Help ! DAC types and sound characteristics

Man I really need help from people with better ears and more audio electronics experience than myself !!!

Do DAC topologies have a audible effect on the analog output, is an r2r DAC going to sound characteristically different to a delta sigma DAC ? , given a compatible line out.

Or do all DACs sound basically the same, with differences in sound attributable to varience in line out impementation, the great open loop vs closed loop debate ,and whatever processing occurs due to the inherent problems associated with da conversion, ie aliasing.

I thought I had a handle on the whole subject , but having recently bought a dx90, and done a shed load of comparison with a bunch of other audio sources, my confidence in my own abbilty to pull all the factors apart , is shaken.

Would it be fair to say all DAC issues come from the transition between one bit value and the next ? Be it from the electrical switching noise or from the fact there is a straight line there, and defiantly not in the analog trace ! In which case topology could make a difference.

This site is full of good engineers and lots of experience, hopefully someone can take the time to reply
Thanks
 
I'll have a go at this !!

First - no one has better ears than yours unless yours are damaged - which I suspect they are not !!;)

In my mind the really clever engineers have an idea, design, assemble, measure and press play and I guess also hoping like hell the measurements do indeed give them the results they expect.
If not they have another solution and keep going with it.
Then we benefit later on as the schema's get published or leaked...haha

I on the other hand can't do any of that but have very basic knowledge and a soldering ability that gets my projects home.
I can however play the harp out of a piano with my eyes closed and always, and rather naively, hoped I'd be able to re create that feeling of live in my lounge by modifying my cd players based on the instructions available here.
That's not always worked out btw but I keep at it in the hope that one day all the people on my discs will walk into the room - they don't and won't :D

So, My Philips 16bit TDA1541a experiments started with vigour and enthusiasm after sizeable and very noticeable gains were made from initially low investment mods.
That chip still does things in the midrange that a lot still struggle to do and there are also a lot of very technical and electronically challenging things that can be done to eek out the last drop - none of which I really understood nor ultimately implemented.
Some people have though and this site is littered with threads from the very clever people who do understand it.

Then there's the bitstream and single bit incarnations that are topologically massively different that still manage to process zero's and one's and deliver different types of sound - albeit very nice with some simple tweaks or major overhauls to power supplies, regulation, clocking and all the other stuff the engineers do.

My foray into this type of chip has been great fun from a learning perspective and very rewarding in sound delivery.
Different to multibit in so many ways and doing certaiin things very well that maybe the TDA 1541a could not.
If I could have the midrange of the TDA 1541a and the bass of a 1549 as an example then I'd be quite happy.....for a while at least !!

I've also had 1545, 1543 chipsets to play with too but hey, for my ears....they certainly don't sound all the same.
So, were back to ears again...and your room, music tastes and the rest of the system.

It seems to me you know a lot more about DAC chips than I do and you could just buy all the ones your intersted in quite cheaply really, rip em up and rebuild based on the projects, experiences and suggestions of the other members here.

You'd hear the differences immediately - the trick is getting the best bits ( bass, mids, treble, air and space, decay, realism blah blah ) of your favourite machines into one....not sure that's possible.

This isn't technically what you expected nor probably wanted in the way of a reply but your post made me think and that's why I like this place.

I do hope some engineers step up and give a better insight

Good luck
 
done a shed load of comparison with a bunch of other audio sources
the minimum to do valid listening tests: http://linearaudionet.solide-ict.nl/sites/linearaudio.net/files/LA Vol 2 Yaniger(1).pdf

if you haven't used level, frequency response matching, a good blinding protocol, direct short delay comparison you really need to start over

NOS DAC without proper sinc roll off correction, adequate reconstruction filtering may indeed sound different from merely competent audio DAC - but most would say that's due to their being broken implementations
 
  • Like
Reactions: 1 user
the minimum to do valid listening tests: http://linearaudionet.solide-ict.nl/sites/linearaudio.net/files/LA Vol 2 Yaniger(1).pdf

if you haven't used level, frequency response matching, a good blinding protocol, direct short delay comparison you really need to start over

NOS DAC without proper sinc roll off correction, adequate reconstruction filtering may indeed sound different from merely competent audio DAC - but most would say that's due to their being broken implementations

Yep agree with it all, tried to match level as best as can, have done some blind tests, need to do more, short delay obviously. But I believe a real implementation of the artical is beyond me at the moment, I just don't have the time.
Also I THINK ( given the fact that placebo and the quality of my ear canals keep making me unsure ) the effect of the line out stages , is too much to compensate for.
I would need a dedicated setup , each DAC would need a sub module, be correctly interfaced, and have similar iv schemes , to get the electronics right would be some work. I have seen such a project but it covered very few DACs and seemed lost in the differences in the digital domain.
The dx90 I mentioned is a twin ESS ( the cheap *** one with twin cores ) implementation , with balanced iv on each channel. It sounds detailed but the snare drum and cymbals is all I'm ******* hearing ! There are reviews everywhere with the best kit known to mankind showing a flat as a pankake frq response, yet I'm just hearing treble , and this is through 20 odd final amps and headphones. So I am wandering about my ears and sanity,
I am in the process of setting up ( got to buy it first ! ) the only r2r I could find , a discrete binary weighted danish design, to try and answer the question for self, but in hope someone had been at this before and save me the effort.
But all these have NFB opamp buffers as outputs , I have more than a passing ( nelson pass get it ! ) distrust of NFB stages driving anything other than other NFB stages with massive input resistance , so maybe bright sound is there ?
Same goes for fio x3, a Wolfson midrange DS DAC , and the ubiquitous opamp buffers. Slightly less bright and slightly less detailed.
So as I said am getting lost in all the variables , and that's why need help ☺
 
I'll have a go at this !!

First - no one has better ears than yours unless yours are damaged - which I suspect they are not !!;)

In my mind the really clever engineers have an idea, design, assemble, measure and press play and I guess also hoping like hell the measurements do indeed give them the results they expect.
If not they have another solution and keep going with it.
Then we benefit later on as the schema's get published or leaked...haha

I on the other hand can't do any of that but have very basic knowledge and a soldering ability that gets my projects home.
I can however play the harp out of a piano with my eyes closed and always, and rather naively, hoped I'd be able to re create that feeling of live in my lounge by modifying my cd players based on the instructions available here.
That's not always worked out btw but I keep at it in the hope that one day all the people on my discs will walk into the room - they don't and won't :D

So, My Philips 16bit TDA1541a experiments started with vigour and enthusiasm after sizeable and very noticeable gains were made from initially low investment mods.
That chip still does things in the midrange that a lot still struggle to do and there are also a lot of very technical and electronically challenging things that can be done to eek out the last drop - none of which I really understood nor ultimately implemented.
Some people have though and this site is littered with threads from the very clever people who do understand it.

Then there's the bitstream and single bit incarnations that are topologically massively different that still manage to process zero's and one's and deliver different types of sound - albeit very nice with some simple tweaks or major overhauls to power supplies, regulation, clocking and all the other stuff the engineers do.

My foray into this type of chip has been great fun from a learning perspective and very rewarding in sound delivery.
Different to multibit in so many ways and doing certaiin things very well that maybe the TDA 1541a could not.
If I could have the midrange of the TDA 1541a and the bass of a 1549 as an example then I'd be quite happy.....for a while at least !!

I've also had 1545, 1543 chipsets to play with too but hey, for my ears....they certainly don't sound all the same.
So, were back to ears again...and your room, music tastes and the rest of the system.

It seems to me you know a lot more about DAC chips than I do and you could just buy all the ones your intersted in quite cheaply really, rip em up and rebuild based on the projects, experiences and suggestions of the other members here.

You'd hear the differences immediately - the trick is getting the best bits ( bass, mids, treble, air and space, decay, realism blah blah ) of your favourite machines into one....not sure that's possible.

This isn't technically what you expected nor probably wanted in the way of a reply but your post made me think and that's why I like this place.

I do hope some engineers step up and give a better insight

Good luck

Than you very much for that , I have a tda in an old arcam DAC , will experiment on the strength of your recommendations , from the data sheet its a weird DAC chip, a kind of multiplexed r2r
What I'm really after is a final result that is detailed yet ballaced, where I'm not hearing the bloody snare drum and cymbal all the bloody time :)
 
I know - all the more reason to find what has already been said rather than ask everyone to say it again. Sorry if that appears unwelcoming, but it would be useful if new folk found the search function.

Yes I did ( use the search function ) it gave me a few hits but all seemed specific to some upgrade or 2 DAC shoot out etc, what I would humbly like and which did not appear is a experienced and knowlegeble statement on the question, I could indeed spend a year getting expert in the digital domain of da conversion, then I would have an opinion. But I have an entire system to design, and at this stage would prefer taking the word of an "expert" , given my listening tests at this point I'm willing to accept anything , it will be better than where I am, in terms of choosing a DAC chip, topology and line drivers.
 
