Any I/V opamps should be on the same ground plane as the dac chip, according to ESS. http://www.esstech.com/files/4514/4095/4306/Application_Note_Component_Selection_and_PCB_Layout.pdf
There can also be a little RF that comes out of a dac chip, which can get into the opamps and affect distortion. Reference designs don't always get everything completely right/optimal, at least at first for a new type of product. Even today there is discussion about eliminating the 'ESS hump' over at ASR, and it seems that it was an ESS distributor who performed some experiments to find a solution, apparently not ESS itself.
The point is that making a very good dac is still not quite something that can be assured on a first iteration. It can be a project in itself that takes time and effort.
Last edited:
Of course! I totally assume that our first iteration won't be good at all, but I'm just trying to reduce the number of things that are totally wrong about it. So, I take it that I should keep the opamps on the same board as the DAC chip, right?
Looking at Figure 83. External I-V Conversion Circuit Example (Mono mode) of 019001308-E-00, is it fair to assume that everything to the right of the I-V Conv. Circuits (the section within a dotted line) could go on the XLR output board, while everything within the dotted line would stay on the DAC board?
Looking at Figure 83. External I-V Conversion Circuit Example (Mono mode) of 019001308-E-00, is it fair to assume that everything to the right of the I-V Conv. Circuits (the section within a dotted line) could go on the XLR output board, while everything within the dotted line would stay on the DAC board?
Assuming you can protect the signals from noise sources and maintain clean grounds, that seems like a reasonable thing to try. Depending on how far and or where the signals might run, you might include a small-ish resistor in series with the I/V outputs to provide some isolation from wiring capacitance.
That makes sense. Thanks!
Looking at AKM's prototype on this article, I am left with one question: if the primary constraint is space (volume) instead of BoM cost, is it better to have as many discrete ultra low noise power supplies as possible (7 in AKM's prototype) with smaller capacitors, or less power supplies using larger capacitors?
Also, is it fair to say that the quality of the power supplies for the ADC board matters more than the quality of the power supplies for the XMOS board?
Looking at AKM's prototype on this article, I am left with one question: if the primary constraint is space (volume) instead of BoM cost, is it better to have as many discrete ultra low noise power supplies as possible (7 in AKM's prototype) with smaller capacitors, or less power supplies using larger capacitors?
Also, is it fair to say that the quality of the power supplies for the ADC board matters more than the quality of the power supplies for the XMOS board?
The eval board can run on +-15v only, however it will sound better with +-15v, +5v, and +3.3v (good to get the digital loads off the +15v rail). They don't have to be super supplies, as there are secondary regulators on the eval board for most things. More or less the same thing is true for Sabre dacs, use local regulators for things that need them, and general purpose power supplies as pre-regulators.
That might depend a lot on where the clocks are located. For a synchronous design everything should run on the same master clocks. Also, I2S signals coming out of the XMOS chip may lose their square edges pretty quickly due to interconnection capacitance. With some added complication that could be dealt with of course. They signals could be converted to LVDS at the XMOS, then converted back to LVCMOS near the dac, as one possible example. In such a case one might decide there is some benefit from using a 'reclocker' circuit to restore all the signals before going into the dac. Don't know if it matters all that much with AKM so long as setup and hold times are met. [On the other hand, getting a Sabre dac to lock in synchronous mode (DPLL = 0) might be more complicated, not so much because of setup and hold, but maybe because unspecified I2S noise sensitivity and or slow edges could cause issues.]
Last edited:
AK4499 Evaluation Board Rev.0
We just received the AK4499 Evaluation Board Rev.0 document for the AKD4499-B evaluation board. This will allow us to move forward with the design of our DAC brick. And the board includes both a 4-channel section and a mono channel section with the addition block that we want to implement. Yay!
We just received the AK4499 Evaluation Board Rev.0 document for the AKD4499-B evaluation board. This will allow us to move forward with the design of our DAC brick. And the board includes both a 4-channel section and a mono channel section with the addition block that we want to implement. Yay!
Last edited:
You still don't seem to understand this.
First, jitter on bit clock, word clock, and data do NOT matter one iota for AK4499 and similar parts. Your point is just plain wrong. The reason why is very simple and it's already been explained by me and Mark's quote from the datasheet. You only need to understand where the data enters the analog domain and what clock drives that action.
Second, the test by JA is done at the external interface level. The point of this test is to hamper the MCLK recovery of SPDIF receivers. It doesn't apply here.
I would use the 22/24 clocks as Mark suggests because 768 kHz is useless other than for bragging rights.
I'd also stick with OPA1612 rather than a budget replacement.
For the power supplies, I haven't seen their eval schematic, but I am pretty sure that LT3042 / LT3045 will work and deliver datasheet-level performance. The LT3042 specs are very similar to the dedicated ESS low noise regulator. I am sure Mark will disagree based on subjective criteria which I don't buy into, but you can make your own decision here. It's safest to copy the evaluation board. It feels like AKM copied the ESS approach, who knows if they tested other alternatives.
I would also make sure you put the LDOs on the boards near the loads. The regulators are most effective then.
I had to contact sales. But I did not have to sign any NDA.
I agree with you and Mark. DSD support is a must.
Agreed!
Would you mount them like LVODR does, or directly on the board?
I am confused by that statement. Which board(s) are you talking about? The PSU board or the different boards being supplied (DAC and XMOS)?
