I don't think a Digital PLL behaves in the same way as an analog PLL?
The whole concept behind the D10 as a base for ESS DAC mods was to have a cheap but well designed design on which mods might be audibly effective but simple enough for everybody to do.
IME, it is possible to make a very good general purpose dac (not requiring HQlayer) of similar architecture to DAC-3, and subjectively better sounding than DAC-3 using AK4137 rather than SRC4392 for the initial upsampling (mainly since DSD256 sounds better than 211kHz PCM, with Sabre dacs anyway), and by clocking AK4137 output synchronous (half or 1/4 the frequency) with the dac chip clock which helps strongly minimize needed DPLL_banwidth. However, getting AK4137 to sound its best remains complicated the way I did it. It would be nice to have an external interpolation filter for PCM as another option, but no work on that recently.
AK4499, noise and all, sounds subjectively better than DAC-3 if using HQplayer. It has an AK4137 on the eval board (they didn't populate the output clock) which I haven't tried yet.
If using a more conventional Reference Voltage power supply, I would say AK4499 roughly sounds pretty similar to typical Sabre dacs, but without that 'ESS sound.' Its maybe more neutral in that regard and will probably measure a little better as the datasheet suggests.
For test and measurement, seems to me Jens probably has it about right. For hi-fi, maybe not as enjoyable an experience that way. The so-called noise filtering at the dac also seems to help filter some of the HF junk that would otherwise come out of the dac and possibly adversely affect sound quality (maybe since the electrolytic caps used for that are lossy at HF).
EDIT: Also, there are other good sounding audio power supplies besides Jung. We will probably find out if a much lower noise one helps and how much.
EDIT 2: For hi-fi development, IMHO the eval board is worthwhile to have, even though expensive.
Last edited:
Both the AK4490 and the AK4499 rely on a clean external reference. So in that respect they are comparable. Some DAC's are much less dependent on that.
There will be more current swing from the AK4499, so that will probably make it even more sensitive to the impedance of the supply than the AK4490.
The noise may cancel to some degree. But the largest cancellation will probably happen when there is no signal. Once there is a signal close to full scale the VREF supply will primarily influence the positive output signal, whereas the negative going signal will draw current to GND.
It will be closer to ground internally, but which way current flows also depends on opamp non-inverting input level (Vref), which may be other than AVCC/2 (to use ESS terminology), no?
I never suggest this or tried to do that - I always said it was a good base for dipping ones toes in the ESS environment
Mark, please don't become aggressive - this has been a civil discourse - let's keep it that way.
Disagreement, debate & discussion is how we ALL share ideas & what we have learned.There is no one oracle on this thread, who must not be questioned or disagreed with - is there?
No oracles, but I would rather not see the thread go from a technical one where more experienced engineers share with and assist less experienced folks wanting to learn, verses a thread for debate like Blowtorch is.
If you have a question, please ask and I'm sure someone will try to help. If you don't want or need help, that's fine too.
If you have a question, please ask and I'm sure someone will try to help. If you don't want or need help, that's fine too.
Last edited:
Yep, I agree with all that & am trying to help where I can & with whatever little knowledge I have. I hope what I have contributed so far is nothing other than technical & I certainly wouldn't like to see it sink to the level of the BT thread.No oracles, but I would rather not see the thread go from a technical one where more experienced engineers share with and assist less experienced folks wanting to learn, verses a thread for debate like Blowtorch is.
If you have a question, please ask and I'm sure someone will try to help. If you don't want or need help, that's fine too.
EDIT: If you do decide to ask, you could possibly get comments or advice included in the response which is intended to be helpful, but that you don't like or don't believe. You don't have to take advice you if you don't want, but you don't have to argue with it either. We try to keep things low key, non-polarized, and friendly if we can.
I intended to report my findings on this thread with my D10 experiments if it's considered appropriate - if not, that will be fine too.
I do hope that this is a thread where technical learning (& hopefully technical questions) is foremost? Technical questions to technical answers should not be considered a challenge or argumentative, simply a desire to be fill in some area of knowledge.
So let's not be anything other than civil, please.
With regard to the whole synch clocking issue with ESS DACs, let's leave it, OK?
Right. When pulling current to GND the op-amp output will of course go to a higher voltage.
I know. The limitation is all mine. The D10 can only offer a limited insight into the ESS environment. A limitation that grows if your starting point is the ES9038Q2M.
My "board" finally measured here Review and Measurements of E1DA 9038S BAL Portable DAC & Amp | Audio Science Review (ASR) Forum
Looks really good, great to see you have tamed the ES Sabre hump - any plans for a version with coax (or toslink or usb) spdif input and stereo (unbalanced) rca output and possibly making it available in a bare bones version and a usb/5/9/12v power line i.e. a Khadas Tone Board competitor ?
Guys, I never go to copy something, especially in such a trivial "design" area like DAC chips usage. I offered the DAC board idea with built-in APx55_like THD analyzer but I meet pretty much zero interest here, as I remember. So I gave it up, almost. A big PCB without housing could be much cheaper vs tiny one assembled into the accurate case. I think toneboard_like PCBA may be $50 effortless, and logistic is minimal: buy parts, send it to the SMD soldering facility, that's it.
Chances are it sounds like it too. Designing for measurements is only one consideration. Ask Gustard, and similar.
By the way, Apple makes a small USB dac that measures very well too, and beats many others on price: Buy USB-C to 3.5 mm Headphone Jack Adapter - Apple
Last edited:
Yeah, I saw the discussion where $300 DAC based on ES9038Q2M shows THD .015% and Apple $9 DAC outperforms it effortlessly. In my home probably any 3.5mm jack output what I only can find will outperform that stupid DAC too ))
Archimago's Musings: MEASUREMENTS: AudioQuest Dragonflies Reviewed! Dragonfly Cobalt, Red, and v1.2.
Archimago's Musings: MEASUREMENTS: AudioQuest Dragonflies Reviewed! Dragonfly Cobalt, Red, and v1.2.
Yes, but it comes in such attractive packaging!
More seriously, designing for great measurements and great subjective sound quality is possible too. Its harder than trying to do only one or the other. Benchmark did a really good job of that with DAC-3, but while very good its age is starting to show, IMHO.
Last edited: