Hi,
I am trying to do my best to maximise the sound quality from my 3e TPA3255 amp build. I have a handle on the basic things like using shielded hook up wire and routing it appropriatelyand having single earth connection to the chassis, but I am wondering whether I should be bothering to use some method of cleaning up the power supply from the SMPS which i'll be using to power my DAC. I have briefly read about options such as LC filters and also ferrite beads. The filters sound complicated and apparently can result in a voltage drop, and the ferrite beads seem to be more effective in controlling noise in the frequency ranges which are much higher than audio range. I am also using a separate SMPS to power my 3255 amp board. Both SMPS switch at about 70kHz.
I had considered using a separate regulated linear power supply just for the DAC but figured there wasn't alot of point given that I am using SMPS for the amp and also I will be using the second 5V SMPS to power a raspberry pi, so I figured that using this to power the DAC would be a simple improvement over using the USB from the RPI to power the DAC.
Generally, the consensus seems to be that SMPS don't affect the sound quality of Class D amps (?much) provided they switch at a high enough frequency. If I am not filtering the power supply to the amp, is there any logic in filtering the supply to the DAC? If so, what is the simplest method.
DAC is a Khadas Tone Board.
SMPS feeding it is this one:
ECM60UT31 | XP Power, 60W AC-DC Converter, 5 V dc, +-12 V dc, Open Frame, Medical Approved | RS Components
Thanks,
Mark
I am trying to do my best to maximise the sound quality from my 3e TPA3255 amp build. I have a handle on the basic things like using shielded hook up wire and routing it appropriatelyand having single earth connection to the chassis, but I am wondering whether I should be bothering to use some method of cleaning up the power supply from the SMPS which i'll be using to power my DAC. I have briefly read about options such as LC filters and also ferrite beads. The filters sound complicated and apparently can result in a voltage drop, and the ferrite beads seem to be more effective in controlling noise in the frequency ranges which are much higher than audio range. I am also using a separate SMPS to power my 3255 amp board. Both SMPS switch at about 70kHz.
I had considered using a separate regulated linear power supply just for the DAC but figured there wasn't alot of point given that I am using SMPS for the amp and also I will be using the second 5V SMPS to power a raspberry pi, so I figured that using this to power the DAC would be a simple improvement over using the USB from the RPI to power the DAC.
Generally, the consensus seems to be that SMPS don't affect the sound quality of Class D amps (?much) provided they switch at a high enough frequency. If I am not filtering the power supply to the amp, is there any logic in filtering the supply to the DAC? If so, what is the simplest method.
DAC is a Khadas Tone Board.
SMPS feeding it is this one:
ECM60UT31 | XP Power, 60W AC-DC Converter, 5 V dc, +-12 V dc, Open Frame, Medical Approved | RS Components
Thanks,
Mark
Easiest and cheapest solution is SMPS, best solution for a DAC (and possibly RPI too) is a linear PSU.
It is a never ending story "linear versus SMPS" but in my experience linear PSU's almost always win in the relevant aspects certainly when we are talking 15VA or less and used for audio sources. Please note that the best audio-SMPS are equipped with a linear LDO regulator for lower noise....
A DAC in general is sensitive to PSU noise so using SMPS here is asking for extra work getting rid of RF hash, noise etc. A plain and simple linear PSU of normal quality already will be cleaner to begin with. A specific low noise linear PSU will be cleanest. Whether this difference justifies the price difference is up to you.
True audio SBC's mostly are equipped with linear regulators and low jitter clocks too. With such devices the emphasis is on good sound quality. Not unimportant in audio 🙂
It is a never ending story "linear versus SMPS" but in my experience linear PSU's almost always win in the relevant aspects certainly when we are talking 15VA or less and used for audio sources. Please note that the best audio-SMPS are equipped with a linear LDO regulator for lower noise....
A DAC in general is sensitive to PSU noise so using SMPS here is asking for extra work getting rid of RF hash, noise etc. A plain and simple linear PSU of normal quality already will be cleaner to begin with. A specific low noise linear PSU will be cleanest. Whether this difference justifies the price difference is up to you.
True audio SBC's mostly are equipped with linear regulators and low jitter clocks too. With such devices the emphasis is on good sound quality. Not unimportant in audio 🙂
Last edited:
The Khadas Tone Board claims to have LDO linear regulators and high performance crystal oscillator built in.
Does that mean that in theory the DAC should address the shortcomings of a less than perfect power supply and poor quality crystal of the RPI? Or is jitter still dictated by the SBC that it is connected to?
There is no point in me powering the DAC separately (rather than via the RPI USB port) unless it will likely improve the sound quality.
"Ultra-low Phase Noise Clock Jitter: The SiTime ultra performance oscillator SiT820&8209 series with just 0.5ps phase jitter and output frequency stability as low as ± 10PPM. We chose it as a reference clock in our audio system so that the data clock jitter < 30ps."
Attachments
Last edited:
I'd say 'yes' - an LC filter is the best option. Ferrite beads are poorly specified for filtering duty in that their inductance drops with load current in ways which manufacturers don't quantify in general. All filters are going to incur a voltage drop - you specify the inductor for whatever voltage drop is acceptable to you. LC filters are not very complicated but magnetics scare most people it seems.
As for your question why filter the DAC supply when not filtering the amp's, I'd say based on my experience that it can pay dividends to filter the amp's supply. Trouble is amps generate their own supply noise so you want the signal stage supplies separate from the power stage ones, something that's normally not straightforward to implement when using chip-based classD with integrated output stages. I got considerable improvement from my IRS2092-based amps with extra attention to the signal stage power supplies.
Last edited: