Re: Re: Re: Re: Re: I2S Data width
Well, no, I am not anybody important as another post by me basically asks: am I the only one around here who does not work for a commerical electronics/audio company or have a degree, masters or PhD in electronics!
I'm essentially an audiophile noob to DAC design who just wants to build the best possible DAC he can learning from everybody else. I used to own a lot of Chord equipment including the DAC64 Mk2. I've talked to John Franks at length before being of an inquisitive mind, but he is a very approachable friendly guy willing to lend his ear to his customers and engage them.
The issues for me are where this thread goes too in-depth too soon, then it is meaningless to me until I re-read it a dozen times and make deductions or research terminology and accompanying theory.
I know the basic steps to building a DAC hifi unit the novice DIYer way: Digital Receiver -> (A)SRC filter -> DAC chip -> output filter and how to read a datasheet well enough to maybe add a decent external master clock and suitable surrounding components.
I write computer software for a living. I do not work for a commercial electronics/audio company or hold an electronics or maths degree of any kind. It just so happens I have fairly good deduction to make educated guesses at to who is who here. Partly that is because I have over 10 years in system which perform information aggregation and analysis. Hopefully that deduction ability helps me learn from the likes of people like you.
I do agree that a super DAC should be in the offing. There certainly appear to be enough people here capable of high-end DAC design (both on chip and the wider hifi unit wise), although some here appear to take theory of practice as unbreakable rule and this sometimes manifests in arguing that a design is stupid when it takes a new and novel approach to circumvent issues that cause the theory of practice to arise in the first place. (Just observation). I do hope that such an item will be a corporate design of DIY Audio members and not a commercial enterprise, or something commercial enterprises tweak enough to be called different and saleable.
If you don't believe I do not work for an electronics firm, just look up the IP address block on something like www.ripe.net and it will show you where I work. Make sure you use the .net extension as the .com one would be extremely embarassing.
FuriousD said:
Well, no, I am not anybody important as another post by me basically asks: am I the only one around here who does not work for a commerical electronics/audio company or have a degree, masters or PhD in electronics!
I'm essentially an audiophile noob to DAC design who just wants to build the best possible DAC he can learning from everybody else. I used to own a lot of Chord equipment including the DAC64 Mk2. I've talked to John Franks at length before being of an inquisitive mind, but he is a very approachable friendly guy willing to lend his ear to his customers and engage them.
The issues for me are where this thread goes too in-depth too soon, then it is meaningless to me until I re-read it a dozen times and make deductions or research terminology and accompanying theory.
I know the basic steps to building a DAC hifi unit the novice DIYer way: Digital Receiver -> (A)SRC filter -> DAC chip -> output filter and how to read a datasheet well enough to maybe add a decent external master clock and suitable surrounding components.
I write computer software for a living. I do not work for a commercial electronics/audio company or hold an electronics or maths degree of any kind. It just so happens I have fairly good deduction to make educated guesses at to who is who here. Partly that is because I have over 10 years in system which perform information aggregation and analysis. Hopefully that deduction ability helps me learn from the likes of people like you.
I do agree that a super DAC should be in the offing. There certainly appear to be enough people here capable of high-end DAC design (both on chip and the wider hifi unit wise), although some here appear to take theory of practice as unbreakable rule and this sometimes manifests in arguing that a design is stupid when it takes a new and novel approach to circumvent issues that cause the theory of practice to arise in the first place. (Just observation). I do hope that such an item will be a corporate design of DIY Audio members and not a commercial enterprise, or something commercial enterprises tweak enough to be called different and saleable.
If you don't believe I do not work for an electronics firm, just look up the IP address block on something like www.ripe.net and it will show you where I work. Make sure you use the .net extension as the .com one would be extremely embarassing.
I've done that simulation last year. For a discrete I/V+filter combo fed 20 kHz at -6 dB, I superimposed same-amplitude signal at the oversampled rate, and also at some MHz. I still got better than -100 dB THD for 20 kHz main signal (I had a many pole filter with ideal components after the I/V to separate out the 20 kHz).FuriousD said:
http://www.diyaudio.com/forums/showthread.php?postid=1425859#post1425859
I merged two threads on this subject so they are now in one spot.
Re: Re: Re: Re: Re: Re: I2S Data width
Please read the thread. All the info is here.
WRT filtering Putzeys does point out some relevant information
here:
http://recforums.prosoundweb.com/index.php/t/18922/0/
I believe his view is that a limitation of most common DAC DF's is that they are half band.
http://recforums.prosoundweb.com/index.php?t=rview&goto=305738&th=19619
Having quoted all this, I'm still not totally convinced blocking -all-
information above fs/2 is absolutely necessary. I've heard digital
filterless DACs which just sound too good.
Has anyone here actually done some serious listening to this ESS
beasty yet?
cheers
Terry
InfiniteGain said:
Please read the thread. All the info is here.
WRT filtering Putzeys does point out some relevant information
here:
http://recforums.prosoundweb.com/index.php/t/18922/0/
I believe his view is that a limitation of most common DAC DF's is that they are half band.
http://recforums.prosoundweb.com/index.php?t=rview&goto=305738&th=19619
Having quoted all this, I'm still not totally convinced blocking -all-
information above fs/2 is absolutely necessary. I've heard digital
filterless DACs which just sound too good.
