The modification suggested requires a little bit of familiarity with the parts.
Mod on the linked sa9023+es9018k2m board (like the one I have here) is to remove the soic opamp (AD823?), tap its input pins and feed those 4 inputs (+/- on both left and right channels) to a separate differential transimpedance stage (typically 3 opamps needed for a SE output, but vicnic does it with just a pair and doubling up a difference on a single opamp).
I'd have to look at any other filters on the board to tell whether there's any caps you'd need to remove.
Mod on the linked sa9023+es9018k2m board (like the one I have here) is to remove the soic opamp (AD823?), tap its input pins and feed those 4 inputs (+/- on both left and right channels) to a separate differential transimpedance stage (typically 3 opamps needed for a SE output, but vicnic does it with just a pair and doubling up a difference on a single opamp).
I'd have to look at any other filters on the board to tell whether there's any caps you'd need to remove.
This is probably your best bet then: ES9018K2M XMOS DSD DXD 384kHz USB DAC with Bit-perfect volume control and SPDIF input - DIYINHK
Pretty nice bit of kit.
Pretty nice bit of kit.
Surely that will depend on whether the DAC has either:
- a voltage controlled amplifier output buffer, or
- expands the bit width to at least 24 bits and then does the scaling
As far as I can tell, its an issue with 16 bit scaling, but neither here nor there if you can scale with at least 24 bits of sample depth and if you can do 64 bit math and resolve to 32 bit samples then the theoretical benefits of analogue scaling are rather hard to justify, even before considering linearity of means to do it.
i'll risk chiming in here. I have a 9023 based (Subbu DAC) that I think sounds great, And I bought a cheapy 9023 usb DAC/dongle for my sons laptop and was impressed by how well it sounded for the money (~49 bucks), not as nice as the Subbu but light years better than the lap top DAC. It would be a safe starting point.
PJN
PJN
The DAC discussed here has enough documentation available to answer these questions - digital volume control in the DSP section, 32bit processing.
I am not saying the digital volume control is wrong/suboptimal, just do not like the misleading/incorrect marketing. Sorry for offtopic, over.
If you would like to know more about some of the factors involved in getting good sound out of a Sabre dac, there is a whole lot of information over here: ES9038Q2M Board
Depends on how deep you want to get into the DIY aspect.
But, trying a bunch of cheap dacs based on chips from different manufacturers and trying to extrapolate from that to what a good one can sound like isn't likely to work very well, IMHO.
Part of the whole equation depends on how good of a dac you want in terms of sound quality, how much you can afford to spend, and how much DIY effort you are interested or willing to put into it. It also depends on how you intend to use the dac. Some people are happy to have an RPi-based music player dac. Other people want a full-featured general purpose dac. More features and more sound quality tend to cost more, of course.
With a fair amount of work and some money, maybe $200 or so in total, it is possible to make a dac the arguably rivals dacs costing over $1,000. Maybe that seems odd or hard to believe, but it appears to be true so far as I am able to tell. How I have come to that opinion is a long-ish story that is still not over, but like I say, it seems to be the case.
Depends on how deep you want to get into the DIY aspect.
But, trying a bunch of cheap dacs based on chips from different manufacturers and trying to extrapolate from that to what a good one can sound like isn't likely to work very well, IMHO.
Part of the whole equation depends on how good of a dac you want in terms of sound quality, how much you can afford to spend, and how much DIY effort you are interested or willing to put into it. It also depends on how you intend to use the dac. Some people are happy to have an RPi-based music player dac. Other people want a full-featured general purpose dac. More features and more sound quality tend to cost more, of course.
With a fair amount of work and some money, maybe $200 or so in total, it is possible to make a dac the arguably rivals dacs costing over $1,000. Maybe that seems odd or hard to believe, but it appears to be true so far as I am able to tell. How I have come to that opinion is a long-ish story that is still not over, but like I say, it seems to be the case.
Last edited:
Why is that a risk?
It's great to hear.
What board is that?
In ebay there are now boards that are SA9023+ES9023, for 16$..
I understand.
If only all boards were optimized..
That's nice.
Maybe there's some manufacturer that does it?
