RME, or Motu. All 3 manufacturers give specs that can be trusted IME, so you can look at prices and spec sheets, and with quite some confidence make your decision on that basis.

 

[JohnYang1997]

@keantoken wouldn't it be best to get an AD/DA rather than just a DA? Kill two birds with one stone? Or do you already have an excellent-measuring AD and are now just looking for a DA to produce the test signals?



Do you have a link please? List price is US$750 AFAIK.

From the Focusrite line, the Clarett 2Pre has slightly better specs than the Forte and (unless I'm mistaken about the Forte's price!) costs a little less.

In general, I'd say your best bet would be with Focusrite, RME, or Motu. All 3 manufacturers give specs that can be trusted IME, so you can look at prices and spec sheets, and with quite some confidence make your decision on that basis.

Second hand. Sorry forgot to mention that.
And forte measures better than 2pre. I tried 2pre not as good as forte. If you want to buy a new one then you can get 2pre. But not as competitive to qa401 you can measure pretty low thd with qa401 at low level too.

 

[andreasmaaan]

And forte measures better than 2pre.

Do you have a source for that? I haven't seen measurements of the Pre2 but it's spec'd slightly better than the Forte by the manufacturer.

For example:

• Line input THD: <0.003% (Forte) vs 0.001% (Pre2)
• Dynamic Range ADC: 117dB (Forte) vs 119dB (Pre2)
• THD Line Outputs: <0.0008% (Forte) vs 0.00075 (Pre2)

And so on and so on, i.e. on virtually all metrics the Pre2 is spec'd slightly better than the Forte.

I tried 2pre not as good as forte.

Could you explain why the Forte is better in your opinion?

 

[JohnYang1997]

Do you have a source for that? I haven't seen measurements of the Pre2 but it's spec'd slightly better than the Forte by the manufacturer.

For example:

• Line input THD: <0.003% (Forte) vs 0.001% (Pre2)
• Dynamic Range ADC: 117dB (Forte) vs 119dB (Pre2)
• THD Line Outputs: <0.0008% (Forte) vs 0.00075 (Pre2)

And so on and so on, i.e. on virtually all metrics the Pre2 is spec'd slightly better than the Forte.



Could you explain why the Forte is better in your opinion?

Simply provides lower thd in loop measurements. The better spec is provided by 2pre is under two conditions: Noise included and at max volume(critical for best thd+noise). Noise can be easily measure with a voltage amplifier and as the op directly specified thd no n. That's the better choice.

More information about adc, forte uses same adc as hilo which performs the same as adi2pro, which uses higher end akm adc than 2pre. Manufacturer's spec sheet don't always tell the whole story. Especially if you are asking for something slightly different from the numbers provided.

 

[andreasmaaan]

Simply provides lower thd in loop measurements.

Ok, but which measurements?

The better spec is provided by 2pre is under two conditions: Noise included and at max volume(critical for best thd+noise).

And the Forte is spec'd differently?

More information about adc, forte uses same adc as hilo which performs the same as adi2pro, which uses higher end akm adc than 2pre.

To my knoweldge, the 2Pre uses the AK5388, which if I'm not mistaken is the same ADC chip used in the Forte. Am I mistaken?

 

[JohnYang1997]

Ok, but which measurements?



And the Forte is spec'd differently?



To my knoweldge, the 2Pre uses the AK5388, which if I'm not mistaken is the same ADC chip used in the Forte. Am I mistaken?

Measurements are done my me like more than a year ago.
And forte uses cs5381 in the adc. I have two fortes.
The specs are thd+n at max level. But we want best thd no n at whatever level. So that's not representative for the measurements we want.

 

[andreasmaaan]

And forte uses cs5381 in the adc. I have two fortes.

That's a Cirrus Logic chip. You said in your previous post:

More information about adc, forte uses same adc as hilo which performs the same as adi2pro, which uses higher end akm adc than 2pre.

Measurements are done my me like more than a year ago.

That clears it up

 

[Blumlein 88]

That's a Cirrus Logic chip. You said in your previous post:







That clears it up

The Forte uses a CS5381 for ADC and CS4398 for DAC.

Focusrite claimed the Claret line used more or less the same hardware as the Forte, but I don't know if that is true. The specs are close and there are slight differences in gain amounts and output levels. I own a Forte, but haven't had hands on a Claret. Two things I don't like about the Forte for measurement is there is an idle tone around the mid 30 khz range, and it has a little low level aliasing oddly reflected around that 30 khz tone and not the sample rate. But for $200-250 2nd hand, it is very good, and can certainly be used for measuring most things to good effect. The really good thing if you do recording is the mic pre's seem unusually nice to me and have 75 db of gain with very good EIN specs.

 

[JohnYang1997]

The Forte uses a CS5381 for ADC and CS4398 for DAC.

Focusrite claimed the Claret line used more or less the same hardware as the Forte, but I don't know if that is true. The specs are close and there are slight differences in gain amounts and output levels. I own a Forte, but haven't had hands on a Claret. Two things I don't like about the Forte for measurement is there is an idle tone around the mid 30 khz range, and it has a little low level aliasing oddly reflected around that 30 khz tone and not the sample rate. But for $200-250 2nd hand, it is pretty good, and can certainly be used for measuring most things to good effect. The really good thing if you do recording is the mic pre's seem unusually nice to me and have 75 db of gain with very good EIN specs.

They are at same level. But cs5381 is just a hidden gem(not really hidden because many high end studio equipments use it.), it has the performance hidden under the thd+n number that's my opinion. (same with cs4398)

 

[keantoken]

Having the AD and DA on the same clock is useful for some things, but also has it's own limitations. My AD doesn't get below 0.0002% THD (at least I don't think so), so a DAC with a good integrated AD would get me closer to my goal. However if the DA and AD are not galvanically isolated there could be positive feedback problems.

 

[JohnYang1997]

Having the AD and DA on the same clock is useful for some things, but also has it's own limitations. My AD doesn't get below 0.0002% THD (at least I don't think so), so a DAC with a good integrated AD would get me closer to my goal. However if the DA and AD are not galvanically isolated there could be positive feedback problems.

SMSL M3 with -6 to -12dbfs should do the job.