Hello all,
to summarize the discussion, my options are:
So what would be a good set of external components to pick up? does anyone have a kicad project to share?
Best Fabian
to summarize the discussion, my options are:
- ES9023 - cheap, easy to integrate but no balanced out, decent quality
- AK4490R - expensive but high quality, might exceed the budget of 30 € harder to integrate.
So what would be a good set of external components to pick up? does anyone have a kicad project to share?
Best Fabian
It isn't hard to create balanced out with ES9023. Just feed true data to one chip and inverted data to another. Sure you need twice the number of ES9023s and an HC04 but those parts are relatively cheap.
You could even gild the lily at the cost of a bunch of shift registers and some muxes which would allow you to re-purpose I2S to send +L and -L to one chip and +R and -R to the second one. Which would give you 'dual mono' operation and perhaps allow some CM noise rejection within each chip. But that might be going too far.
Last edited:
Not directly the ES9023, but ESS has a lot of other ICs DACs that look interesting. Would be interesting to try a bit before making a final decision. As I mentioned, I have no expertise in circuit design, having learning objects would be niceSamples!?? ES9023P costs 1.60 Euro/piece. 0 Euro if you pay shipping within Germany
What if we build a small database for ADCs and DACs commonly used for DIY projects?
My experience looking for a DAC so far is quite cumbersome... There is a lot out there but sometimes documents are hard to get or not complete (looking at you ESS Tech). Some are well tested but reading long threads to find specific information is often not so easy.
We could create a GitHub repo with a page for listing the ICs with some useful information as well as related projects and threads. This might help to find and compare components.
What do you guys and girls think?
Best
My experience looking for a DAC so far is quite cumbersome... There is a lot out there but sometimes documents are hard to get or not complete (looking at you ESS Tech). Some are well tested but reading long threads to find specific information is often not so easy.
We could create a GitHub repo with a page for listing the ICs with some useful information as well as related projects and threads. This might help to find and compare components.
What do you guys and girls think?
Best
At the risk of saying some things that other people might disagree with, it seems to me that if budget is the most important requirement that the project be kept simple and low budget, rather than allowing mush feature creep.
Regarding manufacturer data sheets, these days they tend to be a product of the marketing department. They are intended to entice engineers into giving a product a closer look such as in a prototype, much like a resume is a marketing tool for a job seeker which is designed to garner an interview. Specs are intended to highlight the main selling points of the product. Again much like a resume, information that may not be seen as favorable may be omitted from the document.
For one example, take the ESS dac sound verses AKM sound. Many people have noticed a difference in sound that is hard to fully make sense of if only comparing datasheets. ESS 'hump' distortion is one such characteristic not mentioned in their datasheets. Its taken years for people to figure out how to measure what it actually looks like in the time domain. Its becoming more clear how it poses difficulties in high performance output stage design.
OTOH, AKM has some minimally documented features in some of its dacs to enhance perceived sound quality, albeit at the expense of increased steady-state HD numbers. They don't say what the features are supposed to be doing to affect perception, at least they don't say much in the datasheets. AK4499 might have been the most obscure in that regard, with a peculiar, poorly documented bit of circuitry in the evaluation board that they knew somehow increased measured HD. But they neve explained why they did it or what it it was for.
Anyway, my take on it is that comparing specs can only get you so far, much like comparing resumes. Eventually you have to invest more time and effort to learn more, particularly about what wasn't said up front. Up to you to ask the right questions, test for things that matter to you, etc.
Regarding manufacturer data sheets, these days they tend to be a product of the marketing department. They are intended to entice engineers into giving a product a closer look such as in a prototype, much like a resume is a marketing tool for a job seeker which is designed to garner an interview. Specs are intended to highlight the main selling points of the product. Again much like a resume, information that may not be seen as favorable may be omitted from the document.
For one example, take the ESS dac sound verses AKM sound. Many people have noticed a difference in sound that is hard to fully make sense of if only comparing datasheets. ESS 'hump' distortion is one such characteristic not mentioned in their datasheets. Its taken years for people to figure out how to measure what it actually looks like in the time domain. Its becoming more clear how it poses difficulties in high performance output stage design.
OTOH, AKM has some minimally documented features in some of its dacs to enhance perceived sound quality, albeit at the expense of increased steady-state HD numbers. They don't say what the features are supposed to be doing to affect perception, at least they don't say much in the datasheets. AK4499 might have been the most obscure in that regard, with a peculiar, poorly documented bit of circuitry in the evaluation board that they knew somehow increased measured HD. But they neve explained why they did it or what it it was for.
Anyway, my take on it is that comparing specs can only get you so far, much like comparing resumes. Eventually you have to invest more time and effort to learn more, particularly about what wasn't said up front. Up to you to ask the right questions, test for things that matter to you, etc.
Actually AK4493SEQ datasheet explaines that peculiar circuitry but you may not like the explanation.
Attachments
Not so sure. The term, 'obtain expected analog characteristics,' could mean one thing over at ASR and another thing to an audiophile-oriented designer. Also, when they refer to power supply noise they don't say if its coupled inside the dac chip or outside the dac chip. There is some reason to believe they are being coy about noise coupled inside the dac chip through the substrate, hence the development of AK4499EX as a two chip solution. In any case, the circuit increases measured distortion so not clear how its helping to 'obtain expected analog characteristics' in that sense.
