Let's consider two different power supplies for a DC-10 KHz 4 channel single ended 18-bit DAC being used for a physics experiment. It will in the end feed a high impedance — the principal power draw is the quiescent load of the opamps, DACs, and MCU. For reference, 18-bits corresponds to 108.37 dB of SNR:
In one design I use an external power supply such as Blocked . Inside I use a high quality LDO. 1% pk-pk ripple from the datasheet is -20 dBV voltage, or -40 dB power relative to the supply, and then on top of that I can use a high end LDO such as the TI TPS7A4901 which features 70 dB PSRR, for a total of 110 dB... ideally. I place the brick outside the enclosure and add a ferrite to the DC line to fight common mode noise. Maybe I add some ferrites and pi filters inside the enclosure before the LDO to get the last bit of performance out. Given that the references of the DAC and the DAC itself have their own PSRR, I would imagine I'm in the clear.
In another design, I use a much larger chassis and an Acopian A5MT510 which outputs 5 volts and has 0.25 mV ripple. That's -40 dBV or -80 (?!) dB power relative to the supply. I suppose I could add an LDO there and potentially under ideal circumstances beat 110 dB but 110 dB seems quite low, and I have to imagine I'd be limited by other things at this level.
My question is this... is there really any sense in buying a high end linear power supply outside of simplicity and reliability when you can get a decent switcher and pair it with an LDO and a little filtering and do essentially well enough? If this is true now, when in the history of electrical engineering did it become true and what was the reason?
Finally, is there a cheap and cheerful way to probe a power rail? I’m thinking I want a 50 ohm transmission line with DC blocking capacitor and front termination feeding into a low noise preamp like an SR560 or something like that. Maybe I should instead use a FET probe, but in the end I really have no qualms about loading the rail, as it’s a power rail!
In one design I use an external power supply such as Blocked . Inside I use a high quality LDO. 1% pk-pk ripple from the datasheet is -20 dBV voltage, or -40 dB power relative to the supply, and then on top of that I can use a high end LDO such as the TI TPS7A4901 which features 70 dB PSRR, for a total of 110 dB... ideally. I place the brick outside the enclosure and add a ferrite to the DC line to fight common mode noise. Maybe I add some ferrites and pi filters inside the enclosure before the LDO to get the last bit of performance out. Given that the references of the DAC and the DAC itself have their own PSRR, I would imagine I'm in the clear.
In another design, I use a much larger chassis and an Acopian A5MT510 which outputs 5 volts and has 0.25 mV ripple. That's -40 dBV or -80 (?!) dB power relative to the supply. I suppose I could add an LDO there and potentially under ideal circumstances beat 110 dB but 110 dB seems quite low, and I have to imagine I'd be limited by other things at this level.
My question is this... is there really any sense in buying a high end linear power supply outside of simplicity and reliability when you can get a decent switcher and pair it with an LDO and a little filtering and do essentially well enough? If this is true now, when in the history of electrical engineering did it become true and what was the reason?
Finally, is there a cheap and cheerful way to probe a power rail? I’m thinking I want a 50 ohm transmission line with DC blocking capacitor and front termination feeding into a low noise preamp like an SR560 or something like that. Maybe I should instead use a FET probe, but in the end I really have no qualms about loading the rail, as it’s a power rail!
Have you set up your DAC with just a simply power supply?
What instrumentation do you have to check the DAC signal output, and what would you expect to check and see? Have you done that measurement with any instrumentation?
Why do you need to focus on the power supply performance - have you already measured a DAC output and found it wasn't what you expected?
Are you really wanting to set up a PSRR type measurement of your DAC output, and need to apply a test frequency signal to your power supply output, and measure the level of test frequency signal that has coupled through to your DAC output?
What instrumentation do you have to check the DAC signal output, and what would you expect to check and see? Have you done that measurement with any instrumentation?
Why do you need to focus on the power supply performance - have you already measured a DAC output and found it wasn't what you expected?
