MDAC First Listen (part 00110010)

[DANOFDANGER]

So can this be made clear to me. Since the oversampling modulator removes aliasing as you say, but not all of it. Then the digital LPF is mostly for noise just outside the passband due to noise shaping. So its not or mostly not a anti imaging filter but more just a plain old LPF for noise.
Why is this noise important to filter out ? Does it impact the passband ?

EDIT.
Further research as far as i can tell shows that the oversampling modulator preceding the SD modulator only shifts the spectral images further up the frequency band. Oversampling cannot remove "spectral images" but rather move them somewhere else or possibly transform their energy.
The answer to my question appears to be that the Digital LPF handles only the Quantization noise from the SD moduation and anti aliasing\anti imaging is still handled by the analog filters.

Still yet to figure out why we need to filter the Quantization noise outside the audible bandwidth but i assume it must be because it affects the passband somehow.

 

[dtd]

Being frankly honest, I would put money on it that most peoples existing system setup (speakers, amplifiers, pre-amps, room accoustic, positioning, reflection points etc) will probably not take full advantage of the FDAC prowess without its ADC input, and likely be very happy with the MDAC2 (ignoring bigger better deal illnesses etc).

When you get to that kind of finese at the source, the rest of the chain also has to have care and attention given to it to even hear the differences.

 

[JohnW]

I quite agree that it's a good idea to get something finished. But this is the first time I can remember it has been suggested that the F dac would not be complete this year. Next year sounds a long long way away.

I think that it would be sensible to create a proper real achievable strict road map for the Fdac and to allow people the option to transfer the payments made for the FDac into mdac2s. Maybe the mdac 2 just is the mdac replacement we've always wanted. I was excited at the idea of the complete ADC\dac phono streamer but maybe I'll have to settle for less.

Adam,

I caught between an uncertain delivery time on the FDAC and MDAC2 which I can be certain to deliver within the next 2-3 months - with only the rear panel to worry about (mechanically) - and PCB production here in Europe its a far simpler project.

I personally will be still working on the FDAC design - so it will follow ASAP, but for sure MDAC2 can be first.

For those who chose the MDAC2 now they can still have the FDAC at a reduced development fee - say 50%...

The MDAC2 will have a streamer option - but the hardware will be delivered to owners BEFORE we have completed the software (Volumio) for the streamer - as I don't want to delay the MDAC2 delivery.

The MDAC2 :-

Dual ES9038 "Mobile's"

FDAC / "Superior" MOSFET ClassA output - discrete Gain stage

Balanced Headphone output

Isolated USB to 32KHz 768KHz, and DSD upto 512

Arm A7 Quad core Streamer Option, with HDMI video output

Reuses the MDAC Chassis - with new external PSU

Prototypes built mid October ~ production a month afterwards (dependent on component sourcing).

We have working "pre-prototypes" using Fusions "Superior" PCB & our XMOS development board currently working on the bench...

THD 0.00009%

Dynamic range 131dB Awtd

On the MDAC2 we will reuse the MDAC front panel so no Colour screen as on our small XMOS DAC PCB (but the MDAC's a larger screen).

https://dl.dropboxusercontent.com/u/86116171/Fusions MDAC2 proto1.JPG

https://dl.dropboxusercontent.com/u/86116171/Fusions MDAC2 Proto2.JPG

Jarek will issue the PCB during my China trip for Detox and FDAC chassis etc. so its effectively running in parallel with FDAC.

 

[highfi]

John what is the status of my MDAC repair please?

 

[JohnW]

highfi,

I'll ship your unit at the end of the week.

 

[janerik]

The MDAC2 :-

Dual ES9038 "Mobile's".

Hello John,

Are these the same as the new ES9038Pro DACs ?

Thanks,

Jan

 

[JohnW]

Yes, they are the new Hyperstream II devices

Really, it will be a VERY VERY good dac

 

[DANOFDANGER]

Yes, Correct.



Again correct, the OT "Filters" are simply sample and hold... and as such introduces no Digital domain* pre or post ringing on the playback side. * the analogue domain has a small amount of overshoot / undershoot.

https://dl.dropboxusercontent.com/u/86116171/MDAC Optimal Transient.jpg



There is no clear differentiation between minimum phase and apodizing filters - apodizing filters are more marketing speak and just a variation on minimum phase. (as far as I understand the apodizing filter achieves its "full" digital attenuation at FS/2 (as Shannon theory demands) rather then just 6dB of typically poorly designed digital filters, but this is not a restriction of "minimum phase"....

