Most of the let’s say “often present” DAC people have some kilometers on the odometer when it is about DACs.
If I am allowed to speak for them I guess many have tried almost all old stuff and probably now follow the recent developments and try out/build accordingly. Some may be involved in DAC development for a certain brand.
Then the phenomenon of “listeners” and “measurers” which both are fine people that always (cough) come to an agreement eventually.
All that is a different approach to (sorry) dogmaticly keeping to a certain old chip and/or a certain topology. These guys unfortunately also show up in threads
You are on your own journey but I am sure many will help you out when it is about the currently produced higher performance ESS and AKM ICs as todays paper is yesterdays news. Who cares about ES9018 with ES9038 available!? Since a better chip is a better chip the old ones are forgotten quickly. That database won’t happen soon and what is the use of it?!
Some exceptions like that ES9023P stir up things though with heated discussions. Discussion is good although there may be some truth in “talkers are no builders”. Just teasing, guys.
Last edited:
Another evening of DAC searching led me to the PCM5252/5242 chips from TI.
They seem relatively easy to integrate, deliver good "datasheet specs", and are used in some projects by the community. The documentation is extensive and they seem reasonable prices around 12 €
Is this the good middle ground I have been looking for?
datasheet
They seem relatively easy to integrate, deliver good "datasheet specs", and are used in some projects by the community. The documentation is extensive and they seem reasonable prices around 12 €
Is this the good middle ground I have been looking for?
datasheet
I am a firm believer in gathering knowledge by having a discussion and that's why forums are so important, but this thread already has some offtopic in it that makes it harder to compare the recommendations directly. Databases on the other hand are compressed information and allow easy comparison.
If someone in the future has a similar question he might find it useful to have something to look up. If you look at StackOverflow for example, the answers are rated to allow others to find answers faster.
There is no need to include everything in a DB, beginning with some well-regarded components and having some information about, input (I2S, etc.), output configuration, available documentation, bit-depth and sampling rate as well as related products or forum threads describing the product.
For example, I found the PCM5242 searching TIs catalogue. If I knew that this IC exists I would have found some threads in this forum discussing it. That's why databases are useful.
Why do I have to do it on my own and why should I move up, instead of choosing something specific that suits my needs?
I already did one project on my own, but for this project, I would love to make it more generic and community-driven, I still haven't narrowed down my Idea to a point where I start to prototype. The wider goal is a USB DAC with at least 1 analogue input 2 output + digital out, DSP crossover and equalization, that fits in active speakers and can be used standalone.
Paul Stoffregen has done a fantastic job developing the teensy boards based on the NXP i.MX MCUs, and I took that as an inspiration to see if I can utilize the i.MX RT600, but don't know yet if that will be possible.
The other fantastic boards I take inspiration from are the Hifi Berry boards, delivering good quality DACs and AMPs for the Raspberry Pi, funny enough, they released a new amp featuring the same chip I used in my own amp.
My hope is to use this forum for discussion and sharing ideas to come up with an open-source project that is new and interesting and has a purpose without being too specific. I would love to create a solid platform that delivers reasonable price/performance to hobbyists. Don't want to build the best, cheapest or make money out of it, but I see a niche for what I have in mind.
You are not the first guy with such ambitions. Its mostly quite doable but first you have to tell us a bit more: You say you want a dac inside a speaker box that works with USB. You understand that DACs and ADCs require a time reference and a voltage reference? Where do you envision the time reference residing if there is to be more than one such speaker?
Also, will the equipment work with Windows? If so, what driver model will meet your needs? UAC2 compliance only? Do you know what Windows will do to your audio in that case? ASIO? WASAPI Exclusive Mode?
The above sorts of questions have to be considered at the outset, even before choosing chips.
Tried that PCM DACs and liked ES9023P better. They even went to recycling. WM8741 was also better if I remember correctly.
I think chips that go the extra mile will be discussed of in peaks in fora. People busy with old stuff like to make lists but then the chips are hard to find and they may lack a feature here and there. Then a strong “old stuff is best” sentiment is felt too. Going back in time so to speak. One german member here likes to make such lists. His name is tiefbassuebertr. It’s apparently a german thingie those lists
A database for best performing voltage regulators and good passive parts for sensitive applications like DACs would be perfect. The chip is a choice but how to know which resistors are best?! Electrolytic caps?
I think chips that go the extra mile will be discussed of in peaks in fora. People busy with old stuff like to make lists but then the chips are hard to find and they may lack a feature here and there. Then a strong “old stuff is best” sentiment is felt too. Going back in time so to speak. One german member here likes to make such lists. His name is tiefbassuebertr. It’s apparently a german thingie those lists
A database for best performing voltage regulators and good passive parts for sensitive applications like DACs would be perfect. The chip is a choice but how to know which resistors are best?! Electrolytic caps?
Last edited:
for simplicity, UAC2 compliance will do for the start of the project. The good thing is, that I can build on top of a lot of good libraries.
EDIT: and for clarification on what kind of area I aim for, something that volumio, hifiberry or freeDSP does.
This is not directly in my early goal, I thought about it but there are many options to choose from, and I'm up for some interesting recommendations or ideas. I thought about AES EBU but haven't looked into that subject yet, ethernet might be an option too as long as the MCU provides those options.
I made this post when I started the Idea.
