Hello,
I am using LP2989AIM-5.0 for powering up a REF5050, 2x OPA2735, one INA330 and one XTR111.
All of these IC's require decoupling caps, and since this design is intended to take place in a noisy environment and specs. require tight analog accuracy, we are not saving on caps.
Some IC's, like REF5050 have as much as 4 parallel decoupling caps: 4.7u (tantalum), 1u(tantalum), 100n (cercamic COG) and 10n (cercamic COG).
XTR111 is bypassed with 10u (tantalum), 1u(tantalum), 100n (cercamic COG) and 10n (cercamic COG).
These IC's are quite close to the LP2989 output on the PCB, I assume I can consider them as part of the output caps (considering worst case, very short trace length).
LP2989 datasheet is very clear about the minimum 0.004 ohm ESR allowed for stability. Actually, the main reason I am using LP2989 is because it is stable with the lowest ESR amongst all the others I have looked on. I didn't read all TI LDO's datasheets, but a lot of them, and I have to say, it took some time...
My main concern is that 10n and 100n caps are needed for decoupling every IC in the pcb, and I can't calculate neither measure equivalent ESR.
From some threads here in the forum I've seen suggestions to add a footprint for a series inductor, and use a 0H if not needed. That really seems not elegant or practical.
Other threads, like:
http://e2e.ti.com/support/power_management/linear_regulators/f/321/t/243094.aspx
suggest a series resistance connected to the ceramic caps, but I am not sure how to choose a value for the resistor.
Also, is the resistor supposed to be in series with the ceramic caps (each) only, or in series with all the caps in parallel?
There are other LDO's I am using, that power up other IC's that need ceramic low capacitance values decoupling caps as well.
What is your suggestion? R series? If yes, how to choose it? What possible side effects would it have?
Thanks!