The snag is that if you ask 3 'experts' this sort of question you are likely to get 4 different answers(*), all of them delivered with confidence. You can pick one to believe at random, or you can spend the time gaining understanding so you can eventually make an informed decision yourself.

(*) note that you can sometimes get four different answers even when there is only one correct answer - and this might not be one of the four!
 
The short answer is yes, but it's all depends all many things as power supply, low jitter, good board layout, good ground design. The DAC is one of many parts but very important part. 2r2 DAC I think uses a network of resistors but I don't think nowaday any high end DAC really uses that because it has quite a bit of short coming in the high freq. Sigma-delta seems like a popular architecture for audio but even within them there could be difference in implementation. DCS ring-DAC architecture is a type of sigma-delta but I think it's attention to power supply, low noise design, and probably what's called the "dynamic element matching" aka DEM or DCS called "ring-DAC" which really improves the high frequency characteristic. I have an old Arcam which uses a lesser version of the ring DAC and it sounds fabulous and very natural. The popular BurrBrown is also a type of sigma delta (and maybe the Wofson as well).
I think most digital audio all comes down to high frequency in the treble region. And obviously low jitter is important too but that's true for all things.
 
Last edited:
Do DAC topologies have a audible effect on the analog output, is an r2r DAC going to sound characteristically different to a delta sigma DAC ? , given a compatible line out.

The best way to answer the first question is by finding out for yourself (this is after all DIYAudio). I suggest buy a cheap DAC - listen to it and then try some simple mods. See if you notice anything after applying them.

Here's a wiki describing a highly affordable DAC with a list of mods for you to experiment for yourself - TDA1387 X8 NOS DAC - diyAudio
 
If I could have the midrange of the TDA 1541a and the bass of a 1549 as an example then I'd be quite happy.....for a while at least !!

I struck the same with the TDA1541 whenever I've heard it with different implementations, especially with passive I/V.
Have you tried a PCM1704K R2R multibit?
I have found that with a decent I/V and buffer all direct coupled you can have it all, and even more.

Cheers George
 
Administrator
Joined 2004
Paid Member
I have found great joy in the voltage out DACs like the AKM and Cirrus chips, followed by a 1:1 signal transformer. The Cirrus chips especially do well thru a transformer. They've always been well liked when others have heard them. I know some people rail against this type of chip, but I like them.
Only major cost is the transformers.

Pretty simple stuff, just make a decent power supply and you're off to the races.
 
The best way to answer the first question is by finding out for yourself (this is after all DIYAudio). I suggest buy a cheap DAC - listen to it and then try some simple mods. See if you notice anything after applying them.

Here's a wiki describing a highly affordable DAC with a list of mods for you to experiment for yourself - TDA1387 X8 NOS DAC - diyAudio

To be honest those mods seem a little simple, improving the supply, better decoupling and tweaking opamps
Issues such as those should have been addressed by a good SI and PI analysis of the topology and the PCB layout, signal integrity and power integrity are no longer black arts ☺
, I'm looking more for ES9018s ( mono mode) into a opamp based iv / line out...
VS.....
ES9018s into a current conveyor iv and pure class A line out.
Before I lost all confidence in my ears, the plan was an ESS with current mode all the way to the headphones, no iv at all.
I did a fair amount of surfing and the ESS chip kept getting rave reviews , current mode when driving really weird inductive loads is a no brainer, so topology seemed obvious.
Then got a ESS based device , and all I am hearing is treble , don't get me wrong the rest is detailed and really quite nice, but the snare drum and too much ccymble getting through is making me tired and spoiling the mood !!!
So hence this thread . it has been useful so far, I have a few new DACs to look at , and its very nice having a guy who has heard it recommend it.
 
I have found great joy in the voltage out DACs like the AKM and Cirrus chips, followed by a 1:1 signal transformer. The Cirrus chips especially do well thru a transformer. They've always been well liked when others have heard them. I know some people rail against this type of chip, but I like them.
Only major cost is the transformers.

Pretty simple stuff, just make a decent power supply and you're off to the races.

Chip names ? Cirrus do a lot of chips ! And will look up AKM, never heard of em .
BTW, Cirrus in my experience ( ADCs admitadly ) are a little "uneven" in their quality control at the chip test/selection phase is a little dodgy
Living in Denmark have a vgood transformer guy near so might experiment that way, but they are not exactly small ! Fine for boxed equipment but will Be challenge for portable.
 