Has anyone here actually done some serious listening to this ESS
beasty yet?
cheers
Terry
abzug said:
Ok... was only a guesstimate!
As for the TCXO, are you looking at the Connor Winfield site? That fugure is very good. Is it an RMS figure? What about ageing? Typical drift rate? 5.0V or 3.3V?
For a minute there I thought I read VCXO and was going to ask you why use a VCXO, but TCXO of that caliber for $50 bucks is good. I would take 2.
Out of band
So you had the analogue filter set where? 50K? 100K? Using active electronics as an I-V converter. So feedback loop corruption very much an issue and major effect on the sound. To illustrate this issue, try listening to any hifi system with a wireless LAN transmitter in the room. Switch it on and off. You will hear dynamic compression, squashed image etc.
All either HF demodulation or feedback loop corruption.
Also, an analog filter set so close to 20kHz is impossible to do without serious phase effects in band. All very audible.
Sure if you want to cascade multiple stages together but this will always sound bad and introduces . The lower you squash to the out of band close to 20kHz, the further out you can push the analog filter.
Passive would be best then feeding a JFET front ended output stage. The JFET's don't fold down in the same way as bipolar amps do.
So you had the analogue filter set where? 50K? 100K? Using active electronics as an I-V converter. So feedback loop corruption very much an issue and major effect on the sound. To illustrate this issue, try listening to any hifi system with a wireless LAN transmitter in the room. Switch it on and off. You will hear dynamic compression, squashed image etc.
All either HF demodulation or feedback loop corruption.
Also, an analog filter set so close to 20kHz is impossible to do without serious phase effects in band. All very audible.
Sure if you want to cascade multiple stages together but this will always sound bad and introduces . The lower you squash to the out of band close to 20kHz, the further out you can push the analog filter.
Passive would be best then feeding a JFET front ended output stage. The JFET's don't fold down in the same way as bipolar amps do.
Right...
I went to Shaw Electronics website. The Quote page has no submit button so is useless. I filled out the Info Request page, hit submit and got a 403 error (Forbidden).
That leaves phone and I am many hours adrift in the UK making it tricky as I have a lot to do at most times as I get married on Easter saturday and am making final arrangements.
Any chance anybody is interested in the group buy approach?
I went to Shaw Electronics website. The Quote page has no submit button so is useless. I filled out the Info Request page, hit submit and got a 403 error (Forbidden).
That leaves phone and I am many hours adrift in the UK making it tricky as I have a lot to do at most times as I get married on Easter saturday and am making final arrangements.
Any chance anybody is interested in the group buy approach?
VCXO Clock
Look at http://www.crystekcrystals.com/crystal/spec-sheets/vcxo/CVHD-930.pdf
0.2ps from 10Hz at 45.1584MHz. (1024x 44.1kHz).
I have spec'd it as a jitter reading merely to compare with the TCXO spec above, but this is irrelavent really, and the fall off of the phase noise spec if far more important.
0.2pS at 45MHz is alot harder than 0.2pS at 24MHz. I think that this device is more suited to the ESS DAC part. It also has the advantage that it can be used either as a VCXO to track and buffer the incoming data, thus reducing jitter.
$100 USD each on 8-10 weeks lead time if anyone is interested.
Please let me know as I will be placing the order for them shortly.
Look at http://www.crystekcrystals.com/crystal/spec-sheets/vcxo/CVHD-930.pdf
0.2ps from 10Hz at 45.1584MHz. (1024x 44.1kHz).
I have spec'd it as a jitter reading merely to compare with the TCXO spec above, but this is irrelavent really, and the fall off of the phase noise spec if far more important.
0.2pS at 45MHz is alot harder than 0.2pS at 24MHz. I think that this device is more suited to the ESS DAC part. It also has the advantage that it can be used either as a VCXO to track and buffer the incoming data, thus reducing jitter.
$100 USD each on 8-10 weeks lead time if anyone is interested.
Please let me know as I will be placing the order for them shortly.
BrianDonegan said:
Probably, but you guys have more experience than me. I will no doubt do a part-time degree in audio electronics at some point as I wish it was my job.
BTW: I learned much reading the threads on the Opus DAC and am still forming my ideas in general on my first DAC project, but basically it will be: CS8416 -> CS8421 -> x2 PCM1794 -> tube I/V
This ES9008 chip will probably form a subsequent project.
Re: VCXO Clock
That's a good part, but I always wonder how mains noise or severe voltage fluctuation is going to affect a VCXO due to any shortcoming in PSU design. Maybe I'm paranoid in preferring a TCXO or OCXO. Having said that the stability range looks easily maintainable. I'm torn between this and abzug's TCXO.
FuriousD said:
That's a good part, but I always wonder how mains noise or severe voltage fluctuation is going to affect a VCXO due to any shortcoming in PSU design. Maybe I'm paranoid in preferring a TCXO or OCXO. Having said that the stability range looks easily maintainable. I'm torn between this and abzug's TCXO.
Also, somebody help me out here: I sometimes see DAC chips use a sinewave XO and sometimes use CMOS or TTL. I am aware of the different types of output signal waveform, but can somebody tell me what the benefit/drawback of each is, or is it irrelevant so long as the frequency is stable and accurate?