CHORD ELECTRONICS DAVE

[ObjectiveSubjectivist]

As I said it later:

I would .9db @20khz is not that much especially when you consider music usually contains nothing that high.
But as you said it's definitely audible (but not always, it depends on spectrum of the music that you listen to) as it's broad "roll off".

It's broad so it is audible (very minor note it) but not with every content. Depends on music spectrum.
But I wanted to make it clear, that indeed HF filter affects audible band.

 

[manisandher]

Actually it was measured that HF filter affects audible range, it's tiny yes but still.

That's simply the FR of the filter. My null tests show that the HF filter has a (quite large) detrimental effect throughout the audioband, due to things like phase shift (that do not affect the FR). And it's not possible to 'cheat' these null tests - you're comparing what's coming out of the analogue output of the DAC directly with what's going in. It's the 'gold standard', as far as I'm concerned... but really difficult to do well - probably why they're not more commonly employed.

I'll post the results for the DAVE once I'm back at home...

Mani.

 

[ObjectiveSubjectivist]

That's simply the FR of the filter. My null tests show that the HF filter has a (quite large) detrimental effect throughout the audioband, due to things like phase shift (that do not affect the FR). And it's not possible to 'cheat' these null tests - you're comparing what's coming out of the analogue output of the DAC directly with what's going in. It's the 'gold standard', as far as I'm concerned... but really difficult to do well - probably why they're not more commonly employed.

I'll post the results for the DAVE once I'm back at home...

Mani.

I would love to see your tests with deltawave, could be interesting.
Yes I was just talking about FR as it is something obvious, I did not try to do null test or any other analysis.

 

[GuiltyRocker]

That's simply the FR of the filter. My null tests show that the HF filter has a (quite large) detrimental effect throughout the audioband, due to things like phase shift (that do not affect the FR). And it's not possible to 'cheat' these null tests - you're comparing what's coming out of the analogue output of the DAC directly with what's going in. It's the 'gold standard', as far as I'm concerned... but really difficult to do well - probably why they're not more commonly employed.

I'll post the results for the DAVE once I'm back at home...

Mani.

That I want to see, I don't think @Rob Watts would design a filter to be that "detrimental" creating phase shift, I don't know what kind of testing you're doing, in the past when I activated the HF filter on DAVE I noticed a slight smoothness to the sound, that's about it.

 

[manisandher]

That I want to see...

Will try to post later today.

I don't know what kind of testing you're doing...

You can learn about it here: https://deltaw.org/

Null testing is the 'gold standard' IMO - but a dog to do well.

Edit:

Here's the test setup:

And here are the null results for the DAVE, without and with HF filter engaged, respectively:

(These results are totally repeatable between multiple samples taken over different times and different days.)

It's clear that the (negative) effects of the HF filter extend well down below 1kHz. Furthermore, the 'Δ Waveform' plots in DeltaWave (not shared here) suggest that it's the transients that mainly take the hit - such a shame, as this is where the DAVE generally shines vs. the other DACs I've tested.

(But I was wrong about the filter causing extra phase shifts - DeltaWave suggests this is actually not the case.)

I have the HF filter switched off. It's no skin off my nose if you prefer the sound with it - but make no mistake, it's less 'accurate' to the original sound.

Mani.

 

[559938]

That's simply the FR of the filter. My null tests show that the HF filter has a (quite large) detrimental effect throughout the audioband, due to things like phase shift (that do not affect the FR). And it's not possible to 'cheat' these null tests - you're comparing what's coming out of the analogue output of the DAC directly with what's going in. It's the 'gold standard', as far as I'm concerned... but really difficult to do well - probably why they're not more commonly employed.

I'll post the results for the DAVE once I'm back at home...

Mani.

Forgive my ignorance, but just how do you compare what’s coming out (analogue) with what went in (digital) for a null test?
From the diagram it looks like you run DAVE analogue output into an RME ADC then compare that to the original. Is that right? How do you know I what the DAVE is altering compared to what the RME is altering?
Cheers

 

[manisandher]

Forgive my ignorance, but just how do you compare what’s coming out (analogue) with what went in (digital) for a null test?
From the diagram it looks like you run DAVE analogue output into an RME ADC then compare that to the original. Is that right? How do you know I what the DAVE is altering compared to what the RME is altering?
Cheers

Yes, it's the delta of the 'reference' file to the 'compare' file. The former is the original file. The latter, the file created by the DAC-ADC chain.

The quality of the nulls is massively affected by:
1. the quality of the ADC (I'm using the best-measuring currently available - until perhaps Davina? )
2. the accuracy of matching the two waveforms in time and amplitude

But in this case, it's DAVE vs. DAVE, so many things cancel out.

Edit: Here's what I mean by 'cancel out':

The repeatability of the results suggests that the differences that I showed between the HF filter off/on are accurate.

Mani.

 

[Jawed]

Does the latency for HF filter on equal the latency for HF filter off?

 

[GuiltyRocker]

Yes, it's the delta of the 'reference' file to the 'compare' file. The former is the original file. The latter, the file created by the DAC-ADC chain.

The quality of the nulls is massively affected by:
1. the quality of the ADC (I'm using the best-measuring currently available - until perhaps Davina? )
2. the accuracy of matching the two waveforms in time and amplitude

But in this case, it's DAVE vs. DAVE, so many things cancel out.

Edit: Here's what I mean by 'cancel out':



The repeatability of the results suggests that the differences that I showed between the HF filter off/on are accurate.

Mani.

@Rob Watts what do you think about this?

 

[559938]

Yes, it's the delta of the 'reference' file to the 'compare' file. The former is the original file. The latter, the file created by the DAC-ADC chain.

The quality of the nulls is massively affected by:
1. the quality of the ADC (I'm using the best-measuring currently available - until perhaps Davina? )
2. the accuracy of matching the two waveforms in time and amplitude

But in this case, it's DAVE vs. DAVE, so many things cancel out.

Edit: Here's what I mean by 'cancel out':



The repeatability of the results suggests that the differences that I showed between the HF filter off/on are accurate.

Mani.

A few too many unknowns and variables for me.

 

[manisandher]

A few too many unknowns and variables for me.

Cool. Go with whichever HF setting you prefer.

For those of you who prefer HF off, you now know why - because it's measurably more accurate to the source .

Mani.

 

[559938]

Cool. Go with whichever HF setting you prefer.

For those of you who prefer HF off, you now know why - because it's measurably more accurate to the source .

Mani.

I’m just not sure what impact the ADC is having on the results. Indeed, the DAVE could well be a more accurate device than the RME.

 

[Rob Watts]

@Rob Watts what do you think about this?

The HF filter changes the original data within the audio bandwidth, so it will certainly 'fail' on a null test. But that doesn't mean it fails perceptively. To give an example - I tested the (note not the same filter as Dave) IIR bandwidth limiting filter with one of my decimation filters for the ADC, which eliminates everything above 20k, but is completely flat with zero phase shift up to 20k. With the decimation filter in place, you could hear the effects of the IIR filter easily, as usual. But with the decimation filter working, you could hear no difference at all. This proved that the phase and amplitude shifts of the IIR filter (which are significant in the audio bandwidth) are not perceptively important. This proved that a -1dB or so at 20k with attendant phase shifts are inaudible (at least to my old ears) and that the SQ changes of bandwidth limiting filters are due to reducing out of band noise above 50k and hence reducing noise floor modulation - which is at unmeasurable levels either way.

 

[manisandher]

The HF filter changes the original data within the audio bandwidth, so it will certainly 'fail' on a null test. But that doesn't mean it fails perceptively. To give an example - I tested the (note not the same filter as Dave) IIR bandwidth limiting filter with one of my decimation filters for the ADC, which eliminates everything above 20k, but is completely flat with zero phase shift up to 20k. With the decimation filter in place, you could hear the effects of the IIR filter easily, as usual. But with the decimation filter working, you could hear no difference at all. This proved that the phase and amplitude shifts of the IIR filter (which are significant in the audio bandwidth) are not perceptively important. This proved that a -1dB or so at 20k with attendant phase shifts are inaudible (at least to my old ears) and that the SQ changes of bandwidth limiting filters are due to reducing out of band noise above 50k and hence reducing noise floor modulation - which is at unmeasurable levels either way.

With all due respect (meant absolutely sincerely), to my ears, the HF filter does 'fail perceptively'.

Furthermore, the 'Δ Waveform' plots in DeltaWave show that transients in the music are massively effected with the filter ON. Other testing methods just won't reveal this.

I much prefer the sound of the DAVE with the filter OFF, but if others like it ON, that's totally cool.

Mani.

 

[Jawed]

Furthermore, the 'Δ Waveform' plots in DeltaWave show that transients in the music are massively effected with the filter ON. Other testing methods just won't reveal this.

Can you show an example of the difference that these plots show?