Are you really wanting to set up a PSRR type measurement of your DAC output, and need to apply a test frequency signal to your power supply output, and measure the level of test frequency signal that has coupled through to your DAC output?
If you sync your SMPS to the DAC there will be nada pinjada of any sort.
I am not sure this simplified method of thinking works but...
What is the reference voltage? 2.048V? 18bit and each step is 7.8uV. How noisy is the reference?
If the DACs PSRR is for example 40dB, 0.78mV can possibly create a bit change.
Can you even measure a single bit change?
I am not sure this simplified method of thinking works but...
What is the reference voltage? 2.048V? 18bit and each step is 7.8uV. How noisy is the reference?
If the DACs PSRR is for example 40dB, 0.78mV can possibly create a bit change.
Can you even measure a single bit change?
I should say that this isn’t an audio thing, and the DAC is updated asynchronously relative to any clock. The 10 kHz is just a figure for reference. It needs to update and then stay quiet until the next update, meaning the SMPS can’t be clocked synchronously relative to anything else.
ezalys, does that mean that you have no further comments on the queries?
PS, I see you are cross-posting over on eevblog - does that mean you are going to cross-post relevant comments here too?
PS, I see you are cross-posting over on eevblog - does that mean you are going to cross-post relevant comments here too?
Last edited:
You seem to be asking questions about my ability to measure and about factors that might dwarf the power supply noise. You’re treating this like an X Y problem when what I want are answers relevant to the physics.
I’m sure anyone else googling the relevant questions can see my post on the other forum. I’ve never once seen someone cross-post answers and I think you’re being needlessly deferential and have said nothing relevant to the underlying physics.
I’m sure anyone else googling the relevant questions can see my post on the other forum. I’ve never once seen someone cross-post answers and I think you’re being needlessly deferential and have said nothing relevant to the underlying physics.
The designer of the "DC-10 KHz 4 channel single ended 18-bit DAC" should be able to say if the DAC needs super-slick power or if s/he considered that and put sufficient PSRR in the DAC.
There may be a broader physics-based answer but this appears to be a practical project and not a philosophy question.
There may be a broader physics-based answer but this appears to be a practical project and not a philosophy question.
smps bricks are very cheap nowadays, for me enough reason to use them...low cost is a great motivator for me...
Last edited:
Ezalys, your question was:
Imho, the physics experiment related aspect is does a particular power supply influence the capability of the DAC. Have you measured the DAC output under controlled conditions and identified an abnormal response? Similarly, have you measured the DAC output under the same controlled conditions, but with the two identified power supplies, and see a different response?
Imho if you haven't taken the time to make real world measurements, and identified a concern related to your physics experiment with one or both power supplies, then you are just fishing for hypothetical situations that may have no relationship to your physics experiment.
As you indicate, the DAC PSRR may exclude any power supply from influencing your physics experiment results. I suggest you explore that path further for starters and identify the external power supply perturbations that could influence the physics experimental results. Perturbations could be for example transient or periodic, and could be just related to the power port, or to stray system ingress, and your experimental assessment of DAC output could attenuate certain types of influence so that they are negligible.
... any difference between two types of power supply. [my summary of your question]
Imho, the physics experiment related aspect is does a particular power supply influence the capability of the DAC. Have you measured the DAC output under controlled conditions and identified an abnormal response? Similarly, have you measured the DAC output under the same controlled conditions, but with the two identified power supplies, and see a different response?
Imho if you haven't taken the time to make real world measurements, and identified a concern related to your physics experiment with one or both power supplies, then you are just fishing for hypothetical situations that may have no relationship to your physics experiment.
As you indicate, the DAC PSRR may exclude any power supply from influencing your physics experiment results. I suggest you explore that path further for starters and identify the external power supply perturbations that could influence the physics experimental results. Perturbations could be for example transient or periodic, and could be just related to the power port, or to stray system ingress, and your experimental assessment of DAC output could attenuate certain types of influence so that they are negligible.
Last edited: