Practical Implementations of Alternative Post-DAC Filtering

Status
Not open for further replies.
Throwing the baby with the bath water.

But we do have some pointers. The effect is not unsimilar to reducing audible jitter (which I suppose you agree is for real) and also that it must have something to do with the delta-sigma modulator - any timing errors there is converted into jitter, even power supply noise. As Ken said, I believe earlier, this is something in the time domain.
<snip>
My gut feeling? This is a jitter reduction thingy.

I have never heard jitter in my life. But isn't this post DAC thing about analog?

I'm sure 100% that over sampling effect is audible. And I'm 100% sure that the effect is in HF. But i'm quite sure that this is in audible bandwidth, or below 20 kHz...

An explanation to this could be that the access upsampling noise cannot be removed 100% from the audible band, using the [SD] technique used by the chip manufacturer. This removal of the megahertz noise with non perfectly high Q filter will leave residual "noise", starting from may be 15 kHz and increasing exponentially into the stop band where the digital filter has been designed, probably 20 kHz. It is this content between 20 kHz and 15 kHz that is disturbing and that we want to remove...

Had the passive filter done at above 20 kHz there wouldn't be audible effect, because our ears just don't have such sensitivity. But Joe did this starting from 10 kHz and more or less 2dB down at 20 kHz...

This is audible, but only when there is signal content in that range. That's why oversampling DAC is audible and ladder or NOS DAC is not.

But then after the I/V Joe equalized HF to restore the response using cap in parallel with a resistor. My question is, HOW Joe had measure the response, such to make sure that there is no difference between before first LPF was applied and after the equalization took place. Is this accurate or not?

Okay, assume the measurement was accurate. But frequency response doesn't show everything. There are things that is audible and cannot be seen from frequency response. Up to this point, my assumption is that the EQ will NEVER restore the original signal. Otherwise, it won't be audible, or there is no different than adding LPF above 20 kHz...

May be there is certain mechanism in I/V that makes signal restoration is not 100% perfect, or I think it can be simply phase related. Simple thinking is that when you parallel a resistor with capacitor, the HF going through the cap is delayed while those attenuated by and through the resistor is not.
 
I have never heard jitter in my life.

Really?

But isn't this post DAC thing about analog?

No.

Much of this has been covered earlier. The response deviation is the penalty, not the benefit.

We are clearly talking about something that affects the DAC itself, some kind of loading effect.

But d-s DACs actually sound flatter when no EQ is applied, that it can be EQ'd is simply an option and also getting around the endless discussion whether this is an FR thing only, when clearly it is not. You can still have it, so why argue about it. Indeed, if you read the first six posts of the thread, it even discusses how you can EQ the response in a music server like JRiver (64 bit) and easily hear the difference when you turn EQ 'on' and 'off' - but even 'on' you will get the benefit of the alternative post-DAC filter.


Had the passive filter done at above 20 kHz there wouldn't be audible effect, because our ears just don't have such sensitivity. But Joe did this starting from 10 kHz and more or less 2dB down at 20 kHz...

This is audible, but only when there is signal content in that range. That's why oversampling DAC is audible and ladder or NOS DAC is not.

So you are pro-ladder DACs? And prefer no oversampling?

But frequency response doesn't show everything. There are things that is audible and cannot be seen from frequency response. Up to this point, my assumption is that the EQ will NEVER restore the original signal. Otherwise, it won't be audible, or there is no different than adding LPF above 20 kHz...

Agreed on the first part, FR does not reveal everything. I hear too much energy in the top two octaves in d-s DACs, so I prefer not to EQ. Strange thing is that after adding the post-DAC filter, it doesn't make that much of a difference, even if you EQ back to flat @ 20KHz, the increased clarity is there whether or not EQ is applied. Even with EQ back to flat, the d-s DAC loses that 'heat' in the sound - and tonal flatness, opaque imaging. So EQ back to flat is really just academic, to overcome an argument that really has little to do with the topic.

May be there is certain mechanism in I/V that makes signal restoration is not 100% perfect, or I think it can be simply phase related.

If you are thinking that the pulses are better integrated by putting a passive filter right on the output pins, that certainly has been suggested before by somebody in Linz, Austria. Effectively the idea is that the reconstruction is incomplete - I am quite open to that idea, but I suspect others will find reasons according to current orthodoxy that cannot be.