Creates DACs for a relatively reasonable price, that perform like much more than their cost
(if not, then that's a business potential
)BTW, what would be the chip it is based on?
(with the total cost of ~200$)
Last edited:
It is at the ES9038Q2M board link I just gave you. Obviously, it is based on a Sabre ES9038 variant. Cost of the chip itself is about $15, but getting the best out of it is non-trivial. That being said, most people stop making improvements before they get to the best sound quality possible with one. They get to a point where they are satisfied and leave it there. At the moment we are taking sort of a break or diversion to help beginners do some of the DIY work with easier to use components.
Oh..
I always wonder regarding this chip,
do people really use its 768KHz ability?
You need a source that is 768KHz to actually use that..
How many people actually have a 768KHz source..
I think that there are two things going on with 768kHz. One is that people want to have it in case they want to use it, and the other thing is there is interest in it because of the ever increasing focus on trying to improve sound quality by increasing sample rates.
I think that there is a good argument to the effect that 192kHz should be plenty high as a sample rate and if there are still sound quality problems one needs to look for other explanations and solutions.
However, it remains that the popular view seems to be bigger numbers must mean better sound. To sell dacs, manufacturers therefore need to be able to hit the numbers.
That being said, my own ES9038Q2M dac is fitted with a 100MHz clock, so it can hit the big numbers, and I find it sounds best with 11.2MHz DSD. What good does that do if most content is still on CDs? I use an AK4137 to convert CD and other audio to 11.2MHz DSD. That's the way this particular dac sounds best, IMHO.
Of course, it is entirely possible I could get it to sound better with a 30MHz clock and some other signal processing scheme. Some of this stuff has not been adequately studied at least in the public domain. Some manufacturers have done their own research, but they don't give all of it away for free. They may give some interesting hints though.
I think that there is a good argument to the effect that 192kHz should be plenty high as a sample rate and if there are still sound quality problems one needs to look for other explanations and solutions.
However, it remains that the popular view seems to be bigger numbers must mean better sound. To sell dacs, manufacturers therefore need to be able to hit the numbers.
That being said, my own ES9038Q2M dac is fitted with a 100MHz clock, so it can hit the big numbers, and I find it sounds best with 11.2MHz DSD. What good does that do if most content is still on CDs? I use an AK4137 to convert CD and other audio to 11.2MHz DSD. That's the way this particular dac sounds best, IMHO.
Of course, it is entirely possible I could get it to sound better with a 30MHz clock and some other signal processing scheme. Some of this stuff has not been adequately studied at least in the public domain. Some manufacturers have done their own research, but they don't give all of it away for free. They may give some interesting hints though.
Last edited:
This is a strong part
It should be put on every product page of a DAC that is over 192KHz..
(tho no manufacturer/seller would want that on his product page, since it will decrease sales)
BTW,
mentioning not just the chip used in a DAC but also the implementation,
what do you guys think about the implementations of hifimediy.com?
For example:
HiFimeDIY UAE23 USB DAC (ESS Sabre ES9023+Savitech SA9023)
From the product's description, I liked this part:
Seems that they indeed try to make it better than the average board that uses a 1117
I am considering buying this one, as my ES9023..
Even tho it's 3 times the 16$ board in ebay that also has SA9023+ES9023
I wonder how much difference there would really be, but I hope it will be worth it
Last edited:
Data converters almost universally perform worse as sample rate increases. I'm not even a fan of > 96KHz.
Understood. I said 192 should be plenty high. Doesn't mean 96 might not be high enough in most cases. However, some very carefully researched and designed dacs are now upsampling to 211kHz. I have no doubt those particular dacs measure and sound better there than at 96, but those may be an exception to the general rule.
Personally, I wouldn't waste my money on anything like that. But, that's me.
Also, I don't think I have anything else to add to this thread that I haven't already said. Maybe see you over in the ES9038Q2M board thread sometime. I just popped in here because no one had mentioned the implementation issue. Good night, all.
Last edited:
Hmm
So If you see the HiFimeDIY's 42$ as a waste of money, that means that the boards on ebay for 1/3 the price (with the same chips) would be regarded as a even worse than that?
Thank you for all the tips
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.