Mark, you don't seem to like Amanero..
Permit me to point out that :
It uses 22/24MHz clocks, on the contrary to the Jls boards (and this is why I had chosen it for my Japanese dac kit);
It can be made to use external clocks by a simple firmware change; (very rare feature)
It can be modded (cutting the USB feed) for local power supply (though when used from a better renderer, like an Allo, that is already feeding it well);
The dac kits worth the mention (like Yanasan's) are already implementing galvanic isolation, it's needless to do it twice. And there exist different galvanic isolation boards developed for Amanero - Diyinhk, twisted pear, etc.
After these, I find it working smooth and robust, recently tried it even with 512..
And obviously very good sound and testing low jitter.
Ciao, George
Permit me to point out that :
It uses 22/24MHz clocks, on the contrary to the Jls boards (and this is why I had chosen it for my Japanese dac kit);
It can be made to use external clocks by a simple firmware change; (very rare feature)
It can be modded (cutting the USB feed) for local power supply (though when used from a better renderer, like an Allo, that is already feeding it well);
The dac kits worth the mention (like Yanasan's) are already implementing galvanic isolation, it's needless to do it twice. And there exist different galvanic isolation boards developed for Amanero - Diyinhk, twisted pear, etc.
After these, I find it working smooth and robust, recently tried it even with 512..
And obviously very good sound and testing low jitter.
Ciao, George
Hi George,
I agree Amanero should be fine if all the steps you describe are done in a good implementation. I tried some popular reclocking/isolation boards and was only happy with one, so caution is advised for beginners. Also, there doesn't seem to be any support available for Amanero from the manufacturer, which is a negative in my view.
On the other hand, I2SoverUSB sounds better as-is without corrective steps required. However, although its onboard clocks are pretty good, like Amanero it can benefit from better external clocks. At that point some solution is required if using AKM dacs, given the different clock frequencies, which starts to make it look more similar in complexity to Amanero solutions.
I agree Amanero should be fine if all the steps you describe are done in a good implementation. I tried some popular reclocking/isolation boards and was only happy with one, so caution is advised for beginners. Also, there doesn't seem to be any support available for Amanero from the manufacturer, which is a negative in my view.
On the other hand, I2SoverUSB sounds better as-is without corrective steps required. However, although its onboard clocks are pretty good, like Amanero it can benefit from better external clocks. At that point some solution is required if using AKM dacs, given the different clock frequencies, which starts to make it look more similar in complexity to Amanero solutions.
Which is a consequence of its output not being reclocked. On its own it sounds atrocious, but of course strictly imho. Not really seeing the point of such a device when the Jlsounds is readily available, extremely affordable, outputting a myriad of possible formats, having onboard isolation and reclocking and accepting external clocks.
What do you think of this ?
SMSL M200 AK4497EQ DAC Bluetooth 5.0 Support 32Bit/768KHz DSD512 Decoder with Remote Control
SMSL M200 AK4497EQ DAC Bluetooth 5.0 Support 32Bit/768KHz DSD512 Decoder with Remote Control|Digital-to-Analog Converter| - AliExpress
You can find it on ebay
Is this worth ?
SMSL M200 AK4497EQ DAC Bluetooth 5.0 Support 32Bit/768KHz DSD512 Decoder with Remote Control
SMSL M200 AK4497EQ DAC Bluetooth 5.0 Support 32Bit/768KHz DSD512 Decoder with Remote Control|Digital-to-Analog Converter| - AliExpress
You can find it on ebay
Is this worth ?
Usually you can see the effects of the quality reference supply in the datasheet or at least in notes for the demo board. Some of the effects are really subtle and may not be audible (especially depending on source material, associated equipment and auditioner) but some can be. Layout can create unwanted interactions on the supply grounds since ground currents and ground management are critical on these super high performance DAC's.
Hard/impossible to tell what it sounds like from here. They don't even show a pic of the insides. Most of the stuff like that is sold on the basis of (1) price, and (2) visual appearance.
So splitting DAC -> I-V boards can be tricky, especially with 4499's 'sense' pins.
JW was suggesting IDC type connector, I can see a lot of parallel Gnd pins...
TCD
If you are going to the effort of using something so challenging I think you have no choice but to merge everything on a 6 or 8 layer board. It's a current out dac. Figure at least 140 dB down from its peak current to get a sense of how little or stray ground current will compromise your efforts.
Why go cheap? You can't easily satisfy all requirements and get good EMC performance with 4 layers. I find it funny how audiophiles want to spend so much money on components and then do layouts to minimize layer count at all costs and use cheap PCB services with low Tg / low quality FR4.
Oh, I know. What does it really take to do it right, 11-layers or so?
Besides I didn't say I was cheap, but I don't know about other people that might be interested in a board. We're just talking, not making any decisions at this point. Already we have complaints about cost.
Sorry, I didn't mean to imply you personally were cheap. Could it be done in 4 layers entirely and meet all requirements and pass all EMC testing? Probably, given enough time and revisions. Given that it's probably not going to be mass produced in high volumes and that time is money, I'd just go with 6 or 8 layers to start. I don't know about 11... odd number and all.
I'm interested to hear why you think meeting EMC guidelines would be in any way difficult with such a design. We are, after all, talking about low power, low signal manitude transitions throughout all the design. Where methods to reduce rise times can be used without detriment to the devices performance.
Plus the plethora of cheap EMC compliant hifi products from a multitude of big consumer names, built around 2 layer boards and using clock of a similar frequency?
Plus the plethora of cheap EMC compliant hifi products from a multitude of big consumer names, built around 2 layer boards and using clock of a similar frequency?
I didn't say it would be difficult necessarily to just pass compliance testing, especially if you are in a metal box. Susceptibility and immunity are the other side of that coin and these layouts could have trouble in ESD testing or with radiated immunity depending on what standards they are tested to and what the pass/fail criteria are.
Still, it's really easy to build a product that will fail radiated emissions, especially if it has to be tested with cables attached. Do you think all these boutique manufacturers using MCUs, XMOS chips, etc. have taken the care to lower the risetimes on all non-critical signals?
The big consumer names also have the time and budget to optimize their layouts and re-spin if necessary. They also do pre-compliance testing. I am talking about "high-end" boutique outfits.
I would be surprised if standard two layer boards, with uninterrupted ground planes, would fail EMC testing with the standard fair of low power MCUs and USB interfaces. I mean yes, if you go out of your way to create bad layouts and don't observe commonly accepted practices for signal routing, via PCB traces, or cables, then sure you'll get issues but I don't think anyone is going to be doing that here?
At least I would hope so.
Having said this I have seen a couple of examples from hobbyists and the AK4499 that have performed terribly or looked to have been laid out for practicality of trace layout, rather than performance.
At least I would hope so.
Having said this I have seen a couple of examples from hobbyists and the AK4499 that have performed terribly or looked to have been laid out for practicality of trace layout, rather than performance.
Its beyond challenging to make a 2 layer board with a real ground plane that will work for the high pinout chips. 4 layers possible. 6 layers practical. Today the incremental cost isn't much in light of the cost of the parts. However noting is more important than the skill and experience of the layout guy.