But read those first six posts, find one that suits you, implement it and make up your own mind. I really can't do much more than that. You will find something that is not an FR thing.

If the measurement gurus can find something, then we shall all be satisfied.

Cheers, Joe
 

Yes. I tried it on a non oversampling DAC tho. I thought it was a project by ddouma, the extremist or something. I couldn't hear an improvement by the reclocking. My feeling was that without the extra part the sound was "better" so I removed the reclocking circuitry and never try reclocking anymore. I think jitter is so microscopic that it requires very transparent system [or the opposite] and very good ears to hear some difference.

We are clearly talking about something that affects the DAC itself, some kind of loading effect.

Yes, I read that when you commented on Abraxalito comment regarding overloading the opamp with wideband signal.

Indeed, if you read the first six posts of the thread, it even discusses how you can EQ the response in a music server like JRiver (64 bit) and easily hear the difference when you turn EQ 'on' and 'off' - but even 'on' you will get the benefit of the alternative post-DAC filter.

Yes, I have read them but I focused on the first scheme.

So you are pro-ladder DACs? And prefer no oversampling?

No. Each has their own strength. Weakness of the oversampling DAC is this HF "distortion". If a method like yours can fix this, I think I will be pro OS DAC.

I hear too much energy in the top two octaves in d-s DACs, so I prefer not to EQ.

Yes, I can hear that too. BUT, what is your argument against Scott Wurcer's comment, that his friend rolled off the signal at 10 kHz and liked the result too. I mean, how do you know that it is not similar case.

But read those first six posts, find one that suits you, implement it and make up your own mind. I really can't do much more than that. You will find something that is not an FR thing.

I believe it is not an FR thing. But first you/we have to prove that it is not the same with simply rolling off the top end... [see Scott's comment]

I have mentioned about zobel before. There is a reason why I mentioned it.

In a speaker crossover, you can make a circus with passive components to tailor the frequency response. But I have never liked the idea of putting capacitor and resistor in parallel for equalization. Because I can hear a weakening of the treble energy. It can be positive in term of clarity but negative in term of musicality.
 
Yes. I tried it on a non oversampling DAC tho..

Of course it didn't work.

Sorry, could have told you so. I have my own NOS-DAC here - it doesn't work on it either.

It has been stated continuously not to do this to ladder DACs and in fact it can only make it worse. It only works for delta-sigma DACs, so I am afraid you have wasted your time. Yes, all you have achieved is rolling off the treble, not a good thing to do to a ladder DAC.

In fact, some have said that this actually helps to close the gap between ladder/NOS-DACs and delta-sigma DACs. One guy is Michael Lenehan of Lenehan Audio who was always preferred true ladder DACs and kept away from d-s DACs, so much so that he went into partnership with Clay Gieseler of Gieseler Audio and made the PDX/Gieseler DAC worth near ten grand. Now Mike is listening to delta-sigma DACs he now knows is superior to the one that he sunk thousands of dollars into. The ardent ladder DAC guy has moved on and there is nobody more surprise than him. Yes, he has listening to our post-DAC filter and like what it does.

Cheers, Joe

-
 
Of course it didn't work.

Sorry, could have told you so. I have my own NOS-DAC here - it doesn't work on it either.

Thanks. Then I will try again the reclocking project with OSDAC any time. But, is reclocking really necessary, or is it better to ensure low jittery process and remove the needs for reclocking?

Now Mike is listening to delta-sigma DACs he now knows is superior to the one that he sunk thousands of dollars into. The ardent ladder DAC guy has moved on and there is nobody more surprise than him. Yes, he has listening to our post-DAC filter and like what it does.

I don't know which DAC chip I own that is suitable for this...

But I also like rolling off top end like Scott's friend. Is it not the same??
 
ABX

I think it is easy to record two files. First one is putting the filter right after DAC, second one is after the I/V. Both filters will roll off starting from 10 kHz, with the same Q [more or less], without equalization.

For me this will prove if most of the positive difference is due to DAC loading or not...
 
Find its number and post it here and there will be plenty willing to let you know what kind of DAC it is.

You can also go to [url]www.alldatasheet.com[/URL] and enter it in the search bar and find the datasheet.

You will also need to know whether it is "current" or "voltage" output.

