Phase rotation of 90 degrees is not phase shift. Just look at the 90 degrees phase rotation of a square wave in my post. The hearing insensitivity of phase refers to phase shift (or change in phase angle). If you rotate the phase 90 degrees you have a different signal altogether so your claim that the Purifi files debunk phase-insensitivity myth is debunked.
The problem with having MCU and DAC on the same PCB is that MCUs may emit noise to the ground plane. Having separate regulator for MCU and DAC may not mitigate that problem which is why digital isolators are often used. Regarding LT8619 I don't really see the point of using a switching regulator unless your MCU is a current hog.
Disagree. Phase rotation describes a process that can be used for changing phase. Phase can also be changed by, say, for example, a minimum phase passive filter. Either way some process or other can produce phase change. Changing phase always alters a time domain transient waveform, and its that change of time-domain waveform (or waveshape) that changes the sound. We don't even need to know the process that produced it. We only need to know the end result change in waveshape.
The idea of having a buck is more for having some headroom for other things to come and I have read that it is advised to step down the voltage before using an LDO for more efficiency. If it is not needed or adviced I wont do it.
In that case I will design separate LDOs for the MCU and the DAC, I mentioned above that I will try to have the MCU + USB on a daughter board soldered onto the carrier, one way or the other, GND would be shared though. I don't have space constraints and can separate analogue from digital components, A 4 layer stackup with 2 inner GND layers should help too and I think my power demands are low enough that I get away with routing PWR. The critical area will be the MCU though if I try to keep all components on one side.
In that case I will design separate LDOs for the MCU and the DAC, I mentioned above that I will try to have the MCU + USB on a daughter board soldered onto the carrier, one way or the other, GND would be shared though. I don't have space constraints and can separate analogue from digital components, A 4 layer stackup with 2 inner GND layers should help too and I think my power demands are low enough that I get away with routing PWR. The critical area will be the MCU though if I try to keep all components on one side.
How do you plan to power your board and what is the input voltage? Switching regulators can be another noise source. I'm quite sure you will end up with some revisions of the board anyhow so no need to try to make it future-proof and complicated on the first revision.
The general plan is to use a 12V AC/DC supply, again for the sake of simplicity. This will feed both, the Amp and the DSP. In a standalone mode, the DSP board could also be supplied by a lower voltage. So the question is if it is good practice to step down the voltage to 5V before using an LDO for the 3v3 rail. This would also allow the possibility of using the USB VBUS for driving the DSP, not sure if that might introduce more problems though.
The consumption of the MCU is around 200 mA, the ES9033 seems not to be very hungry with < 50mA in HW mode 8. I was therefore aiming at a 1.5A 5V buck converter.
The Amps will have its own Power regulator, so if it's fine to go with just an LDO, perfect.
The consumption of the MCU is around 200 mA, the ES9033 seems not to be very hungry with < 50mA in HW mode 8. I was therefore aiming at a 1.5A 5V buck converter.
The Amps will have its own Power regulator, so if it's fine to go with just an LDO, perfect.
Once again agree with Mr. bohrok, this time regarding voltage regulation.
Hope it will okay to raise a new question: What test equipment will be available for assessing performance, troubleshooting, etc.?
Also, before going too far with board layout, is there a test plan in mind? Seems to me it might be nice to have some solder jumper pads so the dac portion to be tested with a linear supply and with a common USB board. That would take other variables, such as a shared power supply with a class-D amp, out of the equation as a starting point. Then perhaps introduce one change, say, a dsp chip, and see if any noise problems arise with the dac that weren't present initially. And so on. I suggest that sort of thing because the OP is new to some of this, so it might be useful to be able to assess performance incrementally.
Hope it will okay to raise a new question: What test equipment will be available for assessing performance, troubleshooting, etc.?
Also, before going too far with board layout, is there a test plan in mind? Seems to me it might be nice to have some solder jumper pads so the dac portion to be tested with a linear supply and with a common USB board. That would take other variables, such as a shared power supply with a class-D amp, out of the equation as a starting point. Then perhaps introduce one change, say, a dsp chip, and see if any noise problems arise with the dac that weren't present initially. And so on. I suggest that sort of thing because the OP is new to some of this, so it might be useful to be able to assess performance incrementally.