To be honest those mods seem a little simple, improving the supply, better decoupling and tweaking opamps
Issues such as those should have been addressed by a good SI and PI analysis of the topology and the PCB layout, signal integrity and power integrity are no longer black arts ☺
.

In the real world these tools only get used on either very critical, expensive designs or when something goes wrong, this is a shame because like you say many things can be optimised during the initial design cycles... But is it strictly necessary and can the cost and time of simulation be worth it for the design...
Most DAC layout now (like SMPS controller chips) is often covered by extensive notes and recommendation's, and some designs so simple that SIV etc. it not really necessary. I can throw an simple ESS9023 design on a double sided board with bottom layer full contiguous ground plane and all signal on the top, full SMD. The newer chips are optimised for simple layout with minimum added components for reasonable results as there is a huge market for these devices in just about every hand held device....
The mods mentioned are simple, but it is simple that works best and will probably give you the best results, complicated changes and mods are more likely to go wrong or not achieve the desired results.
 
To be honest those mods seem a little simple, improving the supply, better decoupling and tweaking opamps

Sure - they're common sense mods, nothing esoteric. Common sense not being so common after all.

So is your point that because they're simple they don't work?

Issues such as those should have been addressed by a good SI and PI analysis of the topology and the PCB layout, signal integrity and power integrity are no longer black arts ☺

Is it your view then that Taobao vendors have the latest and greatest CAD software at their disposal for designing sub-$100 audio components in small volumes?

Then got a ESS based device , and all I am hearing is treble , don't get me wrong the rest is detailed and really quite nice, but the snare drum and too much ccymble getting through is making me tired and spoiling the mood !!!

I'd not heard an ES9018 up until I got my new mobile phone, which has the mobile version of the chip. I have to agree - the HF is way too noisy, not transparent. A bit like a chipamp with too much noise on its power rails. The TDA1387 does not have that problem.
 
In the real world these tools only get used on either very critical, expensive designs or when something goes wrong, this is a shame because like you say many things can be optimised during the initial design cycles... But is it strictly necessary and can the cost and time of simulation be worth it for the design...
Most DAC layout now (like SMPS controller chips) is often covered by extensive notes and recommendation's, and some designs so simple that SIV etc. it not really necessary. I can throw an simple ESS9023 design on a double sided board with bottom layer full contiguous ground plane and all signal on the top, full SMD. The newer chips are optimised for simple layout with minimum added components for reasonable results as there is a huge market for these devices in just about every hand held device....
The mods mentioned are simple, but it is simple that works best and will probably give you the best results, complicated changes and mods are more likely to go wrong or not achieve the desired results.

Yea don't trust the reference designs too much mate, especially pcb references. It depends upon company and which internal product line the chip package combo is under, and what specialists they have access to. ESS for example!e on their ref board use a opamp iv instead of a descrete buffered iv , very much a price compromise, yet their layout is immaculate.
Industry standard SI ( not so much PI ) tools are all over the internet, just look with an open mind , tread the paths others will not, and you shall discover wonders ☺ ( even true now in NSA UK )
But back to the point in you experience is the ESS "trebley" , is it me, my ears or am I going senile !!!
 
Sure - they're common sense mods, nothing esoteric. Common sense not being so common after all.

So is your point that because they're simple they don't work?



Is it your view then that Taobao vendors have the latest and greatest CAD software at their disposal for designing sub-$100 audio components in small volumes?



I'd not heard an ES9018 up until I got my new mobile phone, which has the mobile version of the chip. I have to agree - the HF is way too noisy, not transparent. A bit like a chipamp with too much noise on its power rails. The TDA1387 does not have that problem.

No no, obviously they will work, ( to what audible extent would take all the teadious testing mentioned in an earlier post ) and no doubt improve the sound, but does the "character " of the sound really come from the supply ? I am beginning to believe , mainly from this thread and further reading , its coming from the DAC and associated filters etc etc etc.

Ur yes ☺ very much so, everyone in China has access to the newest and best cad, simulation, finite analysis, mesh solvers money can buy, for free, perpetually. they are just lazy in not using them , to busy pushing new products out in an ultra competitive market. China don't give a damn about software modification, but go nuts if you even think the premiere is a tart ☺

THANK YOU so so so so much mate, really, for real, though my hearing was of, or appreciation of music was of , or music taste was crap,
So other sigma deltas don't suffer from this ??? r2rs are discontinued, and won't be back for a while