So far, there haven't been an excessive amount of recommendations here, Most people seem to recommend the same ICs, I can't tell thought if they just like the sound or if they have developed audio products including those components.
It comes down to ES9023P, WM8741, AK4490R, and AK4493 if someone can point me in a direction with an honest comparison discussing pros and cons between them or similar ICs I happily read that.
So why not start here?
Last edited:
Based on what characteristics did you choose the ES9023P over the PCM5242? does it make a difference what PCM5xxx IC has been used or the kind of application? I can't make meaningful decisions on a short statement.
You don’t need to. I just jump in the deep and make the best of it. When first results are OK I go on and try to make it better. Old fashioned, I know. Some DACs are just not good whatever you do.
My application is the standard internal or external DAC. Internal in the source via I2S is the best.
My application is the standard internal or external DAC. Internal in the source via I2S is the best.
Last edited:
ES9023P appears to be very easy and low cost to work with. PCM5242 is more complex, it needs a MCU to fully control it, it has a PLL rather than an ASRC like ESS dacs have. ASRC tends to be better since a PPLL tends to be more stable than a PLL (see AD1890 datasheet for an ASRC tutorial). ESS makes swiss army knife dacs (lots of features built in) that measure well. ES9023P has single ended outputs whereas PCM5242 may need differential summing opamps to work with single ended loads.
All that said some people don't like ESS dac sound and some people like it fine. Personally, I don't like charge pumps. Use them in guitar pedals sometimes but not for hi-fi.
Next step up from those two dacs might be AK4493. That gets used in some mid-pro and or higher end prosumer recording interfaces. But its more complex yet, needs more support circuitry around it, and so on. It can also play DSD which some people prefer over PCM.
EDIT: The big problem with UAC2 is Windows will keep trying mess with your digital audio stream. It may resample your 24/192 hi res recording down to 16/44 and do it without any warning to you. The Windows resampler is not a great sounding one either. Hi-fi reproduction is not the main purpose of Windows sound engine. For hi-fi the Microsoft official solution is WASAPI Exclusive Mode.
All that said some people don't like ESS dac sound and some people like it fine. Personally, I don't like charge pumps. Use them in guitar pedals sometimes but not for hi-fi.
Next step up from those two dacs might be AK4493. That gets used in some mid-pro and or higher end prosumer recording interfaces. But its more complex yet, needs more support circuitry around it, and so on. It can also play DSD which some people prefer over PCM.
EDIT: The big problem with UAC2 is Windows will keep trying mess with your digital audio stream. It may resample your 24/192 hi res recording down to 16/44 and do it without any warning to you. The Windows resampler is not a great sounding one either. Hi-fi reproduction is not the main purpose of Windows sound engine. For hi-fi the Microsoft official solution is WASAPI Exclusive Mode.
Last edited:
It can be configured as i2s, SPI or HW controlled with high flexibility on all options
I was looking for a DAC with differential output because it was recommended to me to use one and because it allows more flexibility. no need to use ith though, there are versions of the PCM5xxx that have single-ended outs. From what I can see in the Datasheet there is no additional output circuit needed besides a filter.
The PCM was pretty easy to sketch, took me about 2 hours to make a first prototype schematic and PCB layout (thanks to the datasheet), which I can build on. That's a huge plus in POV
that doesn't really help to make an informed decision, I could of course trial and error through a lot of different ICs, but I have a job and studies to finish :/
There is no problem with doing whatever you want. The folks here will still try to help as needed. Don't be surprised though if your dac doesn't sound as good as a commercial one you could buy for much less money. There are quite a few people in the forum who have gone through several dac projects before finding satisfaction.
Regarding HW control (aka pin control) verses I2C/SPI, there is no need for software control if HW control could do it all. In most cases, HW control in dacs is a dumbed down set of options for the most cost sensitive applications.
Using the PLL with asynchronous USB will result in degraded sound quality. With USB a fully synchronous master clock design is best for hi-fi, but its more complex and more costly.
Its like, just because the datasheet says you can keep it simple doesn't mean you will like the sound as well if you do.
Using the PLL with asynchronous USB will result in degraded sound quality. With USB a fully synchronous master clock design is best for hi-fi, but its more complex and more costly.
Its like, just because the datasheet says you can keep it simple doesn't mean you will like the sound as well if you do.
RT600 is a capable MCU but unfortunately is only available in packages that are more or less impossible for diyers. So the only option would be to use existing boards but probably those do not cater for e.g. digital isolators or external clock input.
The "biggest" package the RT600 comes in has a pitch of 0.5, that's indeed not easy to do. I don't know yet if I will be able to solder BGA, but only one way to find out XD. I have more concerns that cheap PCB manufacturers are not able to produce and assemble the boards based on the specification, vias get quite small... I think I'm right at the edge of what JLCPCB can produce.
My plan so far is to have the MCU with USB and the Clock on ta daughter board with castellated pins, some Pico and ESP32 boards use this technique. this might reduce the problems with mixed signals and the small board will be cheaper to manufacture by a company. The other parts are not so hard to do, I will mostly 0805 packages and only use 0603 if absolutely needed. The bigger board could then be soldered with hot air, a reflow oven or a hotplate. I have soldered 0603 by hand but that's not fun.
By the external clock, do you mean an external master clock for multiple systems or just an TCXO? What isolators should be used, and where?
Last edited: