iFi iDSD Micro DSD512 / PCM768 DAC and Headphone Amp. Impressions, Reviews and Comments. | Page 204 | Headphone Reviews and Discussion - Head-Fi.org

earfonia · Mar 7, 2015 at 5:29 AM

Today I played around with the filter settings of iDSD micro, monitored by simple USB oscilloscope (Velleman PCSU200). So line output of the micro iDSD goes direct to USB oscilloscope input (1 Mohm input impedance).

Line Output setting: Direct
Power Mode: Normal
Polarity: +
RCA Cable: Wireworld Luna 7 (50 cm)
Adapter: Monster RCA to BNC

Here are the screenshots if you guys interested, showing only the left channel:

12 kHz Sine wave - PCM bit depth and sampling rate: 24 bit - 48 kHz

Bit-Perfect

Minimum Phase

Standard

12 kHz Sine wave - PCM bit depth and sampling rate: 24 bit - 192 kHz

Bit-Perfect

Minimum Phase

Standard

12 kHz Square wave - PCM bit depth and sampling rate: 24 bit - 192 kHz

Bit-Perfect

Minimum Phase

Standard

12 kHz Sawtooth wave - PCM bit depth and sampling rate: 24 bit - 192 kHz

Bit-Perfect

Minimum Phase

Standard

citraian · Mar 7, 2015 at 5:37 AM

Interesting. What about all three waves at 24/41?
Also interested in the freq response. Is this posted somewhere?

earfonia · Mar 7, 2015 at 5:41 AM

citraian said:
Interesting. What about all three waves at 24/41?
Also interested in the freq response. Is this posted somewhere?

Sorry, I'm lazy to redo all the test in 24/41

I only posted it here.

GrahamL · Mar 7, 2015 at 6:23 AM

karistep said:
Did you try this one
http://ifi-audio.com/3-steps-micro-idsd-firmware-upgrade/

Thanks for the reply. Mine's a new unit; the version you linked (v4.06) is already installed.

v4.06 fixes the pop in some use cases, but not mine unfortunately. The changelog for the latest beta (v0.16) says "Fix completely clicks when switching between PCM and DoP DSD" which I can't wait to try.

If it works well then my preferred method of delivering DSD to my iOS devices at home (DoP ALAC served by iTunes Home Sharing) should hopefully work out nicely. It works now, apart from the damn clicks between tracks

Fingers crossed

maricius · Mar 7, 2015 at 7:21 AM

iFi did say BitPerfect with the ****tiest sine wave would have the best impulse response.

justabanana · Mar 7, 2015 at 9:03 PM

Hi ClieOS , so by having iDSD , A15 are able to play DSD audio file simply by adding the file inside A15 ? Or is there setting that you need to adjust ?

ClieOS · Mar 7, 2015 at 10:16 PM

justabanana said:
Hi ClieOS , so by having iDSD , A15 are able to play DSD audio file simply by adding the file inside A15 ? Or is there setting that you need to adjust ?

No, unfortunately. Although iDSD can play DSD files, A15 can't. It won't even recognize the files. You will have to stick to PCM.

iFi audio · Mar 9, 2015 at 5:57 AM

earfonia said:
Today I played around with the filter settings of iDSD micro, monitored by simple USB oscilloscope (Velleman PCSU200). So line output of the micro iDSD goes direct to USB oscilloscope input (1 Mohm input impedance).

Line Output setting: Direct
Power Mode: Normal
Polarity: +
RCA Cable: Wireworld Luna 7 (50 cm)
Adapter: Monster RCA to BNC

Here are the screenshots if you guys interested, showing only the left channel:

12 kHz Sine wave - PCM bit depth and sampling rate: 24 bit - 48 kHz

Bit-Perfect

Minimum Phase

Standard

12 kHz Sine wave - PCM bit depth and sampling rate: 24 bit - 192 kHz

Bit-Perfect

Minimum Phase

Standard

12 kHz Square wave - PCM bit depth and sampling rate: 24 bit - 192 kHz

Bit-Perfect

Minimum Phase

Standard

12 kHz Sawtooth wave - PCM bit depth and sampling rate: 24 bit - 192 kHz

Bit-Perfect

Minimum Phase

Standard

Hi,

Here is some background information providing you all with an overview of the what these charts mean.

These waveforms show ow the iDSD micro reacts to input data with the different filters.

The Bit-Perfect Filter, 12kHz at 48kHz sample rate illustrates exactly what data can be acommodatd by the format. A 48kHz sample rate only has 4 samples available for a 12kHz sinewave, so there are only four datapoints available for the waveform and we see these four steps very clearly.

The other (digital) filters introduce varying amounts of "ringing¹". This smooth out the steady state sinewave. Increasing the sample rate also helps smooth the waveform.

This can be seen with the squarewave and triangular waveforms. The distortion of the waveform with the different digital filters is timedomain distortion, as opposed to amplitude domain distortion on the sinewave cases.

Clearly, the better a filter performs on sinewaves, especially at a low ratio between sample rate and signal, the poorer it performs with square/triangle waves.

It takes very high sample rates (192kHz or higher) to make the difference small enough² to no longer require any compromise between amplitude domain and time domain performance.

At lower sample rates there is no absolute right or wrong answer, any answer is "wrong" somewhere. What works best depends on system context, music, recording and not the least personal taste.

As there is NO such thing as a "perfect" filter (and don't let anyone tell you there is - if they do they are either ignorant of the real world facts on this or they are telling porkies).

AMR/iFi prefers to offer several filters that offer different tradeoffs, so our customers may choose what suits them "best".

¹What is Ringing?: http://amr-audio.co.uk/html/dp777_tech-papers_ringing.html
² For example the Sony Hi-Res Walkman app plays DSD but it actually converts DSD > PCM 176khz so this still sounds very satisfying with no obvious loss.

diamondears · Mar 9, 2015 at 6:31 AM

ifi audio said:
Hi,

Here is some background information providing you all with an overview of the what these charts mean.

These waveforms show ow the iDSD micro reacts to input data with the different filters.

The Bit-Perfect Filter, 12kHz at 48kHz sample rate illustrates exactly what data can be acommodatd by the format. A 48kHz sample rate only has 4 samples available for a 12kHz sinewave, so there are only four datapoints available for the waveform and we see these four steps very clearly.

The other (digital) filters introduce varying amounts of "ringing¹". This smooth out the steady state sinewave. Increasing the sample rate also helps smooth the waveform.

This can be seen with the squarewave and triangular waveforms. The distortion of the waveform with the different digital filters is timedomain distortion, as opposed to amplitude domain distortion on the sinewave cases.

Clearly, the better a filter performs on sinewaves, especially at a low ratio between sample rate and signal, the poorer it performs with square/triangle waves.

It takes very high sample rates (192kHz or higher) to make the difference small enough² to no longer require any compromise between amplitude domain and time domain performance.

At lower sample rates there is no absolute right or wrong answer, any answer is "wrong" somewhere. What works best depends on system context, music, recording and not the least personal taste.

As there is NO such thing as a "perfect" filter (and don't let anyone tell you there is - if they do they are either ignorant of the real world facts on this or they are telling porkies).

AMR/iFi prefers to offer several filters that offer different tradeoffs, so our customers may choose what suits them "best".

¹What is Ringing?: http://amr-audio.co.uk/html/dp777_tech-papers_ringing.html
²For example the Sony Hi-Res Walkman app plays DSD but it actually converts DSD > PCM 176khz so this still sounds very satisfying with no obvious loss.

Hi iFi,

I know that Minimum Phase has time domain distortion as trade off for removal of pre-ringing (but post-ringing still exists); while Standard Filter has both pre and post ringing but has no time domain and frequency distortion. This correct? Correct me if I'm wrong.

How about the Bit-Perfect Filter? No pre and post ringings, but has lots of time domain distortion?

Don't get the graphs. Doesn't the sine waves show time domain distortion a, while square and triangle/sawtooth waves show amplitude or frequency distortion?

Cheers.

plakat · Mar 9, 2015 at 6:35 AM

ifi audio said:
AMR/iFi prefers to offer several filters that offer different tradeoffs, so our customers may choose what suits them "best".

Yes, its always 'pick your poison' -- and people tend to have different views of what sounds good / correct, so having a choice is not a bad thing. And in the end the only thing that matters (at least to me) is how much it lets me enjoy music. Which is a bit hindered at the moment as I forgot the USB cable + CCK at home... but thats not a fault of the product of course :wink:

earfonia · Mar 9, 2015 at 9:22 AM

ifi audio said:
Hi,

Here is some background information providing you all with an overview of the what these charts mean.

These waveforms show ow the iDSD micro reacts to input data with the different filters.

The Bit-Perfect Filter, 12kHz at 48kHz sample rate illustrates exactly what data can be acommodatd by the format. A 48kHz sample rate only has 4 samples available for a 12kHz sinewave, so there are only four datapoints available for the waveform and we see these four steps very clearly.

The other (digital) filters introduce varying amounts of "ringing¹". This smooth out the steady state sinewave. Increasing the sample rate also helps smooth the waveform.

This can be seen with the squarewave and triangular waveforms. The distortion of the waveform with the different digital filters is timedomain distortion, as opposed to amplitude domain distortion on the sinewave cases.

Clearly, the better a filter performs on sinewaves, especially at a low ratio between sample rate and signal, the poorer it performs with square/triangle waves.

It takes very high sample rates (192kHz or higher) to make the difference small enough² to no longer require any compromise between amplitude domain and time domain performance.

At lower sample rates there is no absolute right or wrong answer, any answer is "wrong" somewhere. What works best depends on system context, music, recording and not the least personal taste.

As there is NO such thing as a "perfect" filter (and don't let anyone tell you there is - if they do they are either ignorant of the real world facts on this or they are telling porkies).

AMR/iFi prefers to offer several filters that offer different tradeoffs, so our customers may choose what suits them "best".

