Dac chips - some kind request of advice

[terranigma]

It take it audioman54 hasn't taken time out to listen to what a transformer does in the output stage of a DAC (or an input stage of an amp for that matter) preferring to rely on the usual objectivist objections over 'linearity' and 'distortion' which I'm tempted to term 'prejudice' I had those concerns myself for many years until I tried trafos.

I agree on this. I'm also using transformers at output stage of my dual mono AK4399 kit (but within an unconventional connection scheme that I mentioned here). Relying on most fundamental AC signal principles makes transformers sent-from-heaven devices for differential outputs of modern dacs.

Hi ... just to say that i could not resist
and i bought this one ...

24bit 96K USB 2 0 SA9023 OPA2132 CS4398 DAC USB to RCA Headphone | eBay

Nice board but consider that power supply quality and analog stage are decisive on final sound signature and noise performance regardless the model of dac chip.

I personally prefer the kits with external linear power supplies thus you can build your own linear power supplies instead relying on computer supplied 'switching' DC voltage.

Example:
http://www.aliexpress.com/store/pro...rd-suppoert-PCM-model/420320_32261422437.html

Aliexpress.com : Buy YJ dual AC12V 0 12V AK4399+CS8422 + AD827 Decoder board from Reliable board shirt suppliers on Best Partner Mall | Alibaba Group

Unless you have noise issues (hum or hiss), there is no real need - but being able to use more of the range of the volume control wouldn't hurt.

Yes, I also have massive hiss right beyond of 50%. I should check its service manual to find out its gain ratio. Thanks.

 

[abraxalito]

It is usb powered only.

That's a major weakness - unless the chip runs well at 3.3V. Otherwise there's no headroom available for local regulation and filtering. What terranigma said about external supplies +100.

 

[ginetto61]

... Nice board but consider that power supply quality and analog stage are decisive on final sound signature and noise performance regardless the model of dac chip.
I personally prefer the kits with external linear power supplies thus you can build your own linear power supplies instead relying on computer supplied 'switching' DC voltage. Example: ...

Hi and thanks alot for the very kind advice. Bought one
You are perfectly right. But i am also looking for external usb power supplies that can replace the pc power.
I still have the feeling that +5VDC are very low ... but i know of some very high level dac relying only on 5VDC ... so i am curious.
That dac chip then it could sound fantastic ... if correctly used of course.
It is at heart of very well received pro units from Korg and Prism Audio.
thanks a lot again for the advice.
I will let you know when i receive both the boards.
Kind regards, gino

 

[ginetto61]

That's a major weakness - unless the chip runs well at 3.3V. Otherwise there's no headroom available for local regulation and filtering. What terranigma said about external supplies +100.

Hi and thanks a lot.
So can we safely say that a dac relying only on +5VDC supply to work is limited by design ?
Or can it have some kind of dc-dc converter inside increasing voltage ?
Thanks again, gino

 

[terranigma]

Hi and thanks alot for the very kind advice. Bought one

I see. Consider that it has no usb, but toslink and coax spdif which is preferable over usb if you have spdif output from your pc. If you are using a notebook, probably one of 3.5mm jack featured with mini spdif output.

This is mini spdif cable in case you have this feature:
Aliexpress.com : Buy Gold Plated Connector Optical Toslink to Toslink Mini Fiber Optic Digital Cable( Male to Male ),Black from Reliable cable wire connector suppliers on Mogoi

 

[abraxalito]

So can we safely say that a dac relying only on +5VDC supply to work is limited by design ?

That wasn't quite my meaning - the limitation comes from a DAC relying on the USB supply, which is notoriously noisy and hence needs filtering and regulating.

Or can it have some kind of dc-dc converter inside increasing voltage ?

Yes - this I think is the best solution if USB power is the only choice. But DC/DC converters are noisy too - the quietest are switched-capacitor ones like ICL7660.

 

[phase]

I would reason to bet that the headphone jack out of an average PC would be at least as good as many of those DACs with basic power supplies/signal filtering. Take a look at the datasheets of some of the chips and see what the sample circuits look like, that would be a place to start.

Sounds like a fun project to learn about digital power supplies, something that would have saved me a lot of $/effort trying to figure that out with the more expensive ones...

I did have some fun playing around with line conditioning/power supply for the PC itself, made decent improvement.

 

[Julf]

I did have some fun playing around with line conditioning/power supply for the PC itself, made decent improvement.

What was actually improved?

