Also why sales of vinyl are on the increase again. There is just something lacking in the sound produced by recent digital audio chips, regardless of the specs.
I bought a cheap USB to I2S interface to try listening to Youtube through one of my Dacs out of curiosity. I am using my smart tv and home theatre amp at present, and whilst the sound is good, it is also characterless less and not really very involving. Without the video to watch at the same time, I would lose interest very quickly.
(sorry off topic)
So what you're describing is before the DAC I assume..
I hope after conencting the DAC it'll be better.
You mentioned before the AD1862 and AD1865 which are old,
I also saw sometime ago the AD1955 being mentioned..
So to narrow my experiments, which of these 3 do you suggest to take, as my Analog Devices representative?
"Cheap". "Youtube". On that basis you dismiss digital audio?batteryman said:
Don't visit certain High Street 'restaurants'. If you do, on the basis of having to eat their food with your fingers in a noisy atmosphere with uncomfortable seats you might decide to stop eating food.
I agree with this completely.
A technical advantage of small packages is their small current loop areas, that help to keep magnetic coupling under control. They also reduce parasitic inductance between external decoupling capacitors and internal circuitry.
Sorry it does not work that way, the silicon is expensive. The same die is in every package option. Imagine the old 40pin DIP's the tiny die has LONG bond wires to the corners, not good.
Re CS4398 vs CS4344, I don't know which sounds better because specs aren't a definite guide, but I'll bet the CS4398 does.
AD1955 is an excellent chip. It's one of only two I know (the other is PCM179? my memory fails me) that can be set to bypass the oversampling so with 384KHz 24 bit data it becomes non-oversampling (NOS). NOS vs OS sound quite different so definitely try a NOS DAC to see if it's your cup of tea. AD1862 is worth a look for NOS too.
Fancy a curve ball? If you are using a class D amplifier, then you can think about not bothering with a DAC - get a different kind of amp - google "full digital amplifier". No op amps at all in them.
This hobby can get expensive.
AD1955 is an excellent chip. It's one of only two I know (the other is PCM179? my memory fails me) that can be set to bypass the oversampling so with 384KHz 24 bit data it becomes non-oversampling (NOS). NOS vs OS sound quite different so definitely try a NOS DAC to see if it's your cup of tea. AD1862 is worth a look for NOS too.
Fancy a curve ball? If you are using a class D amplifier, then you can think about not bothering with a DAC - get a different kind of amp - google "full digital amplifier". No op amps at all in them.
This hobby can get expensive.
Last edited:
Yeah, seems likely more are available now - last time I looked at this was a few years ago. I've got a dual AD1955 NOS 384Khz DAC - one of a kind afaik. Which chips exactly have OS bypass?
This is what google shows for a starting text of "PCM17":
Yes I was told about Fully Digital Amplifiers when I asked about SMSL AD18..
Amazon.com: SMSL AD18 80W2 Bluetooth 4.2 HIFI USB DSP Digital Decoding Power Amplifier: MP3 Players & Accessories
It looks like a raelly nice device,
and definitely interesting, because of the Fully Digital Amplifier.
It's 145$..
It would've been nicer to start and hear a Fully Digital Amplifier that comes as a small PCB, for eBay/AliExpress, for a cheap price,
like they do with everything else.
Is there such one?
You'll have to visit the relevant websites. Between AKM, Cirrus and TI the number approaches a dozen.
Probably should be mentioned that most (all?) DAC's that report performance at 44/48 vs higher sampling rates behave best at the 44/48 kHz end. I suppose having NOS at higher sampling rates means you can do your own oversampling and filtering, though, if you should find the provided filters unfavorable. (which, I'd love to know how you're hearing up that high, unless you like aliasing artifacts?!)
Cirrus makes the chip for apple, usually their best chip at the time of manufacture.
since it is external to the SOC noise is kept minimal as is jitter.
As I see it, any digital player from very cheap to expensive it is the clock accuracy that will most affect the sound quality. This is because the chip buffers and tests the reader data for accuracy. the sound you hear has been read into buffer memory several seconds before it is played.
isolate the DAC power supply and replace the timing crystal with better units.
since it is external to the SOC noise is kept minimal as is jitter.
As I see it, any digital player from very cheap to expensive it is the clock accuracy that will most affect the sound quality. This is because the chip buffers and tests the reader data for accuracy. the sound you hear has been read into buffer memory several seconds before it is played.
isolate the DAC power supply and replace the timing crystal with better units.
Last edited:
Good point.
BTW I've never seen such reports,
are they included in a DAC's datasheet, or some place else?
Nice
Of course at a very fundamental level this makes no sense at all. Filtering a DSD stream to an analog signal is still a DAC. A digital representation of a signal on a memory device converted to an analog signal at headphones or speakers is a D to A conversion period no matter how it is achieved.
Usually the provided filters of oversampling DACs don't fully suppress imaging, only imaging above 0.55 fs or so at low sample rates and sometimes even higher at high sample rates.
I know of one Japanese experiment that seemed to show that Japanese gamelan players can subliminally hear above 26 kHz, but I'm not sure if the experiment was double blind or only single blind (which would render it useless).
If you can't hear above 20 kHz and have no pets, high sample rate recordings are useless and there is no need to worry about imaging as long as it isn't bad enough to fry your tweeters or cause TIM in your amplifier.
If the Japanese are right, it makes sense to buy high sample rate recordings, to make your own oversampling filter to get rid of the images between 0.5 fs and 0.55 fs and to experiment with apodizing filters.
This is all assuming that the issue of intersample overshoots has been dealt with somehow, as these are less severe with higher sample rate recordings (independent of whether you can hear anything above 20 kHz).
Last edited:
- Status
- Not open for further replies.