BGA is next to impossible for diyers.
If you plan to use RT600 as USB-to-I2S it is much better if the audio clock (MCK) used for generating I2S signals is fed externally rather that rely on the MCU PLL. But I haven't studied the datasheet in detail so I don't know if it is possible with RT600. Digital isolators should be used for in/out I2S lines between USB-to-I2S board and adc/dac.
If you plan to use RT600 as USB-to-I2S it is much better if the audio clock (MCK) used for generating I2S signals is fed externally rather that rely on the MCU PLL. But I haven't studied the datasheet in detail so I don't know if it is possible with RT600. Digital isolators should be used for in/out I2S lines between USB-to-I2S board and adc/dac.
I will include it in my design. will general-purpose components do or should I look at something specific?
I plan on using a good and reliable clock for the MCU
Last edited:
The faster and less jitter the better.
Good MCU clock does not translate into good I2S signals. And with integer PLL the I2S frequencies will probably be slightly off.
Why? I started with circuit and PCB design 8 months ago and I'm not getting better if I don't try. If you would have asked me last year, true. But with enough dedication it's possible. I'm fairly confident in reading datasheets and designing symbols and footprints for the components including the RT600 MCU, and can follow good advice. I also learned when to stop digging my own rabbit hole (maybe the most important lesson) and I have decent tools to solder SMD QFN components. (not knowing if it will be possible with BGA tho). I definitely have not enough experience, especially in simulation and measuring and I bet it will be necessary with the first DAC.
To be honest, I don't think the DAC is my biggest problem regarding complexity. I fear the DSP implementation will be the real challenge
just had a glance at Digikey, they don't come cheap... at least what's available there.
I don't know if I want to tackle the ASRC vs. PLL idea. Seems pretty complicated to me
Last edited:
There is some controversy regarding audio clock close-in phase noise verses far-out phase noise (aka jitter). IME clocks such as Crystek 957 have to be babied to get the best out of them. Of course I'm mostly using DSD which is said to be more sensitive to phase noise than is PCM. Regarding jitter, on average it affects the noise floor as seen on an audio FFT. Close-in phase noise is often treated a separate consideration, such as in, say, for example, radar. It tends to show up more as skirt-widening at the base of spectral peaks.
https://www.analog.com/media/en/technical-documentation/application-notes/AN-756.pdf
https://secure.aes.org/forum/pubs/conventions/?elib=5611
https://www.qorvo.com/products/d/da007548
http://www.museatex.com/phase.htm
https://www.analog.com/media/en/technical-documentation/application-notes/AN-756.pdf
https://secure.aes.org/forum/pubs/conventions/?elib=5611
https://www.qorvo.com/products/d/da007548
http://www.museatex.com/phase.htm
It has actually nothing to do with ASRC. Some MCUs can generate I2S signals from external clock so it is synchronous and not involving any sample rate conversion.
I haven't yet decided, the ES9023 is still a very likely option. I got a lot of good feedback and I am very thankful for the discussion I'm having here!
Skirt-widening caused by phase noise and Vref noise can also be seen with the dreaded FFT using very large FFT size allowing increased frequency resolution (smaller bin size). There is an awful lot you can do with FFT.
Amir from ASR does some very good and structured analysis, all with the help of FFT, i learned quite a lot from his educational videos
ES9033Q seems to be the replacement of ES9023P but with more features (if you need those). Why would using ES9023P make it easier to switch to a "better" ESS chip? A DAC chip that is considered good today still is good next year. ES9033Q seems interesting and it has good specs so maybe a good one to jump into the deep with. Haven't noticed "full documentation".
If you want more or less guaranteed success with low cost, easy solderability, simply good sound and no hassle the choice is simple. If the plan is to make it as complex a challenge as possible then the choice for ES9023P would be somewhat odd
Last edited:
Okay, but have you considered what that approach misses or does wrong? Considered why overreliance on FFT can be misleading? Suggest to listen to the audio files at: https://purifi-audio.com/2019/12/07/amfm/ Then tell us if FFT is reliable for predicting how something will sound?
Also, in another thread I tried to explain some other thoughts about FFTs that I hoped might be useful for some people: https://www.diyaudio.com/community/...ffer-slaved-i2s-reclocker.348074/post-7093036 The idea was just to try to develop some insights about how FFT works and what that may imply.
Regarding the broad utility of FFT, once again I will agree with Mr. bohrok. It is useful, but we need to have some sense of what its not so good for as well.
Last edited:
ES9033: https://www.esstech.com/wp-content/uploads/2021/10/ES9033_DS_0.3.pdf Beyond that early version, an NDA would probably be required.
What do you mean? They have published a datasheet on their website that is way more detailed than the one for the ES9023.
looks like a standard document to me
I said that someone with good tools can make good measurements. What's your point?
I know that with FFT we can do very reliable image processing, edge detection for example. I did some courses on digital image processing at the university, but I don't have any experience with audio measurements.