¹What is Ringing?: http://amr-audio.co.uk/html/dp777_tech-papers_ringing.html
²For example the Sony Hi-Res Walkman app plays DSD but it actually converts DSD > PCM 176khz so this still sounds very satisfying with no obvious loss.

Thanks! I know that. To me all looks good and as expected.
I really appreciate ifi to have included the 'bit perfect' filter which is to me looks like 'no filter', so I can use iDSD micro not only for audio, but also for signal generator, generating many types of waveform without low pass filter. Useful especially for generating low frequency square waves.

One more thing, also thanks a lot for the DC coupled line output, exactly what I want

1 Hz square wave:

What I wish to be improved is the output tolerance of the line output. There is slight imbalance between left and right channel of the line output, even when the line output switch is set to 'Direct':

Channel 1: Left Channel
Channel 2: Right Channel

1.0 Full scale

0.75 Full scale

0.5 Full scale

0.25 Full scale

At average of 0.26 dB different between left and right channel, it is still considered low, and not something alarming, but I wish the tolerance could be tighten to 0.1 dB. Probably on iDSD Pro

john57 · Mar 9, 2015 at 9:41 AM

Plus we do not know how well the calibration or accuracy between the two channels on the scope measuring voltage.

earfonia · Mar 9, 2015 at 9:49 AM

diamondears said:
How about the Bit-Perfect Filter? No pre and post ringings, but has lots of time domain distortion?

What shown from the first graph, the 12kHz sine wave from 24bit/48kHz PCM, is not a distortion. The DAC faithfully reconstructs the data from the PCM file. What shown is exactly the information that is stored in 24bit/48kHz PCM.
24bit/48kHz PCM simply doesn't have enough capacity to store better resolution of the 12 kHz sine wave, as compared to higher resolution such as the 24bit/192kHz PCM or higher.
But the graph doesn't show what 'we want to see / hear'. We want smoother sine wave beyond what 24bit/48kHz PCM is capable of. This is where low pass filter plays its part, to smoothen the jagged edges to make better looking sine wave. But as explained by ifi, there is no such thing as perfect filter. At least as of now. So low pass filter has it's own pros and cons. Everything is a compromise. Depending on the purpose, all the available filters are very useful if we understand when to use them. For standard bit rate, 44.1 and 48 kHz, or music in general, I stick with 'Minimum Phase'. When using iDSD micro for other measurement purpose, or playing DXD files, I use 'Bit Perfect'. I use 'Standard' filter only when I hear the recording sounds harsh or too bright. I really like to have all the filters option available. Thanks ifi!

diamondears · Mar 9, 2015 at 9:59 AM

earfonia said:
What shown from the first graph, the 12kHz sine wave from 24bit/48kHz PCM, is not a distortion. The DAC faithfully reconstructs the data from the PCM file. What shown is exactly the information that is stored in 24bit/48kHz PCM.
24bit/48kHz PCM simply doesn't have enough capacity to store better resolution of the 12 kHz sine wave, as compared to higher resolution such as the 24bit/192kHz PCM or higher.
But the graph doesn't show what 'we want to see / hear'. We want smoother sine wave beyond what 24bit/48kHz PCM is capable of. This is where low pass filter plays its part, to smoothen the jagged edges to make better looking sine wave. But as explained by ifi, there is no such thing as perfect filter. At least as of now. So low pass filter has it's own pros and cons. Everything is a compromise. Depending on the purpose, all the available filters are very useful if we understand when to use them. For standard bit rate, 44.1 and 48 kHz, or music in general, I stick with 'Minimum Phase'. When using iDSD micro for other measurement purpose, or playing DXD files, I use 'Bit Perfect'. I use 'Standard' filter only when I hear the recording sounds harsh or too bright. I really like to have all the filters option available. Thanks ifi!

Apologies, still don't get it totally. What are the actual pros and cons of each filter in terms of time domain distortion and amplitude/frequency response distortion? I mean, the trade offs?

earfonia · Mar 9, 2015 at 10:01 AM

john57 said:
Plus we do not know how well the calibration or accuracy between the two channels on the scope measuring voltage.

Very good point!

Here is the graph of the left channel, split to both channel1 and channel2. So same source for both channels. Total cable length about 1 feet.

The scope I use is just a cheap and simple USB scope, and shouldn't be considered accurate. But so far quite useful.
I've bought better one last week, on the way from Amazon

Latest Thread Images

earfonia

Headphoneus Supremus

citraian

1000+ Head-Fier

earfonia

Headphoneus Supremus

GrahamL

New Head-Fier

maricius

500+ Head-Fier

justabanana

New Head-Fier

ClieOS

IEM Reviewer Extraordinaire

iFi audio

Sponsor: iFi audio

diamondears

Headphoneus Supremus

plakat

Headphoneus Supremus

earfonia

Headphoneus Supremus

john57

Headphoneus Supremus

earfonia

Headphoneus Supremus

diamondears

Headphoneus Supremus

earfonia

Headphoneus Supremus

Users who are viewing this thread