LDO regulators in very low noise rf designs

I Have noticed a problem regarding the use of LDO regulators in very low noise rf designs. It seems that a LDO regulator from say ROHM or NJRC under normal conditions, (decoupling caps as manufacturers recomendation)give a very low ripple (or possibly a damped oscillation) when placed under load, say 100 - 300 mA. The ripple is around 150mV for a load of 200mA. If i increase the caps (not good since board area is tight) the ripple reduces a little. I am using X7R ceramics so the ESR is very low.
If i replace the reg with a standard linear type, the ripple goes completely.
I know this must be the way in which the regulator has been designed, but does any one know why this happens?

The ripple is 50Khz for the ROHM ldo and 100Khz for the NJRC type.

RGS

Bagster

The noise you are seing is indeed most likely a feedback oscillation of the LDO's internal op-amp circuitry.

Have you tried using tantalums on both input and output? Ceramics are great, but sometimes you need a little beef in your supply, especially if you have any significant transients in the load.

I ran into a similar problem with a Linear Tech LDO a few years back and the only thing that helped was adding 100uF decoupling to the *input* terminal of the regulator.

It is difficult to use tants as this application is very very limited in space, to the extent where an 0603 is a problem to fit in. I think the solution is to improve the headroom voltage and go for a standard reg or even a shunt reg!

Thanks anyway

Bagster

You may be able to just increase the voltage and make the problem "go away" with the LDO (I didn't know that was an option for you).

If you are using the LDO near the low drop-out (typically 1 - 2 volts) then it will increase output oscillation if you don't have adequate decoupling at both the input and output terminals.