Thanks john.
But isnt the appdizing filters linear phase, compared to nonlinear phase with minimum phase. In turn removing the dispersion problem that minimum phase has but gaining in its benefits of better pre ringing.
This would seem to me, if true, be ideal if dispersion is really a problem as it seems to be. I was just so shocked at my experience with this filter and how well it matched to the dispersion characteristics detailed in said article.

 

[highfi]

highfi,

I'll ship your unit at the end of the week.

Thanks John,that would be fantastic!

 

[ChrisPa]

Dan

If you're researching DACs and filtering, then you should have a look at sinc filtering with (NOS) DACs. This is an underlying reason why (digital) filters are are applied to DACs.

 

[DANOFDANGER]

Dan

If you're researching DACs and filtering, then you should have a look at sinc filtering with (NOS) DACs. This is an underlying reason why (digital) filters are are applied to DACs.

Thanks for the tip. Its killing my brain haha. I learned about filters for conventional NOS dacs years ago now. What i understood was the need for low pass filters is to remove sampling images and prevent aliasing. Then with oversampling came more relaxed and linear filter design(analog or digital). But with sigma delta dacs it appears that digital low pass filters is for quantization noise not HF sampling errors. All papers i have read say the same thing. Its suggested that digital filters handle this noise and the analog output filter deals with the anti imaging and reconstruction.
So unless i am missing something it seems to be the case.

 

[adamdea]

I am sure you mean the OT oversample without "digital filters" they still indeed filter though, from the frequency plots of the filters they show all low pass filtering inside the passband. Utilizing the passband instead of leaving it mostly intact like most other filters. Its essentially sacrificing passband linearity for time domain linearity. I am not even sure it does not use digital filters, i think it still does.

Yes. Sample and hold actually is a filter of sorts with a drooping frequency response but that's not how it is marketed.

When oversampling came along in conventional r2r dacs, it pushed sampling images further away from the passband. Allowing a more simpler anti imaging analog filter to be user.

yes. It removes the images from the orginal nyquist to the nyquist of the oversampling rate. It leaves those above

When sigma delta arrived, it just did this exessively more.

For ADCs not for Dacs AFAIK. I can sort of see that you could somehow have a filter built into the S/D process but AFAIK it doesn't happen. (Could be worng though)

When they changed the LPF to digital this allowed the analog filter to act only as a reconstruction filter, allowing it to be a simple rc circuit rather than a multi order filter for the 20-24khz band.

I think those dsd images have the filter before the SD modulator to show how it is bypassed, if they had the filter after they could not demonstrate their point as the line would show it bypassing the SD modulator as well.
Flick through this white paper for me. I have seen a few others with them all showing the digital filter after the modulator.
https://www.google.co.nz/url?sa=t&source=web&rct=j&url=http://www.analog.com/media/en/technical-documentation/application-notes/292524291525717245054923680458171AN283.pdf&ved=0ahUKEwjjm-TDj43PAhUEHpQKHdRiBWEQFgggMAI&usg=AFQjCNFum4gPsrSfpi3gNtaOsMhe_iswBg&sig2=_Nta0VYobAL1ya-oQn1Ozw
*EDIT*. After thoroughly looking at multiple papers i have concluded that the AA\AI filter is before ADC and "AFTER" DAC. The anti aliasing and anti imaging filters are handled by the analog LPF stages not the digital LPF. As far as i can tell, oversampling does not remove the spectral images of the source, but merely moves them further from the passband. I can understand that digital filters may remove unwanted sampling images, but they dont appear inside its filter bandwidth ?. Also regarding the digital low pass filter, i assume the filter is done in the oversampler/interploation modulator. Please confirm.

Dan you seem to me to be mixing up digital filter and analog filters. The digital filter in the Dac is quite clearly indicated in figure 6.34 using the words interpolation filter. In the ADC it is after the modulator.
Yes ADCs and Dacs also have analog filters. This is obvious.Both and anolog and the Digital filters have a function of removing Aliases in the ADC and images in the Dac. [hence AI/AA filter could refer to either]The analog filter in the Dac also removes S/D noise.