 

[ginetto61]

I see. Consider that it has no usb, but toslink and coax spdif which is preferable over usb if you have spdif output from your pc. If you are using a notebook, probably one of 3.5mm jack featured with mini spdif output.
This is mini spdif cable in case you have this feature:
Aliexpress.com : Buy Gold Plated Connector Optical Toslink to Toslink Mini Fiber Optic Digital Cable( Male to Male ),Black from Reliable cable wire connector suppliers on Mogoi

Hi and not problem here ... i have already some usb to spdif converters to try with it. From what i understand usb has a big potential being an asynchronous interface, avoiding in this way possible jitter issues.
But not easy to do rightly.
What i miss often with digital is the sense of "ease" and flow i get usually from analog. There is of course much more noise and even hiss with analog but also this sense of continuum ... especially with ballads that make me want to dance. And tempo of course.
Thanks a lot again. Regards, gino

 

[ginetto61]

That wasn't quite my meaning - the limitation comes from a DAC relying on the USB supply, which is notoriously noisy and hence needs filtering and regulating.

Hi and i understand but i have come to this conclusion. 5V are too low voltage for circuits ... there is no enough delta for regulation.
And in case of op-amps they always have better figures at higher V supply.
And also caps store more energy ... everything is better in the end.
I will keep this in mind next time.

Yes - this I think is the best solution if USB power is the only choice. But DC/DC converters are noisy too - the quietest are switched-capacitor ones like ICL7660.

thanks a lot for the very valuable suggestion.
I will stay away from usb only powered dacs from now on.
Thanks a lot again. Kind regards, gino

 

[abraxalito]

Hi and i understand but i have come to this conclusion. 5V are too low voltage for circuits ... there is no enough delta for regulation.

Yes. Hence USB supplied DACs are going to be at a disadvantage to ones with their own PSUs.

And in case of op-amps they always have better figures at higher V supply.
And also caps store more energy ... everything is better in the end.

This is all true - caps store more energy per unit volume at higher voltages. So decoupling is easier - but only provided a trafo is used at the output to reduce the current draw. Otherwise no real advantage in running higher rails, might as well run just enough rails to handle the signal (normally 2VRMS from a DAC). Since opamps generally run up to 36V typically this calls for a 5:1 step-down trafo at the output (10:1 for balanced) and then running signals at 10VRMS for maximum dynamics.

 

[ginetto61]

Yes. Hence USB supplied DACs are going to be at a disadvantage to ones with their own PSUs.
This is all true - caps store more energy per unit volume at higher voltages. So decoupling is easier - but only provided a trafo is used at the output to reduce the current draw. Otherwise no real advantage in running higher rails, might as well run just enough rails to handle the signal (normally 2VRMS from a DAC). Since opamps generally run up to 36V typically this calls for a 5:1 step-down trafo at the output (10:1 for balanced) and then running signals at 10VRMS for maximum dynamics.

Hi and thank you again for the very valuable advice.
No more questions on usb power only dac ... i will look somewhere else.
Kind regards, gino

 

[aos]

Its not all equally audible though, which is where it seems to me Julf is coming from. So its quite possible to get >96dB perceptual dynamic range at 16bits and 44kHz by clearing noise out of the 2-4kHz band where the ear is most sensitive and shoving it above 10kHz where the ear's sensitivity is rapidly falling off (according to Fletcher-Munson).

But that was my assumption, that you won't be touching the 0-20kHz range (at least not drastically). Sure you could noise shape it like that but then you are not going to have a full dynamic range outside of 2-4kHz and if you are going to do that, you might as well start doing psychoacoustic encoding (e.g. AAC). A file drastically altered like could not be called a reference recording any longer as while it may be useful for listening, it would not be useful for much else. Also you won't be able to get such a drastic result by using only one or even two bits, which was my another stipulation.

 

[aos]

It is actually trivially easy to prove you wrong. Have a look at this file: 1 kHz tone at -105 dB.

It is a 1 kHz tone at -105 dB (16 bit / 48 kHz). According to you it shouldn't contain anything (apart from dithering noise), right?

Well, funny enough:

An externally hosted image should be here but it was not working when we last tested it.

Guess what? It is still there, audible, at -105 dB.

Actually I suppose what I was referring to is signal to noise in the 0-20kHz range which comes to -96dB. For some reason I assumed the "noise floor" you see on FFT would be that, but of course it is not true (that depends on how large is the FFT window, larger window on a periodic signal like this gives "better noise floor"). It makes no sense to define noise without specifying bandwidth so yes if you focus on, say, only 100Hz bandwidth then your noise level will be well below -96dB. So yes, sure, you are right, you can represent a -105dB tone.

 

[Julf]

Actually I suppose what I was referring to is signal to noise in the 0-20kHz range which comes to -96dB. For some reason I assumed the "noise floor" you see on FFT would be that, but of course it is not true (that depends on how large is the FFT window, larger window on a periodic signal like this gives "better noise floor"). It makes no sense to define noise without specifying bandwidth so yes if you focus on, say, only 100Hz bandwidth then your noise level will be well below -96dB. So yes, sure, you are right, you can represent a -105dB tone.

Right - that was my point. The overall noise level over the whole 0-20kHz range will be -96 dB, but you can represent and reproduce approximately 120 dB of dynamic range of a signal - that -105 dB tone will still be audible through the -96 dB *overall* noise floor.

There aren't that many recordings that require more than 120 dB of dynamic range.