TPS63001 to power arduino

Hi,
I have been looking at using a TI TPS63001 to get the last bit of juice out of my batteries and I was wondering because it doesnt produce a completely flat output do I need a linear regulator afterwards?

No, you don't need a linear regulator afterwards. The ripple produced by that regulator is 10mV or less according to the datasheet. If you are using the ADC then you might want to feed the Aref pin via an L-C filter, otherwise the ripply may cause a small amount of fluctuation in the ADC reading.

Oh ok thank you, I am unsure of how to calculate what I need. Any advice would be appreciated. I am running everything at 3.3V

If you are going to filter the supply, then I suggest feeding the regulator output through a choke of between 1uH and 10uH followed by a second 10uF tantalum capacitor to ground. The choke should have a DC resistance of around 0.5 to 2 ohms ohm to provide sufficient damping without undue voltage drop.

Have you measured the actual output or are you basing the ripple on just what the datasheet says?

I'm going from the datasheet - that's what it's for.

dc42:
I'm going from the datasheet - that's what it's for.

The question was not directed at you, it was directed at the original poster. The reason for the question will help determine the correct answer.

Hi, Yes I was going from the datasheet and was wondering whether the ripple would affect the operation of the MCU.

If you were looking for +/- 1 bit accuracy from an ADC powered by this switcher without any filtering, IMO you 'might' have an issue.
@ 10 mV noise the split is +/- 5 mV. with a little filtering provided by the board and some small oversampling, I don't think there is any issue at all.
even with the 1V1 ref on an Uno or 2V56 aref configured on a mega. a 100 uHy choke and a 47uF cap as an input filter and a 22uF cap as output filter you should be fine
at 50KHz a 100 uHy choke has about 3k ohms of reactance at 150 KHz it is 3 X that with a dc resistance of under an ohm a switcher inductor of that value is perfect
in that the magnetics are shielded, there's no radiation from the choke and no need to shield anything... IMO and in the case of several prototypes I built with small
Linear tech LT1172 switchers never an issue.

Doc

{I neglected to mention that I removed the filtering (3 1206 parts) in production as they weren't required at all... Edit}

mrjonny2:
Hi, Yes I was going from the datasheet and was wondering whether the ripple would affect the operation of the MCU.

Yes, that's why I asked the question I did.

This principal applies to hardwarew and software: don't optimize unless you have to.

Most linear regulators like the TPS series have such low ripple, you don't need to over complicate the decoupling solution. I think you'll find other aspects of your design are more critical than the 10mV of ripple from the regulator.'

Explain more about how the rest of your circuit works and maybe the suggestions so far might make more sense.

Well as part of the rest of my circuit I have ethernet and SD on SPI, I have 2 devices on I2c and I am then bit banging 7 more devices on virtual I2C channels