John has said they are all the same but with mathematical differences. Also from memory john could measure no differences in time or frequency domain between the three filters, i could be wrong though.


Thanks for that link adam, it looks very very detailed. I will do some reading and get back to you on all this. Thanks again.

So can this be made clear to me. Since the oversampling modulator removes aliasing as you say, but not all of it. Then the digital LPF is mostly for noise just outside the passband due to noise shaping. So its not or mostly not a anti imaging filter but more just a plain old LPF for noise.
Why is this noise important to filter out ? Does it impact the passband ?

EDIT.
Further research as far as i can tell shows that the oversampling modulator preceding the SD modulator only shifts the spectral images further up the frequency band. Oversampling cannot remove "spectral images" but rather move them somewhere else or possibly transform their energy.
The answer to my question appears to be that the Digital LPF handles only the Quantization noise from the SD moduation

Nope that's the analog filter in a Dac

and anti aliasing\anti imaging is still handled by the analog filters.

the digital and analog filters share the lifting. They have to because
1. a digital filter can only remove spectral images between .....
2. at the front end of an ADC you hsave to filter before digitising
3. at the back end of a DAC you may want to remove noise from the D/A process and quantisation noise from the S/D conversion, which takes us neatly to

Still yet to figure out why we need to filter the Quantization noise outside the audible bandwidth but i assume it must be because it affects the passband somehow.

It might, but the main reason is AFAIK because it might make your amp unstable. In any event why bother amplifying loads of spurious energy.

 

[DANOFDANGER]

For ADCs not for Dacs AFAIK. I can sort of see that you could somehow have a filter built into the S/D process but AFAIK it doesn't happen. (Could be worng though)

Prior to digital low pass filters the entire task was up to a analog filter. This filter with redbook had to sit between 20 and 24.1khz and provide a steep linear attenuation of 96db by the time it reached 24.1khz. This filter was needed by both the adc and dac to prevent aliasing and imaging. Then oversampling came in, provided that larger gap. Simpler analog filters could be used. With sigma delta same thing.

Dan you seem to me to be mixing up digital filter and analog filters. The digital filter in the Dac is quite clearly indicated in figure 6.34 using the words interpolation filter. In the ADC it is after the modulator.

No as i suggested in my EDIT, i said that i thought the digital low pass filter was done in the oversampling/interpolation modulator behind the dac.
What i was talking about was the AA and AI filters, which i believe you think is done on the digital and analog filters. But this i think is wrong. With oversampling the sampling images are shifted so far from nyquist frequency that the digital low pass filter doesnt even touch it. The digital filter from as far as i can tell is required only for the quantization noise just outside the passband(for reasons beknown to me). It has to low pass filter within the same guidelines as the older filters, between 20 and 24.1khz. Then the analog filter does the last job of reconstruction with it filtering out the discrete noise patterns from the digital output into a continous signal. Something that cannot be done by digital filters because it is inherently digital itself.

Yes ADCs and Dacs also have analog filters. This is obvious.Both and anolog and the Digital filters have a function of removing Aliases in the ADC and images in the Dac. [hence AI/AA filter could refer to either]The analog filter in the Dac also removes S/D noise.

I do not beleive the analog filter goes anywhere near the passband. Not even on ADC. The signal upsampled by the modulator allows for a simple analog LPF that starts far beyond the passband, closer to where the new sample images are located.

the digital and analog filters share the lifting.
They have to because
1. a digital filter can only remove spectral images between .....

The digital filter is a low pass filter characterised by design, it is not limited to spectral images. How do you explain the difference between slow roll off and fast roll off.

2. at the front end of an ADC you hsave to filter before digitising

The front end of the analog adc used to handle the bandlimiting function but now this has been largely taken over by the digital low pass filter and the analog LPF simply acts as the AA for the oversampled images.

3. at the back end of a DAC you may want to remove noise from the D/A process) and quantisationing noise from the S/D conversion, which takes us neatly to

It might, but the main reason is AFAIK because it might make your amp unstable. In any event why bother amplifying loads of spurious energy.

