Nice quiet regulators for ADC/DAC?

[ElectricDruid]

Hi All,

I'm looking for some advice. When building digital audio circuits, I've always just grabbed a basic 78L05 or 78L33 regulator to get the low voltage power supply. They're what I have about, so...

But I've been reading about it, and it sounds like these ancient regulators are far from a good option for modern digital stuff, and that I should be using something much lower noise if I hope to get decent performance out of my ADC/DACs.

Has anyone got any good tips or parts they can recommend?

Thanks,
Tom

[taudio]

Hi All,

I'm looking for some advice. When building digital audio circuits, I've always just grabbed a basic 78L05 or 78L33 regulator to get the low voltage power supply. They're what I have about, so...

But I've been reading about it, and it sounds like these ancient regulators are far from a good option for modern digital stuff, and that I should be using something much lower noise if I hope to get decent performance out of my ADC/DACs.

Has anyone got any good tips or parts they can recommend?

Thanks,
Tom

Hi Tom,

I've used the AP2202 in some designs (like the Easy Spin pedal) and found that it works well.

The AP2202 only goes to 3.3V output max but there is an AP2205 that will deliver 5V output. I've not used the AP2205 (yet).

Both are SMD parts.

Cheers,
T.

[niektb]

Problem with a lot of LDOs is that the PSRR quickly drops over frequency, the AP2202 is no different in that regard, see this plot from the datasheet:



You could even argue that a well-designed buck converter has less noise because their PSRR still has decent values well into the MHz range
As far as a good low-noise LDO I would recommend the LT3045, it's not cheap but definitely high performance.

[PRR]

There is no problem filtering wall AC to DC so clean you can leak it into your audio.

Capacitors are your friends.

[Rob Strand]

Problem with a lot of LDOs is that the PSRR quickly drops over frequency, the AP2202 is no different in that regard, see this plot from the datasheet

Yes there's no point having low noise if the PSRR is letting more noise pass through.   PSRR can be horrible on some devices.

PSRR can also vary with the current.  In fact you can add a dummy load on some device to place it in the optimal operating range (there's at least one app note about it).

When comparing regulator performance the conditions might not be equal.  Different loads, different output caps.

There's a whole heap of "ultra low noise" LDO's and "low noise" LDO's out there.   Pretty much all manufactures have them.  Many are "modern" packages like mini-BGAs.
[eg.  https://www.analog.com/en/products/adm7150.html]

The thing I find about regulators:  At the start you see an overwhelming number of regulators.  However, when you impose input voltages ranges and currents it narrow things down (especially the case with modern devices).  Then you impose packages, availability, and MOQ's the list reduces.   Then you impose drop-out, quiescent current, noise, even ease of stabilization (many LDO's are fussy on the output load) and the list gets smaller.    After doing all that you have to be your own devils advocate and ask have I gained anything over my old LM78xx or LM317 (or LP2992)..

[ElectricDruid]

Yes there's no point having low noise if the PSRR is letting more noise pass through.  PSRR can be horrible on some devices

Are we talking about the PSRR of the *regulator* here? What's the point of a *regulator* if it's PSRR is terrible? Isn't that what they're for?!? There must be something I'm missing...

[Rob Strand]

Are we talking about the PSRR of the *regulator* here? What's the point of a *regulator* if it's PSRR is terrible? Isn't that what they're for?!? There must be something I'm missing...

Yes PSRR of the regulator.    And there's some regulators with quite bad PSRR out there, either the outright PSRR or the poor PSRR bandwidth - more often than not low power types.

If you think about a system that needs 5V and 3.3V rails.    If you use a 3.3V regulator with a poor PSRR and  it's fed from the regulated 5V rail you can get away with a lot and you might make that choice for the regulator if the quiescent power was low (as required by some hypothetical circuit).   That's where these regulator might find a home.

The reason I mentioned it if you aren't aware of the problem you can get stung.   When things start misbehaving most people don't think the PSRR regulator could be causing the problem.

[ElectricDruid]

Ok, I think I get it. So you're saying that these regulators might reduce the voltage, but they're fairly transparent to noise, essentially? That any muck on the rails is going to finish up on the reduced-voltage output rail too? Have I got that right?

Thanks,
T

[Rob Strand]

Ok, I think I get it. So you're saying that these regulators might reduce the voltage, but they're fairly transparent to noise, essentially? That any muck on the rails is going to finish up on the reduced-voltage output rail too? Have I got that right?

That's the idea.   I wouldn't go so far to say they are transparent but they are certainly a lot poorer than what we are used to seeing.  (I can't remember any specific part numbers.)

[Ksander]

I've tried some high frequency adjustable switching regulators recently, like 'XL4005 LM2596 DC-DC', stuff from Ali, to power the heater filaments in a Valvecaster. They switch at 180kHz, maybe it was even higher, and I've found them to be perfect for the job. No audible noise whatsoever.

[ElectricDruid]

I've tried some high frequency adjustable switching regulators recently, like 'XL4005 LM2596 DC-DC', stuff from Ali, to power the heater filaments in a Valvecaster. They switch at 180kHz, maybe it was even higher, and I've found them to be perfect for the job. No audible noise whatsoever.

Those sound useful for some other jobs, but Valvecaster heaters is a very different application from ADC/DAC, and I'm very nervous about having switched power for a codec.

[peterc]

A mate of mine always used the Linear Tech regs
https://www.analog.com/media/en/technical-documentation/data-sheets/108345fh.pdf
I could never afford them....

[JTEX]

Just use an ADM7150 for the clean +5V rail. It's what AKM uses in their eval board for AK5397. You can't go wrong.

[ElectricDruid]

Just use an ADM7150 for the clean +5V rail. It's what AKM uses in their eval board for AK5397. You can't go wrong.

Sounds good. Do they do a 3.3V version too?

[PRR]

3.3V version too?

https://www.analog.com/en/products/adm7150.html
Fixed output voltage options: 1.8 V, 2.8 V, 3.0 V, 3.3 V and 5.0 V

but also:
Alternative Parts    Product Life Cycle    Description
ADM7151 RECOMMENDED FOR NEW DESIGNS    800 mA Ultralow Noise, High PSRR, RF Linear Regulator