Hehehe... of course I know that. I have many DA chips but the ones comparable to the mentioned chip on page1 are all SMDs. If there is no resistor between DAC output and opamp input then the trace should be cut [?]

Hmmm.... will an average DAC worth this mod? May be I can try with big size DAC such as the old PCM6x so I can breadboard and compare the 2 schemes where the filter is put before and after the I/V...

My favorite OSDAC [the one that I own] is I think uses CS4362 [hard to remember their models]...
 
May be I can try with big size DAC such as the old PCM6x so I can breadboard and compare the 2 schemes where the filter is put before and after the I/V...

PCM6x is a ladder DAC, so again, not suitable.

My favorite OSDAC [the one that I own] is I think uses CS4362 [hard to remember their models]...

Yes, all Cirrus Logic DACs are delta-sigma and "voltage" - so need to do something like back in Post #2 and repeated here:

AK4382_V-Out_RC-2.gif


Adjust 120R to get near -2dB at 20KHz and make sure they are the same value.

Yes, in pretty much all situations, you have to cut the tracks.

Cheers, Joe
 
PCM6x is a ladder DAC, so again, not suitable.

Yes, all Cirrus Logic DACs are delta-sigma and "voltage" - so need to do something like back in Post #2 and repeated here:

AK4382_V-Out_RC-2.gif


Adjust 120R to get near -2dB at 20KHz and make sure they are the same value.

Yes, in pretty much all situations, you have to cut the tracks.

Cheers, Joe

Joe,

In all honesty if you need to smash the audio with a filter that is already -2
to 3dB at 20kHz to make things sound subjectively right, you very likely
have problems elsewhere in the DAC / player that are needing to be covered
up.

Looking at your path of development this seems like it could be the case.

From all the data I have seen, those SAW clocks actually add jitter
compared to a really good low phase noise XO. Secondly ES9018 is best run
in synchronous mode with a really good clock.

Running the 844 like you do open loop and with a small I-V resistor at IP
will add huge amounts of distortion to an already non linear I-V. I dare you
to measure the noise and distortion of this I-V, it will not be pretty.

After many years of playing with all this stuff, it is my experience that if you
get all the other things right, ie; very low jitter clock, very low distortion
open loop I-V, the right power supplies etc the DAC just doesn't need a huge
amount of low passing. Even for DS DAC's additional noise, 3rd order
Bessell at 80 to 100k should do the trick.

I await your barrage of customer testimonials to support your argument LOL 🙂 🙂 🙂

cheers

Tez
 
Looking at your path of development this seems like it could be the case.

Actually, this is all based on assumptions, in fact it works with straight on conventional IC I/V converters and indeed transformers, passive I/V with no post-DAC electronics and so on; if you go back at the beginning and look at the first six posts, you cannot make any accusation of laziness - this is way further down the road that you seem to think. Rather wasting time typing and explain, come around or phone... much has happened.

Cheers, Joe

PS: SAW clocking "dramatically" increases the video quality in the Oppo 105 according AV World in Auckland, NZ. They are the largest store of its type in all Australasia and that shop sells more Oppos than any in the Southern Hemisphere. I suppose that will be regarded as another useless 'anecdote' by the nay-sayers here, sigh.
 
Last edited:
Well when you put it that way, GM and VW sell a lot of cars, does that mean they are trustworthy. You present your anecdotes as prove, but they are not!

And you are going to get more 'anecdotes' and you are not going to like it. Except they are not anecdotes, they are genuine observations that can be tested by anybody given to the means, opprotunity and the tools - that includes you.

So the proof you crave so much is within your grasp - take it !!!

I sit in front of a computer screen and you ask for proof? Isn't that a bit nonsensical as I cannot use that screen as a Star Trek transporter?

Anyone who turn up here will be warmly welcomed, there are some not so far away who may be redaing this, who can get here in very reasonable travel time, just call Joe on 0412-203382 and tell me you are coming.

If that then leads to more 'anecdotes' and you don't like that, then not really my problem as my conscience tells me very clearly that I have done everything that I reasonably could have done.

So again to you Mark - the proof you crave is within your own grasp and only you can do somethng about that, so don't ask anything unreasonable from me, OK?

Cheerily Joe 🙂
 
Status
Not open for further replies.