For many types of analog and RF circuits, low-noise voltage supplies are important for achieving the proper performance. A high-noise floor on RF, audio, or other ADC/DAC and analog circuits can doom a project. Whether it's a precision analog data-acquisition system (like reading transducers to 10 bits and above), HDTV, a radio, or a music system, good voltage regulation is critical. To that end TI has come out with a new series of low-noise, low-drop-out (LDO) voltage regulators.
Consider that you typically use an LDO in circuit locations where there is already electrical noise on the power bus, so a device with a good power supply rejection ratio (PSRR) is important. With the PSRR, it's important to see at what frequencies the measurements were made. Older technology LDOs would show a good PSRR at 120Hz, but not one so good at the higher frequencies. Also of importance is the inherent noise of the device.
From the LP38798 landing page on the TI site and from the data sheet, we get some interesting pieces of info. The device has very low output noise -- 5 μV in the 10 Hz to 100 kHz band. It has high a PSRR -- 90 dB at 10 kHz. Even at 100 kHz the PSRR is still at 60 dB. Those specs meet the important criteria mentioned above.
Also worth noting: TI has fabricated the device on a medium-voltage process that can operate up to 20 V. That means this part is good for systems with an intermediate bus voltage of 12 V. There are other parts that are probably a little lower cost than this part, but they are fabricated on a low-voltage process. They can't operate much above 6 V. Regarding cost, it sells for $1.30 in lots of 1,000 pieces. It's not dirt cheap, but not outrageously expensive either.
Stability is important with any regulator. This one will work properly with tantalum or ceramic output caps, so that gives you more flexibility regarding component cost, size, and availability.
If you need a little better PSRR at high frequencies, you can add a bit more external filtering. That's not a perfect solution of course, but it's not too bad (two more caps and a small inductor).
You would probably use an LDO like this after a switch-mode power supply (SMPS) to provide clean DC for VCOs, PLLs, and ADCs. These are also good where there can be transients on the SMPS output due to other loads stepping/changing with time. They are also able to solve other issues, such as susceptibility to power line-conducted noise (noise rejection issues).
Note that if you have a 16-bit data-acquisition system, you need 96 dB of noise-free range to get a full 16 bits. With today’s lower voltage rails, this noise level can be trouble, even with (for example) an op-amp with a 60 dB PSRR spec. Many other analog and RF applications will benefit from an LDO like this.
What uses do you see for an LDO like this? Can an LDO like this replace older multi-part discrete filters and give an equivalent or better performance in your application?