The AK4118A actually has a fairly good jitter performance. Better than some of the TI equivalents, I have used. I don't know the experience level of the OP, but if this is a first attempt at building a DAC, it should probably be kept simple. And the AK4118A with the AK4493 can achieve a good performance with the right circuit, a good layout and a clean reference supply. Yes, you can achieve a better performance with a better DAC, clock circuits etc., but then we are not talking about a simple design anymore.
AK4118, AK4137 and AK4493 can all be used without I2C in pin control mode. But I agree with Jens in that AK4137 is an overcomplication for first version. BTW I have never seen other than AK4118A version available at Digikey.
Has it been shown that adding the AK4137 or any other ASRC is a guaranteed improvement? If it hasn't then perhaps the thread would be best served by getting the best out the OPs chosen devices.
Last edited:
If you do it really wrong its probably guaranteed to sound bad. Besides, IIRC it has an SRC bypass mode if you want to use it.
Last edited:
Absolutely. I'm fairly comfortable programming MCUs but in this case I'd like to avoid adding one unless absolutely needed since it adds complexity, more routing, noise etc. In fact if I was to add this to a DAC, I'd go all the way with input switching, muting and all the good stuff.
I think it's at least on par with DIR9001/PCM9211. They specify 50ps in their datasheet, but they're not advertising it as much as TI does.
What I'm doing in this dac is some simple reclocking of the I2S signals, purely to "condition" them. There is no clock recovery in this case.
If I were you I'd start with doing a basic dac with the AK4118 providing the MCLK to the AK4493. If you would like something better sounding, I would go for an SRC4392 feeding an AK4493. The SRC4392 is a combination s/pdif receiver and ASRC, so you could easily get it to do what you need.
A combination of AK4118 + AK4137 would also work of course, but a single SRC4392 is more elegant, and there is a chance that it will sound better too..
From the total number of components?
No, because the AK4118/4137/4493 can operate without a microcontroller in pin control mode, while the SRC4392 requires control via I2 or SPI.
So, the total number of components can be even higher.
Also its performance is not as good as AK4118 and AK4137.
Alex.
If we are talking about pin control mode, yes. But it would be a shame to design such a machine and not do software mode. So much more fun to be had. 🙂
Regarding the performance, yes, the 4118 & 4137 combo is more powerful on paper. But I'm not sure that it sounds better in practice, at least when talking about simple s/pdif use.
I have built prototypes of both solutions, quite some time ago in fact, but have not got around to doing a proper A-B test. Mainly because I do not use s/pdif anymore.
Regarding the performance, yes, the 4118 & 4137 combo is more powerful on paper. But I'm not sure that it sounds better in practice, at least when talking about simple s/pdif use.
I have built prototypes of both solutions, quite some time ago in fact, but have not got around to doing a proper A-B test. Mainly because I do not use s/pdif anymore.
Not only.
Used similar combos many times, 4113 and 4118 and 4437 for the simple solutions (for not simple I used FIFO on STM32F407).
Works we well.
Me too, spdif it is not so good in performance, no DSD, no multichannel, etc.
But I'm not doing these things just for myself...
There is some old stuff that was made to be very introductory for people with no experience programming.
Its likely too basic for you, but here it is if you want to take a look:
(some of the links are more or less in reverse order relative to when they were posted)
I2C bus programming with Arduino :
https://www.diyaudio.com/community/threads/es9038q2m-board.314935/post-6386977
I2C Library used for examples, URL:
https://github.com/felias-fogg/SoftI2CMaster
Troubleshooting the I2C bus:
https://www.diyaudio.com/community/threads/es9038q2m-board.314935/post-6392768
Introduction to number systems, etc.:
https://www.diyaudio.com/community/threads/es9038q2m-board.314935/post-6393768
https://www.diyaudio.com/community/threads/es9038q2m-board.314935/post-6393809
More about Arduino I2C bus pins:
https://www.diyaudio.com/community/threads/es9038q2m-board.314935/post-6386941
Arduino pinout diagrams:
https://www.diyaudio.com/community/threads/es9038q2m-board.314935/post-6382409
https://www.diyaudio.com/community/threads/es9038q2m-board.314935/post-6382475
Last edited:
SRC4392 is one of the better ASRCs. Maybe not quite as good as AK4137, but IME its close and maybe the next best thing. It also includes a SPDIF receiver. On the downside, it does need an extra local voltage regulator, and I2C bus programming.
Regarding pin programming versus I2C bus programming, only limited features are available with pin programming. Maybe good to look at the datasheets and see which features will be unavailable.
I2C & MCU brings along a whole new set of issues such as noise. Mitigating those issues properly with e.g. isolators makes the design much more complicated and expensive.
Seen that here too during some tests if I2C bus is active while analog outputs are not muted. However, its often possible that devices can be programmed at power on and then left alone. Especially if using ASRC there may be no need to reprogram a dac, nor a need to run a fancy real time display.
Also it can help some to use larger value pull-up resistors on I2C bus and run the bus slower. It the digital edge risetime that tends to couple noise into analog circuitry. Not so different in some ways from the substrate coupled noise in dac chips, if AKM is to be believed.
Also it can help some to use larger value pull-up resistors on I2C bus and run the bus slower. It the digital edge risetime that tends to couple noise into analog circuitry. Not so different in some ways from the substrate coupled noise in dac chips, if AKM is to be believed.
Last edited:
So open the chassis, hook up a MCU and reprogram every time when powered on. Very impractical and not what I would call proper mitigation.
That's a little extreme. The MCU gets an isolator, and is shielded and or located away from sensitive circuitry if need be. Noise management is one reason a lot of people use pic processors which can be put in a very low power standby state to minimize noise. Turning a volume encoder can generate an interrupt which wakes the MCU to handle it, again if need be.
However, if a dac is sitting out in the open (or maybe in a thin aluminum case) to begin with its probably going to be subject to various radiated noise sources anyway.
However, if a dac is sitting out in the open (or maybe in a thin aluminum case) to begin with its probably going to be subject to various radiated noise sources anyway.
Thankfully, the I2C bus is essentially just a steady dc voltage when there is no traffic going through it.
And the traffic is very low.. in my designs I get traffic only when there is a change in SR or if I make a change in some IC's parameter.
Still, I use galvanic isolation between the uC and the audio boards.
I'm not aware of any commercial audio equipment going this far to minimize uC noise (except for a couple of very special designs).
And the traffic is very low.. in my designs I get traffic only when there is a change in SR or if I make a change in some IC's parameter.
Still, I use galvanic isolation between the uC and the audio boards.
I'm not aware of any commercial audio equipment going this far to minimize uC noise (except for a couple of very special designs).