If the digital filter is enabled at the oversampling (AK4137) stage then at the AK4490:1 digital filter should be setup as External?
or
2 digital filter should be setup as the same as at AK4137?
I'm still unable to initialize the DACs properly and try the External filter setup. My 3-wire control sequence looks perfect / according to the AKM spec, but looks like nothing get programmed well: there is no output signal.
I have read this thread with much interest and am keen to know how your Ak4490 board performed with both the Amanero usb and the AK4317
I have a single ak4490 with XMOS usb and the Chip outputs directly driving through a Jensen JT-10KB-D (around US$70) transformer with excellent results
The things to note with transformer coupling are:
Flat phase response, even though the Frequency Response is down 3dB at 200kHz (you can't get this with RC based opamp filtering)
High enough input impedance to drive from AK4490 DAC direct (no opamps) -input Z = 40kohm with 2.5 kohm primary resistance
and the lower the driving source impedance, as with a system like yours, the better
I have a single ak4490 with XMOS usb and the Chip outputs directly driving through a Jensen JT-10KB-D (around US$70) transformer with excellent results
The things to note with transformer coupling are:
Flat phase response, even though the Frequency Response is down 3dB at 200kHz (you can't get this with RC based opamp filtering)
High enough input impedance to drive from AK4490 DAC direct (no opamps) -input Z = 40kohm with 2.5 kohm primary resistance
and the lower the driving source impedance, as with a system like yours, the better
I like how it sounds in PCM mode (compared to ES9018). I'm using original Amanero board, Firmware 1.074 (Sync), Windows 10 WASAPI. Every type of the digital filter works and every oversampling rate up to 384K. At the inverting stage using ADA4898-1. Also, compared to original build I've replaced JapanRadio 3.3V regulators into LT3045.
So far the only problem is that ASIO does not work at the moment. I remember it worked once.
Need a clue with DSD. I'm able to play DSD or PCM converted to DSD 2.8M. But it sounds not that nice like PCM, there are the cracks at the beginning of the tracks an it takes a while to lock on. Also, the DSD 5.6 and 11.2 are very distorted.
The only thing I suspect it the low quality of the oscillators on the AK4137 board (Master Clock to 4490 goes from there). Did anybody tried to replace them on that New AK4137 DAC SRC flagship high end audio 786K 32Bit DSD 256 DSD IIS conversion-in Connectors from Lights & Lighting on Aliexpress.com | Alibaba Group
?
Or what would be other bug in there?
Thanks!
The only thing I suspect it the low quality of the oscillators on the AK4137 board (Master Clock to 4490 goes from there). Did anybody tried to replace them on that New AK4137 DAC SRC flagship high end audio 786K 32Bit DSD 256 DSD IIS conversion-in Connectors from Lights & Lighting on Aliexpress.com | Alibaba Group
?
Or what would be other bug in there?
Thanks!
Does the AK4490 have any way of knowing what kind of DSD you are feeding into it? It needs to know that in order to properly process the signal.
Thanks for that evaluation
I have another AK4490 in the shed and I have only used it in parallel mode
I will have to connect it up in software mode to try DSD for which I have an early version of XMOS usb receiver flashed by DIYINHK - hope it works
As for your system I would expect the problems you are having are software and not hardware ie the register set up in the AK4490
Refer to the datasheet pages 53 to 59 for an explanation of bits that need to be set for correct operation
eg DSDD bit set to 1 for DSD direct or native DSD - which bypasses digital processing - but it would appear that the dsd rate also needs to be selected as per table 16 - which is ok if you can feed the dac with a fixed dsd rate via the AK4137 or a software player such as daphile
Someone might have already done this somewhere???
I have another AK4490 in the shed and I have only used it in parallel mode
I will have to connect it up in software mode to try DSD for which I have an early version of XMOS usb receiver flashed by DIYINHK - hope it works
As for your system I would expect the problems you are having are software and not hardware ie the register set up in the AK4490
Refer to the datasheet pages 53 to 59 for an explanation of bits that need to be set for correct operation
eg DSDD bit set to 1 for DSD direct or native DSD - which bypasses digital processing - but it would appear that the dsd rate also needs to be selected as per table 16 - which is ok if you can feed the dac with a fixed dsd rate via the AK4137 or a software player such as daphile
Someone might have already done this somewhere???
I assume that by the DSD mode bit you are referring to bit D7 control 3
This would then proceed with the default setting of 0,0 for DSDSEL1 AND DSDSEL0 = 64DSD
For different DSD rates you need to set control registers as 0 and 1 = 128DSD
and 1 and 0 =256DSD
DSDSEL1 IS BIT D0 OF CONTROL 8 (09H)
DSDSEL0 IS BIT D0 OF CONTROL 5 (06H)
which is all a bit much, so thats why I suggested upsampling everything to 256 DSD and the registers
do not need to be re programmed
Only my suggestions because at this stage I do not have an I2C programmer
Oh by the way the bit I was referring to in my previous post was DSDD bit which is bit D1 of control5 (06H) which is bypassing digital processing so becomes similar to NOS DACs
This would then proceed with the default setting of 0,0 for DSDSEL1 AND DSDSEL0 = 64DSD
For different DSD rates you need to set control registers as 0 and 1 = 128DSD
and 1 and 0 =256DSD
DSDSEL1 IS BIT D0 OF CONTROL 8 (09H)
DSDSEL0 IS BIT D0 OF CONTROL 5 (06H)
which is all a bit much, so thats why I suggested upsampling everything to 256 DSD and the registers
do not need to be re programmed
Only my suggestions because at this stage I do not have an I2C programmer
Oh by the way the bit I was referring to in my previous post was DSDD bit which is bit D1 of control5 (06H) which is bypassing digital processing so becomes similar to NOS DACs
You might want to take a look around pages 12-14 of the following and see what you think: http://www.cranesong.com/solaris_manual_version_1.pdf
Already convinced 10Hz is critical, myself. Amazingly or not, Allo Katana dac has lower measured jitter than Benchmark DAC-3 (according to ASR), and DAC-3 and Crane Song Solaris are said to sound very similar to one another. Solaris uses an AKM dac chip, and DAC-3 uses ESS. But they both use SRC4392 clocked at 27MHz to reduce jitter by upsampling, and then use external interpolation filtering.
Anyway, the amazing thing is that the jitter difference between Katana and DAC-3 is audible. I know what the sound of it is since I was able to get my ES9038Q2M dac project to jitter as low as Katana for short periods of time, as measured by DPLL bandwidth stability (which is a function of incoming jitter).
Jitter at 12kHz is a meaningless number for audio, since jitter tends to increase, often drastically, as offset frequency goes down (similar to what we see with 1/f noise). I will only use clocks that are as low jitter as I can get (at reasonable cost of course <$50) down at 10Hz and 1Hz. If they aren't specified down there, I have no interest in them. And, that is only a starting point, then I try to get jitter even lower with upsampling. It turns out not too trivial to get really good sound quality, IME and IMHO. Others may differ in their opinions, of course. (Master mode like Katana uses helps a lot, but is not good for a general purpose dac like we more often are trying to make.)
Already convinced 10Hz is critical, myself. Amazingly or not, Allo Katana dac has lower measured jitter than Benchmark DAC-3 (according to ASR), and DAC-3 and Crane Song Solaris are said to sound very similar to one another. Solaris uses an AKM dac chip, and DAC-3 uses ESS. But they both use SRC4392 clocked at 27MHz to reduce jitter by upsampling, and then use external interpolation filtering.
Anyway, the amazing thing is that the jitter difference between Katana and DAC-3 is audible. I know what the sound of it is since I was able to get my ES9038Q2M dac project to jitter as low as Katana for short periods of time, as measured by DPLL bandwidth stability (which is a function of incoming jitter).
Jitter at 12kHz is a meaningless number for audio, since jitter tends to increase, often drastically, as offset frequency goes down (similar to what we see with 1/f noise). I will only use clocks that are as low jitter as I can get (at reasonable cost of course <$50) down at 10Hz and 1Hz. If they aren't specified down there, I have no interest in them. And, that is only a starting point, then I try to get jitter even lower with upsampling. It turns out not too trivial to get really good sound quality, IME and IMHO. Others may differ in their opinions, of course. (Master mode like Katana uses helps a lot, but is not good for a general purpose dac like we more often are trying to make.)
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.