The eval board manual says to sequence the external power supplies, is all.
IIRC, AK4499 datasheet says to sequence the power supplies, or else to bring them up at the same time.
For both dacs the manufacturers say to sequence the clocks after the power supplies. Doesn't seem to matter in practice with AK4499.
So far haven't been able to get DSD working with the USB board that the eval board was designed to accommodate. ASIO driver crashes and HQ Player freezes when I hit play.
Will try rigging up I2SoverUSB to the alternate I2S port on the eval board. Eval board manual says in DSD mode to apply BLCK to both the MCLK and the BLCK pins, which appears to imply non-phase mode DSD operation.
IIRC, AK4499 datasheet says to sequence the power supplies, or else to bring them up at the same time.
For both dacs the manufacturers say to sequence the clocks after the power supplies. Doesn't seem to matter in practice with AK4499.
So far haven't been able to get DSD working with the USB board that the eval board was designed to accommodate. ASIO driver crashes and HQ Player freezes when I hit play.
Will try rigging up I2SoverUSB to the alternate I2S port on the eval board. Eval board manual says in DSD mode to apply BLCK to both the MCLK and the BLCK pins, which appears to imply non-phase mode DSD operation.
Last edited:
It is not possible at eval boiard, because it has only one external power for DAC: +15V (CON5, and CON4 for GND).
The second external power +/-15V (CON1-2-3) is for OP only, not for the DAC itself.
For the dac itself, sequencing can be controlled by regulator filter cap choices. Don't know if they are doing that or not, haven't looked at the schematic with that in mind. Otherwise, it must be okay to bring the rails up all at once. Doing it specifically out of sequence might not produce reliable operation...or worse.
Last edited:
Capacitors are not a good substitute for a supervisor, unless you're making a disposable and need to shave 3 cents off it. Wouldn't be the first time I've seen an evaluation board that has obvious didn't-read-the-datasheet mistakes, though.
Agreed.
Regarding the eval board, 40-hours have passed since it was powered on. I am willing to say a few preliminary things about it. The SE and XLR outputs are connected to a Neurochrome HP-1 feeding a pair of Audeze LCD-X. Prefer the XLR sound, which is more clear to my perception. Meaning the differential summing is probably similar to another Neurochome HPA I have, RF input filtering followed by LME49720 differential summing in that case.
Its roughly around where AK4499 eval board was at this stage. Currently running on three bench supplies which are plugged into a power conditioner with some combination of common mode chokes for each supply and two of the supplies on back to back isolation transformers. Third supply is for the USB board. Know from experience that this is not ideal powering, need to get it all running from one R-core, need to bypass the I2S isolator, need it all in a shielded steel case, and probably need to try OPA1612. Then maybe a better idea of how good the chip is. Better clocking usually helps too; easiest way to get it better quickly would be by changing the USB board to I2SoverUSB. Of course, better than I2SoverUSB is also possible; current version of AK4499 uses I2SoverUSB with external Crystek clocks (which are implemented a little cleaner than I2SoverUSB clocks).
Regarding the eval board, 40-hours have passed since it was powered on. I am willing to say a few preliminary things about it. The SE and XLR outputs are connected to a Neurochrome HP-1 feeding a pair of Audeze LCD-X. Prefer the XLR sound, which is more clear to my perception. Meaning the differential summing is probably similar to another Neurochome HPA I have, RF input filtering followed by LME49720 differential summing in that case.
Its roughly around where AK4499 eval board was at this stage. Currently running on three bench supplies which are plugged into a power conditioner with some combination of common mode chokes for each supply and two of the supplies on back to back isolation transformers. Third supply is for the USB board. Know from experience that this is not ideal powering, need to get it all running from one R-core, need to bypass the I2S isolator, need it all in a shielded steel case, and probably need to try OPA1612. Then maybe a better idea of how good the chip is. Better clocking usually helps too; easiest way to get it better quickly would be by changing the USB board to I2SoverUSB. Of course, better than I2SoverUSB is also possible; current version of AK4499 uses I2SoverUSB with external Crystek clocks (which are implemented a little cleaner than I2SoverUSB clocks).
Wouldn't shielding be against the "spirit" of the way the ground plane on that board is designed?
They all need shielding where I live, up on top of hill. Some days in clear weather one can see the Coastal Mountain Range peaks ~100mi away across the Sacramento Valley. In the valley there are airports, a military base, etc. The RFI all comes from off in that general direction and shielding makes clearly noticeable difference.
Finally have Rohm eval board running on I2SoverUSB going into the I2S 'test port' pin header that goes direct to the dac chip. Also removed 5532s and installed sockets with OPA1612s on dip adapters. In addition, power is now all on R-core transformers.
Haven't put it in the steel shielding case yet. Want to test out both DSD modes first, phase mode, and BCLK feeding the MCLK input.
Only got it going about half an hour ago, but its starting to sound much better to me. Can tell there will need to be more done, but at least its a little progress.
Haven't put it in the steel shielding case yet. Want to test out both DSD modes first, phase mode, and BCLK feeding the MCLK input.
Only got it going about half an hour ago, but its starting to sound much better to me. Can tell there will need to be more done, but at least its a little progress.
The old Chinese Amanero boards were stuck at an old firmware revision which limited their capabilities and which would brick if new firmware was applied. Since then I have heard the newer Chinese boards may have gotten past that problem. As to maximum sample rate, most of them still rely on real Amanero firmware and drivers. So, maybe whatever a real Amanero can do.
That said, there is what appears to be an Amanero clone with Accusilicon AS318-B clocks, for which they have apparently gotten their own ASIO driver. What the ad copy says about that one includes (from back before the got their own ASIO driver):
This interface also opens configuration mode for advanced players. The MCLK output frequency can be individually configured for any format. From 64Fs to 512FS, from 2.8224MHZ to 49.152MHz, let the DAC chip play excellent performance. You can also write VID and PID and device strings arbitrarily to adapt to any standard USB2.0 audio driver.
The supported formats and sampling rates are as follows: 32-bit PCM (backward compatible with 24 and 16-bit) formats support 44.1KHz, 48KHz, 88.2KHz, 96KHz, 176.4KHz, 192KHz, 352.8KHz, 384KHz, 705.6KHz, 768KHz, 1411.2KHz and 1536KHz sampling rates. The DSD format supports DSD64, DSD128, DSD256, DSD512, and DSD1024, and is transmitted losslessly using DoP (DSD64 ~ 512) or Native (DSD64 ~ 1024) protocols."
That said, there is what appears to be an Amanero clone with Accusilicon AS318-B clocks, for which they have apparently gotten their own ASIO driver. What the ad copy says about that one includes (from back before the got their own ASIO driver):
This interface also opens configuration mode for advanced players. The MCLK output frequency can be individually configured for any format. From 64Fs to 512FS, from 2.8224MHZ to 49.152MHz, let the DAC chip play excellent performance. You can also write VID and PID and device strings arbitrarily to adapt to any standard USB2.0 audio driver.
The supported formats and sampling rates are as follows: 32-bit PCM (backward compatible with 24 and 16-bit) formats support 44.1KHz, 48KHz, 88.2KHz, 96KHz, 176.4KHz, 192KHz, 352.8KHz, 384KHz, 705.6KHz, 768KHz, 1411.2KHz and 1536KHz sampling rates. The DSD format supports DSD64, DSD128, DSD256, DSD512, and DSD1024, and is transmitted losslessly using DoP (DSD64 ~ 512) or Native (DSD64 ~ 1024) protocols."
Last edited:
I use I2SoverUSB for that.
Might be some more info about configuration options at the website for the Accusilicon-clocked Chinese USB board: 下载 Download – XING
EDIT: USB digital interface AS318B PCM1536 DSD1024 compatible with Amanero Italy XMOS to I2S|Digital-to-Analog Converter| - AliExpress
Might be some more info about configuration options at the website for the Accusilicon-clocked Chinese USB board: 下载 Download – XING
EDIT: USB digital interface AS318B PCM1536 DSD1024 compatible with Amanero Italy XMOS to I2S|Digital-to-Analog Converter| - AliExpress
Last edited:
By the way, got DSD256 working using BCLK for MCLK and for BCLK. Shielding is in place too. Sounding better...
Seems likely the analog output coupling cap design, along with the AVCC supply, are the most likely places to find another level of improvement. (Differential summing is still done in Neurochrome HP-1.)
Seems likely the analog output coupling cap design, along with the AVCC supply, are the most likely places to find another level of improvement. (Differential summing is still done in Neurochrome HP-1.)
I see, thanks. I have no problem buying a real Amanero to test but I had ruled it out because it only officially supports 384 kHz that I see.
Did a subjective listening test at this point with two listeners comparing notes. A little rolling of opamps ranks them in the following order by preference:
1. TI NE5532
2. OPA1612
3. OPA1656
NE5532 was the most balanced tonally with good bass and midrange, and with fairly clean cymbal sounds.
OPA1612 was more 'splashy' on cymbals (more distorted cymbal sound).
OPA1656 gave cleaner cymbals and a lot of detail, but midrange and bass were thin and understated.
Again, this was with Neurochrome HP-1 acting as differential summing amp. Since its input RF filter uses ferrites and I am leery of the possibility of hysteresis distortion, maybe some difference at that point. Sound differences might not be solely due to the opamps is all I'm thinking.
Regarding the dac chip itself, its still too early to tell very much but seems like it might better at reproducing low level room decay sounds than AK4499.
More later.
1. TI NE5532
2. OPA1612
3. OPA1656
NE5532 was the most balanced tonally with good bass and midrange, and with fairly clean cymbal sounds.
OPA1612 was more 'splashy' on cymbals (more distorted cymbal sound).
OPA1656 gave cleaner cymbals and a lot of detail, but midrange and bass were thin and understated.
Again, this was with Neurochrome HP-1 acting as differential summing amp. Since its input RF filter uses ferrites and I am leery of the possibility of hysteresis distortion, maybe some difference at that point. Sound differences might not be solely due to the opamps is all I'm thinking.
Regarding the dac chip itself, its still too early to tell very much but seems like it might better at reproducing low level room decay sounds than AK4499.
More later.
Last edited:
Strange.. OPA1656 in my application sounds meaty, with a lot of weight. If the dac supports that. That aspect changes a lot, depending on clock handling, for example, but also other parts.
What passive parts do you have in there, by the way?
And Peufeu is rigth, - both these chips (1656 & 1612) are fast, correct bypassing is needed..?
What passive parts do you have in there, by the way?
And Peufeu is rigth, - both these chips (1656 & 1612) are fast, correct bypassing is needed..?