Yes unstable amp problems might be it, i have read other reasons too. But this noise is not delt with by the analog filter, this is handled by the digital LPF. The analog filter has no beering on out of band noise filtering. At least not with the modulators we are talking about.

I am all very tired, i could be completely wrong but i just dont think so. Perhaps you are a bit lost on what filters do what exactly ?. I dont mean to offend. I just have a nasty headache haha. ^.^

 

[adamdea]

I am all very tired, i could be completely wrong but i just dont think so. Perhaps you are a bit lost on what filters do what exactly ?. I dont mean to offend. I just have a nasty headache haha. ^.^

Let's leave it there. I hope this process has clarified for you whatever it was that you wanted to clarify.

 

[DANOFDANGER]

Let's leave it there. I hope this process has clarified for you whatever it was that you wanted to clarify.

Most of it to me is answered and makes sense in my head, just would hate if it is all wrong. Would be useless information. But all the papers describing D S modulation and filters i understand so i feel i have a good deal correct. Thanks for trying to help. Maybe someone could confirm or deny my thinking.
Panadol for the head, peace out.

 

[kennyh]

Great thread but boy oh boy is it complicated.

I'd love to see and hear a true blind test with a 20 quid behringer, an MDAC, a Benchmark 2, an FDAC and a/another higher end job. That would answer many a question for me that theory can't.

 

[adamdea]

Most of it to me is answered and makes sense in my head, just would hate if it is all wrong. Would be useless information. But all the papers describing D S modulation and filters i understand so i feel i have a good deal correct. Thanks for trying to help. Maybe someone could confirm or deny my thinking.
Panadol for the head, peace out.

On mature reflection I think I may have been wrong when I said

For ADCs not for Dacs AFAIK. I can sort of see that you could somehow have a filter built into the S/D process but AFAIK it doesn't happen. (Could be worng though)

Clearly there is a digital interpolation filter before the DAC itself. But is there further oversampling filter before the D/S modulator?

I was assuming not but then I thought that actually if it works on a sample and hold basis on the (say) 8Fs input when working its magic at say 64fs or 128 Fs or whatever then that is a sort of filter. And I guess it has to do something (and I think it would be perverse just to zero stuff the input and leave it) .
I had a dig and came across this

http://citeseerx.ist.psu.edu/viewdo...1932682?doi=10.1.1.126.6251&rep=rep1&type=pdf

Which contains this useful passage:
"However, unlike for ADCs, where the high-sample rate signal is decimated and the lowpassfilter requirements, specifically the stopband attenuation, are strictly defined by the allowed aliasing-error, a typical DAC application has no subsequent sampling of the signal after the interpolator, and the replicated spectra from the zero-stuffing will not fold into the baseband.
Thus, the low-pass filter could theoretically be omitted. This would however have negative implications for the overall performance, as the large content of high-frequency energy would saturate the delta-sigma modulator, make the discrete-time to continuous-time interface extremely jitter-sensitive and cause the analog circuitry to slew. Non-linearity would also
result in modulation products in the baseband.
Still, the requirement for low-pass filtering in an oversampled DAC is much more ambiguous than that of an ADC. How much high-frequency content the delta-sigma modulator tolerates for instance, is dependent on the modulator design. At its output, the DSM will also introduce much high-frequency content of its own, so in the case of the discrete-to-continuous time
interface and subsequent analog filtering, the images will be of no influence if they are attenuated to a level where they are well below the DSM high-frequency noise-floor. "

This makes the point that the filtering is not there as a benefit of S-D, it's just there in case not filtering might mess it up, and that the noise of the S-D process is likely to cover it up. Once you have already upsampled to 8Fs the analog filter requirement is pretty weak, so S-D isn;t really there to make it easier.

Nonethless it looks as though there may be some more filtering. The rest of the article goes on in some detail, although it seem to note that zero order hold or linear interpolation should be good enough and any steep filter would be too computationally intensive.

Either way though this happens before the modulator itself.

 

[timola]

Yes, they are the new Hyperstream II devices

Really, it will be a VERY VERY good dac

Can't wait....

If possible John would you look into a nice looking PSU for the MDAC2-Superior...perhaps a similar size or form-factor as the original MDAC itself so that it can be stacked underneath...My MDAC is black so a black case would be nice too....if it's going to be like the original (hide it out of site)...then any shape will do I guess.

Maybe my MCRU could be used?...but I will still be going for the new JW-PSU as I don't want to add another variable.

I think everyone should upgrade their MDACs (if they still have them)...since they are now many years old and in the MDAC2 form (Dual ESS chips and ARM, etc) will be almost as good as an FDAC..it will give your MDACs a new lease of life (and perhaps add value should you wish to sell-on later...the MDAC2 will be a limited run I'm guesing)...I'm gonna use it in my second system when the FDAC arrives...assuming the FDAc sounds better (Just joking!)

At the costs John is offering; for me its a no-brainier to have both DACs (MDAC2 and FDAC)....*(and the DETOX of course).

All the talk of filters is hurting my brain!...I'm glad to know some people here know their stuff....Personally I find it difficult to distinguish (sonically) BETWEEN most of the filters on the MDAC (or the firmware versions)....I'll be trusting you guys to ensure I'm using the best filters on the FDAC....

 

[DANOFDANGER]

http://citeseerx.ist.psu.edu/viewdo...1932682?doi=10.1.1.126.6251&rep=rep1&type=pdf

This makes the point that the filtering is not there as a benefit of S-D, it's just there in case not filtering might mess it up, and that the noise of the S-D process is likely to cover it up. Once you have already upsampled to 8Fs the analog filter requirement is pretty weak, so S-D isn;t really there to make it easier.

Nonethless it looks as though there may be some more filtering. The rest of the article goes on in some detail, although it seem to note that zero order hold or linear interpolation should be good enough and any steep filter would be too computationally intensive.

Either way though this happens before the modulator itself.

Great article. Yes the sigma delta is not there to help with the analog filter, it is just a welcomed consequence.
As it happens the quantization noise (HF energy) from the modulator could cause instability. But it cannot be realised with analog filters in such a short transition band. This is where digital filters are required, because they can be made to such a high order more accurately compared to said analog filters. But there are still high frequency energy generated by the output of the sigma delta modualtor. This discrete time energy is far outside the area of the transition band and also cannot be realised into a continious time signal by digital filters. So a analog filter is needed for these "other" high frequency images.

This article was quite what i needed. I may have less of a headache today ^.^

Here try these.
http://www.resonessencelabs.com/digital-filters/
This one is poorly written, but has the information if you can extract it.

https://www.google.co.nz/url?sa=t&s...GqD57NqK5wWrmWvFA&sig2=QXKBvvkINsUajd6_NuWuvQ

This one talks about the digital LPF and decimator.

http://www.beis.de/Elektronik/DeltaSigma/DeltaSigma.html

Flick down to alias affects for some detail on the filters.

https://www.google.co.nz/url?sa=t&s...Fow7BQyWcFVydbYiQ&sig2=8lKdxao0puoHzpRc52-1pQ

Another one about sigma delta and more info about digital filters.

Cheers adam.

 

[adamdea]

Great article. Yes the sigma delta is not there to help with the analog filter, it is just a welcomed consequence.
As it happens the quantization noise (HF energy) from the modulator could cause instability. But it cannot be realised with analog filters in such a short transition band. This is where digital filters are required, because they can be made to such a high order more accurately compared to said analog filters. But there are still high frequency energy generated by the output of the sigma delta modualtor. This discrete time energy is far outside the area of the transition band and also cannot be realised into a continious time signal by digital filters. So a analog filter is needed for these "other" high frequency images.

This article was quite what i needed. I may have less of a headache today ^.^

Here try these.
http://www.resonessencelabs.com/digital-filters/
This one is poorly written, but has the information if you can extract it.

https://www.google.co.nz/url?sa=t&s...GqD57NqK5wWrmWvFA&sig2=QXKBvvkINsUajd6_NuWuvQ

This one talks about the digital LPF and decimator.

http://www.beis.de/Elektronik/DeltaSigma/DeltaSigma.html

Flick down to alias affects for some detail on the filters.

https://www.google.co.nz/url?sa=t&s...Fow7BQyWcFVydbYiQ&sig2=8lKdxao0puoHzpRc52-1pQ

Another one about sigma delta and more info about digital filters.

Cheers adam.

Thanks Dan, lots for me to read