I’d mount them on your board directly. Any adapter is probably a compromise. I’d put whatever your “final stage” regulators are on their own boards near what they power. It’s fine to pre-regulate your power supply board outputs, but unless you use the remote sense functions, the regulators work best at the point of load.
When looking at the schematic and PCB design for the small 106mm × 96mm PCB on which the AK4499 is mounted as part of the AKD4499-B evaluation board, one thing becomes clear: it does not contain very many components, and we might be able to fit two of them side by side within one of our bricks. You can see that on this picture of the reference board. The components in questions are:
- 1 × AK4499 DAC
- 4 × OPA1612 OpAmps
- 4 large capacitors
- 18 medium capacitors
- A handful of small capacitors, resistors, and diodes
The big difference compared to what the original datasheet shows is that only 4 opamps are needed instead of 8, making the design a lot more compact. Moving forward, cramming everything within a 70mm × 35mm PCB will be our goal.
Just to be clear, you would mount them like that?
- 1 × AK4499 DAC
- 4 × OPA1612 OpAmps
- 4 large capacitors
- 18 medium capacitors
- A handful of small capacitors, resistors, and diodes
The big difference compared to what the original datasheet shows is that only 4 opamps are needed instead of 8, making the design a lot more compact. Moving forward, cramming everything within a 70mm × 35mm PCB will be our goal.
Just to be clear, you would mount them like that?
It is interesting as it looks like AKM used a lot of thru hole electrolytic capacitors around the
DAC rather than SMT caps. The regulators are a long way from the DAC chip too.
I actually designed & populated my own AK4499 test pcb & it is working fine. The AK4499
certainly draws a lot more current than the AK4497.
Markw4 - you were commenting about DSD rate detection.
I found the AK4499 auto detects DSD fine - same as the AK4493.
It’s good as with my AK4497 DAC I had to use info from the Amanero board to set the register for DSD which was extra work.
DAC rather than SMT caps. The regulators are a long way from the DAC chip too.
I actually designed & populated my own AK4499 test pcb & it is working fine. The AK4499
certainly draws a lot more current than the AK4497.
Markw4 - you were commenting about DSD rate detection.
I found the AK4499 auto detects DSD fine - same as the AK4493.
It’s good as with my AK4497 DAC I had to use info from the Amanero board to set the register for DSD which was extra work.
I was planning to use SMT capacitors because of space constraints. And the regulators are a long way from the DAC chip indeed. This is what got me confused with regards to chris719's comments (which made total sense nonetheless). Now, I'm not sure where I should put these regulators, but my hunch is to put them on the DAC boards themselves, which would allow me to use some through hole capacitors (vertically-mounted regulators are tall, therefore would make plenty of vertical space available for through hole capacitors).
Now, if we were to do that, would it make sense to put the power supply's opamps on a separate board as well, or should we mount them on the DAC boards directly?
By the way, thanks a ton for your posts on stereonet, they were really helpful.
Now, if we were to do that, would it make sense to put the power supply's opamps on a separate board as well, or should we mount them on the DAC boards directly?
By the way, thanks a ton for your posts on stereonet, they were really helpful.
ishizeno - so are you going to build a test/evaluation board first or go straight to the final design? I’m in full agreement with Chris, you should have the final regulators as close as possible to the DAC chip. I actually prefer using the Analog Devices ADM7150 series regulators PMIC - Voltage Regulators - Linear | Integrated Circuits (ICs) | DigiKey They are 16v max on the input so you can a bit of headroom which I think Markw4 felt sounded better. For the 1uf & 10uf caps around the chip I used MLCC smt caps but I think in my final version I will use Vishay Tantlum polymer as they are less microphonic than the MLCC. They are also available in ultra low esr too which is good. I’m glad you enjoyed the Stereonet post Actually now my prototype is working I will update that thread with some more good info & photos.
Last edited:
I'd put them on your PCB directly, no module, I mean. Just as it is on the LT evaluation board.
Thanks a lot for all the tips, very much appreciated.
I will get the evaluation board, play with it, then build some prototypes for the different boards that I need, make some measurements, then design the final configuration. And with a bit of luck, I will have access to an Audio Precision APX555 audio analyzer, so I should be able to make quite a few tweaks before settling on a design.
Got it! So now the next question is: LT3045 or ADM7150?
And by the way, the Vishay Tantalum Capacitors have been on my wishlist for a long, long time: I live in the Bay Area, and I've seen the Vishay sign on highway 101 so many times that I must use their products, one way or another. It's either these, or high-precision resistors for an R2R DAC...
Attachments
Definitely! One of our boards is ordinarily powered over PoE and now we measured that the THD+N is more than 30dB worse than when it is supplied by a laboratory supply.
But on the contrary one of our old systems was packed with costly components and low noise LDOs for each channels supply and the TDH+N wasn't better than it is with one of our new systems that uses switching regulators for each rail that is supplying all analog circuits within the system.
Most parts used are standard parts and therefore we were able to cut cost and space dramatically.
That sounds like a good plan - evaluation board then prototypes.
There are a lot of registry options so I guess it will take take a while to find the optimum setup.
Are you going to use the digital attenuator in the AK4499?
The Audio Precision APX555 will be very nice for testing.
Now how can I fit one of those in to my R&D budget - I wish!
There are a lot of registry options so I guess it will take take a while to find the optimum setup.
Are you going to use the digital attenuator in the AK4499?
The Audio Precision APX555 will be very nice for testing.
Now how can I fit one of those in to my R&D budget - I wish!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.