Moreover, they chose a corner frequency that has a very specific perceptual effect. They also didn't bother to bypass the 470uf electrolytic with a small film cap, which if done helps reduce HF distortion. The only reason I can find for not bypassing it is that doing so reduces the perceptual effect of soundstage enhancement. Now I just told you something else I wasn't planning on getting into...
Moreover, they chose a corner frequency that has a very specific perceptual effect. They also didn't bother to bypass the 470uf electrolytic with a small film cap, which if done helps reduce HF distortion. The only reason I can find for not bypassing it is that doing so reduces the perceptual effect of soundstage enhancement. Now I just told you something else I wasn't planning on getting into...
For most of the components that TI provides, there are extensive datasheets, including PCB routing and placement guidelines, a good description of recommended power supply and recommended external components as well as pin descriptions and measurements.
As someone who is willing to read it, one can have good results by just following the datasheet. This was also the case for the datasheets from Infineon when I build my ma12070 Amp.
When I started to look into DACs, especially ESS has VERY inconsistent sheets, sometimes just something that might be useful for marketing.
I asked Mouser for more detailed information about 1 ESS DAC, we will see if I get something...
So why collect Data? If it is just for me, I just would not build something that doesn't have good documentation, therefore, no ESS Tech stuff, and probably something from TI, I will look into that in the next few days. But while doing so, I already started a list of possible candidates for my project.
There are a lot of people in this forum with a lot of knowledge about stuff and it would be interesting to combine and sort it for me and others to make more informed decisions when choosing a product.
When I started, I thought I had my requirements, but now I feel my biggest concern is complexity. This could come from multiple factors, one is the number of components in general, and the other is, that most DACs require some Opamps at the output, which means I have to do some research regarding opamps and related circuitry, also using multiple power rails, etc.
The cost isn't the biggest problem anymore, the question is how it can get it done without spending months on research.
This is why I'm interested in building a small database, seeing what options are out there and pick a solution that fits the needs. I don't think I'm the only one having this problem.
Best
Chapter 8 in datasheet describes the analog characteristics. Nothing ambiguous about those. Also datasheets are meant to be read as is so no need to read between the lines. If the RC filter is to reduce noise that is the purpose. Corner frequency needs to be higher than in AK4493 (36Hz vs. 17Hz) due to much higher Vref current in AK4499 which would result in a larger voltage drop with 10 ohm resistors.
My advice is to start with a simple dac chip such as ES9023. Once you gain practical experience you can delve into more complex dac chips such as AK449x or ES9038xxx. Those not only require a higher number of components but also require careful layout and 4-layer PCB boards. Thats is not something I would start with.
Regarding datasheets, they have changed over the years. Scott Wurcer, former VP of engineering at Analog Devices, was very active in this forum for many years. IIRC at one point he lamented changes in datasheets and application engineering support that he said was an industry-wide trend. Back in the 80's there was sort of golden age of manufacturer published application literature. They would send you books for free, datasheets were designed in educate engineers, software was written to support evaluation boards under Windows XP, etc. I would agree that TI is still pretty good, at least some of the time. Depends though. One other thing that goes is is that products are periodically redesigned so they can be manufactured using newer semiconductor process lines. Its too expensive to keep old process lines open for legacy products, so they either have be discontinued or redesigned. When that happens the datasheets tend to get more sparse. Fewer specs are guaranteed, etc.
Regarding opamps for dacs, usually there is only one basic choice: OPA1611/OPA1612. Some dac chips appear to be designed to work with those exact opamps. Datasheet example circuits use those opamps, and so on.
On the subject of complexity, I agree with Mr. bohrok's advice. Dacs are complicated. Its hard to design a good one first time around. Experience can help a lot. Having said that, I will disagree with Mr. bohrok about some aspects of dac design, say, make a distinction between a dac intended for listening pleasure verses a dac for intended for standardized measurement use. The reasons are both philosophical and practical. Some people believe a measurement dac should make for the most ideal listening dac, the most truthful dac. Other people, like me for instance, might consider that view overly reductionist. Its not just about steady-state FFT measurements, since that's not exactly how the human ear/brain system hears. No instrumentation exactly captures performance in a way that that perfectly correlates the way humans hear. So, anyway, don't be surprised if you see some disagreements. When he and I do agree its probably safe advice.
Regarding opamps for dacs, usually there is only one basic choice: OPA1611/OPA1612. Some dac chips appear to be designed to work with those exact opamps. Datasheet example circuits use those opamps, and so on.
On the subject of complexity, I agree with Mr. bohrok's advice. Dacs are complicated. Its hard to design a good one first time around. Experience can help a lot. Having said that, I will disagree with Mr. bohrok about some aspects of dac design, say, make a distinction between a dac intended for listening pleasure verses a dac for intended for standardized measurement use. The reasons are both philosophical and practical. Some people believe a measurement dac should make for the most ideal listening dac, the most truthful dac. Other people, like me for instance, might consider that view overly reductionist. Its not just about steady-state FFT measurements, since that's not exactly how the human ear/brain system hears. No instrumentation exactly captures performance in a way that that perfectly correlates the way humans hear. So, anyway, don't be surprised if you see some disagreements. When he and I do agree its probably safe advice.