So, to be sure if I understood correctly,
changing the pre-regulator later on is easier than changing the whole PWR section.
- having a pre-regulator is fine, but for a first iteration, use a simple one like the 7805.
- using different LDOs with 3v3v for MCU and DAC is generally not necessary, but could help. Is it advised though?
changing the pre-regulator later on is easier than changing the whole PWR section.
ESS recommended exactly that in a document they used have for download on their website. A copy is attached below. It is suggested to study it carefully.
EDIT: Copy of ES9038Q2M evaluation board schematic attached in case any interest in seeing it.
Last edited:
As I have said a couple of times best option is to have digital isolation between MCU and DAC which also requires separate regulators on both. If you insist on using them without digital isolation then I would still recommend separate regulators. As a bare minimum one regulator could be used with ferrite bead on the MCU to isolate it from DAC.
My equipment isn't that good but I might invest a bit more if it is necessary. Right now I own a simple 4ch oscilloscope and rather cheap lab PSU. A function generator and a better PSU will probably be my next buy. I'm looking into something like the Keysight 1000x series in the future.
I have planned to expose all mode pins of the DAC with jumpers for hardware configuration.
Having the power rails configurable seems like a good idea too.
Why have a separate analogue and a digital section on one board at all then? That means that creating 2 separate boards will be probably the better option. For the first iteration, this might be a good idea, reducing problems, and combining them, later on, seems not practical to me.
Instead of function generator a decent soundcard may be more useful.
In ES9033 I2C uses same pins as HW mode so it may be better to cater for that as well.
I'm not sure I understand what you refer to with analogue and digital section as DAC chips typically have both. Anyhow most modern DAC chips require multiple power supplies (digital, analog, Vref, ...). Usually it is beneficial to use separate regulator for each.
If I understood you correctly, you said I should isolate the MCU from the DAC, in that case, PWR and GND would be isolated too, resulting in a separate board.
That's what I thought but wasn't sure.
Regarding the sound card, I got a 2x2 M-audio atm. but windows 10 refused to let me install the drivers I want... 😡 buying a new sound card for building a sound card ...the irony...
Last edited:
Agree with Mr. bohrok about soundcard, I2C register accessibility, and digital isolation (provided added phase noise does not become an issue).
Regarding test equipment, if you are mixing DSP and Class-D amps along with a dac, you might want to look at RF spectra. Cheap and somewhat useful could be:
https://nanovna.com/
https://www.amazon.com/Seesii-Analy...t=&hvlocphy=9032419&hvtargid=pla-869965344213
https://www.amazon.com/Near-Field-A...QJNZZ5G/ref=pd_lpo_2?pd_rd_i=B08QJNZZ5G&psc=1
Regarding test equipment, if you are mixing DSP and Class-D amps along with a dac, you might want to look at RF spectra. Cheap and somewhat useful could be:
https://nanovna.com/
https://www.amazon.com/Seesii-Analy...t=&hvlocphy=9032419&hvtargid=pla-869965344213
https://www.amazon.com/Near-Field-A...QJNZZ5G/ref=pd_lpo_2?pd_rd_i=B08QJNZZ5G&psc=1
Not necessarily a separate board. You can have isolated GND and PWR planes for the MCU (or DAC) on the same board.
Can be but doesn't have to be. A single board can be partitioned.
EDIT: Oops. Cross posted...
From what I have seen and read so far, splitting GND on one board is never advised nowadays. Is this really a good idea?
Splitting GND and having digitally isolated GND and PWR are 2 separate issues.
Splitting GND refers to having cuts in GND plane when both sides of the cut are actually part of the same GND plane. Isolated GND planes means that there is no electrical connection between the GND planes.
BTW even splitted GND planes have their uses e.g. in switching regulators. But the layout can be quite tricky.
Splitting GND refers to having cuts in GND plane when both sides of the cut are actually part of the same GND plane. Isolated GND planes means that there is no electrical connection between the GND planes.
BTW even splitted GND planes have their uses e.g. in switching regulators. But the layout can be quite tricky.
Last edited: