Probably not going to happen, there is no money in such an ADC, the DAC business cases are related to mobile devices and by the “mine is bigger” principle.
That’s why I’m placing my bet on the 20bit SAR ADCs which exactly meet the requirements you mentioned.
Hello ADC Folks,
sorry if I'm shifting gears inappropriately, but maybe I can get a few evaluating comments on my question anyway (being relatively inexperienced in ADC technology as discussed here):
Owning a TI PCM4222 Evaluation board (which I haven't put to work in my stereo set-up yet). My plan is to "just" AD convert 24Bis/96kHz and connect to my DEQX PDC and to my USB Interface via AES3.
1. I'd like to how well the PCM4222 is up to the job of AD converting a phono line signal (source phono preamp), equally to other or newer ADC chips.
2. To optimise it's performance I was planning to provide well regulated external power supplies and a better, external (maybe OCXO) clock.
What can you folks advice on board and plans?
Thanks a lot!
Winfried
sorry if I'm shifting gears inappropriately, but maybe I can get a few evaluating comments on my question anyway (being relatively inexperienced in ADC technology as discussed here):
Owning a TI PCM4222 Evaluation board (which I haven't put to work in my stereo set-up yet). My plan is to "just" AD convert 24Bis/96kHz and connect to my DEQX PDC and to my USB Interface via AES3.
1. I'd like to how well the PCM4222 is up to the job of AD converting a phono line signal (source phono preamp), equally to other or newer ADC chips.
2. To optimise it's performance I was planning to provide well regulated external power supplies and a better, external (maybe OCXO) clock.
What can you folks advice on board and plans?
Thanks a lot!
Winfried
None of AKM's current production ADCs are overall as good as AK5394A to this day. If they truly cared about performance they wouldn't have killed it, or used filters with ugly passband ripple in the 557x, or made sure that their newer products weren't regressions in various aspects.
They have done exactly what they set out to do: produce an ADC line (557x) that beats the competition on headline SNR / DNR / THD specs in the audio band and is cheap to manufacture.
An interesting chip for digital audio processing/switching/converting https://www.nxp.com/docs/en/user-guide/SCF5250_USERS_MANUAL.pdf Unfortunately it is now obsolete, anybody knows of a functional equivalent part?
Last edited:
ADI has a number of similar parts. So does AKM and TI. It depends on what type of processing you need to do. I'm using one AKM chip that has 4 in, 4 out good quality SRC's on each input and output and the ability to do digital ANC with a small battery.
A more user friendly part would be any of the SigmaDSP's from ADI. AKM used one to mix multiple ADC's together to lower the noise floor and the distortion a little.
A more user friendly part would be any of the SigmaDSP's from ADI. AKM used one to mix multiple ADC's together to lower the noise floor and the distortion a little.
Still looking for a solution to convert SPI to I2S, Left Justified and perhaps Right Justified, without using a FPGA (already done that, but it's rather stupid solution; a FPGA chip is expensive and under used).
The Freescale chip was doing this almost natively, both in Master and Slave modes. As far as I can tell, none of the chips from AD or TI have this feature. Of course, any decent speed MCU/DSP can be programmed to do this job, but if I can get the result without going through the hooplas of coding DMA access and double buffering from scratch, then I'll go for it.
In fact, the Freescale chip was doing everything that XMOS and C-Media chips are doing purely in software. The only missing thing to be the perfect solution for audio was USB support.
This chip was likely killed when NXP took over Freescale. A story way to common in the today's semi industry consolidation environment.
I can imagine that the rather high power dissipation and sensitivity to temperature (if I am remembering correctly from Syn08s measurements) may have put people off. It may also have been seen as making the part obsolete in AKMs eyes.
I can understand AKMs thinking behind the 557x series, they are really designed for multichannel audio equipment, such as line or microphone preamplifiers for studio use. The fact you can use the 5578 in stereo or mono mode is more a by product of having so many channels on the chip rather than it being a user case that many will actually take advatage of.
Indeed I've built a stereo 5578 and when searching on the part only ever saw 8 channel use, none were wanting to parallel multiple inputs for greater SnR.
As a multichannel studio recording mic ADC it fits the bill perfectly.
The 5397 is really what was supposed to be the replacement for the 5394A but that falls short. The interesting thing here is I'm not entirely sure why. Certainly it doesn't show -120dB thd across all sampling frequencies like we'd like but still.
It has flat noise like the 5394A. It has superior intrinsic noise specifications albeit with a capable input buffer. On paper, according to the datasheet, it should be able to manage -108 to -106dB thd+N when using their specified input buffer.
The 5394 specifies -110dB for 48/96 but only -94dB for 192... Granted the evm for the 5394 shows significantly better performance than that at -1dBfs but maybe there were consistency issues with the 5394 at higher sampling frequencies, or maybe the datasheet is in error.
Still going by AKMs -106dB claim for 192kHz, at -1dBfs, with that specified input buffer on the 5397. On paper, at least, it has superiority for higher frequency work.
It's troubling that AKM don't show the measured performance for the lower distortion input buffer and also troubling that it's apparently very hard to drive (syns own measurements showing that too).
But if, with the appropriate application, it can be made to show very similar performance to the 5394A at -1dBfs, with comparable noise, similar passband specifications, comparable high frequency noise, but, with lower power consumption and, at times (-10dBfs etc), slightly lower distortion, you can see why AKM dropped the 5394A. Of course that level of performance is yet to be seen for this device but apparently it can be done.
Any of the high performance ADCs mentioned will capture phono faithfully. You just need to make sure that any spurious pops and crackles don't overload the input to the ADC. On occasion phono has the potential to throw out very large signal amplitudes if your preamp is capable. You do not want these to find their way into the ADC and clip it.
The 5394A was discontinued because of the package. To migrate to a newer package would have required a new mask set. They just did not go down that route. Its a big, power hungry chip so not well suited for portable electronics. However its still the lowest distortion and flattest noise floor of the audio ADC's. There is a new Cirrus (Crystal) ADC that Lynx is using in their latest stuff. They claim the best ever but the marketing group must do that.
Yeah, they consume 1W or around there right? I completely understand why the market for that part was small.
Do you know what the new Cirrus part is? The CS5381 still seems to be the top ADC they make unless it's not on their product page or I missed it.
For measurements - may be, but for the sound - I'm personally do not hear any benefits of 4499 vs 4497 (except using as 4 channels), and the difference between 4497 and 4493 is not very big/ 4490 and 4493 are also very good DAC for hearing.
I used STM32L476 to do this, but I was also need some DSP operations between SPI and I2S. Otherwise - very cheap STM32F0 can be used.
But, need to program.
Were you able to do it easily with the functions provided by the STM32 HAL libraries? That's the attraction - I am too lazy to write anything myself these days if I can help it.
I recently interfaced I2S to the SPI port (TI mode) and a lot of the work was done in the CubeMX wizard. I just needed a few lines of assembler to read the data out of SPI and out again into the I2S ( using STM32G030). No interrupts, no DMA just poll, read and write. The tricky bit would be handling multiple sample rates as that's not dealt with in the CubeMX.
I used CubeMX+HAL just for CPU initialization (clocks, etc.), other was made with HAL but w/o CubeMX? and partially w/o HAL at all.
@5th element
Thank you for answering on my question. I do have experience with digitizing phono signals and so far digitized with the ADC section of my MDS-JA555ES with 16Bit/44.1kHz resolution so far. But I want to change away from this bulky solution, to go with a dedicated ADC feeding a 24/96 signal into my DEQX. The PCM4222 demo board was welcome as a not too pricey solution back at the time when I bought it...
Greetings,
Winfried