Since Mouser updated the delivery of ESS DACs to 03/13, and AKM is dead for the foreseeble future, ES9038PRO it is.
I've started by designing the power supply, with some extra feature: feeding from USB 5V only (using a boost switcher) switcher tracking for the IV stage +/-15V to minimize dissipation, and power sequencing (per the ES9038PRO datasheet). Schematic is attached. Currently designing a 2 layer board, it will be separate from the DAC board to minimize chances to rework a big expensive 4 layers board. In the final version I would expect each regulator to be close to the pin(s) it is feeding.
I've started by designing the power supply, with some extra feature: feeding from USB 5V only (using a boost switcher) switcher tracking for the IV stage +/-15V to minimize dissipation, and power sequencing (per the ES9038PRO datasheet). Schematic is attached. Currently designing a 2 layer board, it will be separate from the DAC board to minimize chances to rework a big expensive 4 layers board. In the final version I would expect each regulator to be close to the pin(s) it is feeding.
Attachments
Interesting observation. It makes me wonder if the CS43198 has the same
characteristics but with better DR. Of note, CS43198 has not really gained
the cult following of other (ESS / AKM) DACs but it is a very high performance
part from the measurements I've seen. Easily as good as 4499 or 9038.
TCD
Unfortunately JLCPCB started charging extra for any board with vias that are smaller than 0.4mm (imagine using .4mm vias for a board with 0.5mm pitch ICs). And the charge a lot, the price of a 3x4" 4 layers pretty much doubles. Still worth for a final board, but as long as it's in development is an useless expense.
Maybe, but not according to the datasheets. But for sure from a "sound quality" perspective these chips are all way under any known thresholds. Since this is about instrumentation I can safely ignore any subjective malarkey.
From what I have seen, outperforms data sheet by significant margin.
One FFT showed nothing above -140 at 0dBFS, another was not far behind.
It does appear to have more wide band noise than, in particular AKM.
Horses for courses.
TCD
Yeah I’m not sure, but I think it’s a little unfair to compare CS43198 to 4499 or 9038PRO. It’s definitely targeting the lower power segment. It is a nice part, but there are some compromises at high sample rates and it has no external filter mode. The measurements of some dongles using it or CS43131 seem to show an earlier rise of out of band noise vs something like AK4493. Looks like you beat me to it.
. Unrelated, but I find it amusing that (some) Hi-Fi people complained for years about Pro Tools interfaces, and yet when we look under the hood we find AK5394A and CS4398. Seems like they knew what they were doing when they selected them as both have stood the test of time and exceed their datasheet performance in several products.
Last edited:
Can I say “I don’t care”? You would think the same if you would question yourself why would Cirrus underspec this part, if they would have a chance to match or beat their competitors, at least in some respects.
And are you sure the AKM and ESS parts don’t also “outperform data sheet by significant margin“?
Like everybody else, I’m waiting for the link to those CS43198 measurements.
From what I have seen, most outperform the data sheet with the best
implementations, some by a small margin, some by more. I presumed you
would have observed this. AK5394 ADC is a classic example. ASR (and other
sites) have plenty of numbers to peruse.
TCD
I have had mixed experiences with the CS43198.
On the evaluation board I have measured THD below -130 dB at 0 dBFS.
I have made my own design with the CS43198, but the THD is much higher, around -109 dB. So far I have not been able to figure out why.
Is it variation from chip to chip?
Is it my layout?
Are there differences in the components used?
The schematic is very similar to the one used on the evaluation board except that a lot of the interface stuff has been removed, leaving only the CS43198 and the passive componens around it + some interface and control. I feed the chip with an I2S signal and a clean MCLK. I used the RTX6001 as a test system, with the DAC board replaced by the dual CS43198 design.
Regarding OOB noise it does have a higher noise level above 40 kHz. It can be reduced by using the "Wide Band Flatness Mode", but this gives some other trade offs regarding distortion.
Does anyone else have experience with the CS43198 and can comment on the big difference seen in the distortion?
On the evaluation board I have measured THD below -130 dB at 0 dBFS.
I have made my own design with the CS43198, but the THD is much higher, around -109 dB. So far I have not been able to figure out why.
Is it variation from chip to chip?
Is it my layout?
Are there differences in the components used?
The schematic is very similar to the one used on the evaluation board except that a lot of the interface stuff has been removed, leaving only the CS43198 and the passive componens around it + some interface and control. I feed the chip with an I2S signal and a clean MCLK. I used the RTX6001 as a test system, with the DAC board replaced by the dual CS43198 design.
Regarding OOB noise it does have a higher noise level above 40 kHz. It can be reduced by using the "Wide Band Flatness Mode", but this gives some other trade offs regarding distortion.
Does anyone else have experience with the CS43198 and can comment on the big difference seen in the distortion?
How about LT8582 as boost/inverter? Available output current would be higher (500mA) although not from USB. I've used LT3471 before but haven't tried VIOC.
It's another possibility, the decision for the ADP5070 part was strictly related to it's TSSOP outline availability. I have no problems soldering DFNs, but at least in the development stages TSSOPs are much more convenient.
The ADP5070 inverter switch is 0.6A which I would think it will be enough to feed the IV stages, but just in case there is the 1.2A version, pin to pin compatible, the ADP5071.