Voltage regulator, LDO + PSRR (10+ kHz), does they exist?

[MasterT]

For a project I'm working on, I need a regulator +7.5V that satisfy two simple requirements LDO & PSRR. 
I see, there are lm317 and all 78xx series,  that have good PSRR above 10 kHz (-70 dB or so up to 20+ kHz) but not LDO.
On the other side modern mosfet p-channel LDO's that have miserable low PSRR, about -10 dB, some companies even don't publish any data except 120 Hz. And its strange for me, unlike 40 years ago nobody should care much about 120 Hz rejection since almost all power supplies switchable, and PSRR in the sub-MHz range should be primary goal for good properly designed voltage regulator. And I don't envision  any obstacle to build fast circuits , since technology step far away from lm317 manufacturing process.
What, inductors makers rule the market?

[Boscoe]

http://www.ti.com/lit/ds/symlink/lp38798.pdf

I found this using TI's parametric search in a minute or so, finding parts is dead easy when you master these things. I use a lot of TI parts as their website and search is particularly good.

[ogden]

Champion of LDO's (today): LT3042.

RMS Noise: 0.8µVRMS (10Hz to 100kHz)
PSRR: 79dB at 1MHz

[ConKbot]

Make sure you check the datasheet curves for PSRR as you approach dropout, and at your intended running voltages. A regulator may specify a 60+ db PSRR, but that may be at Vin-Vout = 1V, not when it's at 0.2V or lower that the dropout voltage is specified at.

[MasterT]

http://www.ti.com/lit/ds/symlink/lp38798.pdf

I found this using TI's parametric search in a minute or so, finding parts is dead easy when you master these things. I use a lot of TI parts as their website and search is particularly good.

Thanks for the tip, TI search engine is much better than at mouser. Though seeding through hobby friendly package groups, I was able to locate non TO-92, and just 1 part in soic-8 adjustable that goes above 5.0V, with good PSRR data. It's LM2931.  There are two more parts LP2951 & LP4951 that have unreliable PSRR over mA current range.  Will order 2931, to see how it goes.

[ogden]

Look at the stupid price.

Sure. It's the best LDO one can get (today).

LP5907 can do 99% the thing LT3042 can do if you don't need high voltage output, and it's a fraction of a buck.

LOL. With same success one can say that LM1117 can do 99% the thing LP5907 can do.

LT3042 PSRR: 79dB at 1MHz
LP5907 PSRR: ~39dB or so at 1MHz

Difference is ~40dB which is 100 times worse

[ogden]

Why do you need PSRR at 1MHz? Anything above 100kHz should be filtered by an LC stage.

Goalpost shifting into LC filtering? OK. Let's assume that main subject of conversation, PSRR is not hat important anymore. Let's compare RMS noise of both.

LT3042 : 0.8µVRMS (10Hz to 100kHz)
LP5907 : 6.5µVRMS (10Hz to 100kHz)

Still LT3042 is better by HUGE margin, thus price is fair.

[ogden]

LT3042 : 0.8µVRMS (10Hz to 100kHz)
LP5907 : 6.5µVRMS (10Hz to 100kHz)

A DSBGA packaged OPA2376 and two NPN pass transistors. The other channel is for free.

The cost is considerably higher quiescent current, and at least 1V of voltage drop, but the price can be still sub $1 and smaller than your LT chip.

The cost is worse PSRR performance. All this argument finding nonsense just to disagree that LT3042 is best LDO around and it's price is fair? 

[ogden]

The cost is worse PSRR performance. All this argument finding nonsense just to disagree that LT3042 is best LDO around and it's price is fair? 

I do not like a magic black box chip that's outrageously priced.

Right. Instead you build LDO out of opamp, pair of transistors and no voltage reference

FYI, I have sub-uV LDOs implemented with an OPAMP built-in an MCU.

Sure. You build sub-uV RMS moise LDOs just like that. I get your point and we are way too off-topic now.

[tszaboo]

Place an RC filter in front of the LDO, and magically now you have good PSRR for high frequencies. 1 Ohm 100uF gives you about -20dB @ 10KHz. With minimal loss in voltage.

[MasterT]

Place an RC filter in front of the LDO, and magically now you have good PSRR for high frequencies. 1 Ohm 100uF gives you about -20dB @ 10KHz. With minimal loss in voltage.

Could be an option, for 20 mA I can push resistor up to 10 Ohm and get -40dB. Still too low,   18-bits ADC 1.33 msps (EVAL-AD7984-Pxxx) is very sensitive.

My first try to get clean -2.5V / +7.5V   power using dc-dc isolated +5/ +-12V than downregulate with lm317l /lm337l turns out a disaster on EMI interference side.
Second build using TC7660 charge pump converter get 6-10 times lower EMI, but I get into the limits of lm317l, loaded converter (with 2x schottky multiplier) supply only 8.8V, and regulator switched itself into bypass - mode.

I put -70db as a base line, and was wondering why there are no lm78xx / lm317 generation-2, or G-3 etc. Some TO-92 easy to use upgrades of the common and very popular ICs.  Though, at least I found one (lm2931) with the help of this forum.