Linear regulator ripple

[graeme1]

We have a +/-12V regulated power supply as part of our board design.

The board takes a 5V input and passes this though a TI DCP020515DU DCDC. This gives us +/-15V unregulated. Each 15V rail has been tested with loads up to 60mA (our requirement is 30mA).

The +ve output from the DCDC is regulated to +12V using an LM3480IM3X. This is nice and stable.

The -ve output is regulated to -12V using an L79L12A.

Even with no load there is noticeable -12Vdc ripple. As the load increase, then the ripple becomes large and the voltage goes beyond -12V to -13V.

Your L79L datasheet states that no external components are required. The test circuit (figure 3) has input and output caps. We added the caps to our circuit.

Can you suggest anything that could be the cause of this behaviour?

We have temporarily replaced the ST regulator with another manufacturer which has a completely stable output.

[AScha.3]

I would go the easy way: just use a DC/DC converter with regulated outputs.

Cost is a little bit higher, but need no further regulators. So should be the better way.

[graeme1]

Thanks. Yep, that would certainly be a solution.

Or, I could find out why the simple linear regulator just doesn't work as expected.

I did ask ST support directly, but I didn't get any. For now, I'm using a reg from ON Semi which is stable.

[MasterT]

Be advised, that L78xx & L79xx regulators are not an LDO type, so minimum drop voltage required. I see L79L mention 1.7V in data sheet. Usualy they listed in data sheet (all other makers as well, TI OnSemi etc.) PSRR with significant input-output voltage difference, for example 

https://www.st.com/resource/en/datasheet/l79l.pdf 

VI = -15 to -25 V, f = 120 Hz IO = 40 mA, TJ = 25 °C

as you can see, -15 is the minimum for PSRR specification, and I'm sure your converter likely fail this under load. 

Also pay attention what freq. PSSR is measured, 120 Hz is likely not what converter works (100kHz ???)

and the issue with all regulators (except some new LDO) that PSRR at high freq. range be likely much lower, 10-20 dB depends on output capacitor.

And last, L79xx series was developed when MLCC capacitors were not available, and regulator likely be unstable if  ESR value very low, < 1Ohm range. Try electrolytic or insert R in series with output MLCC cap.

Op-amp "ultra low noise regulator" I designed for AD adc module:

[graeme1]

Many thanks for your time and the detailed answer.

The L79L datasheet does suggest that the output voltage should be between -11.4 and -12.6 with an input of -14.5 or higher. Our DCDC (which switches at 800kHz) has a steady 15V when measured under full load (without the L79L connected).

We can't really avoid low ESR caps on the output, as most of the other devices that it powers require these.

It's funny that the onsemi MC79L behaves much better. I would have thought they are pretty much the same inside.

[MasterT]

The fact is, even mc79l performs well, doesn't change anything. Both parts were not designed for high freq. switching application. And things may be quite different in real enviroment, their parameters are not predictable and parts should not be in use.

I see TI has some high voltage/ high PSRR at 1 MHz regulator:

https://www.ti.com/product/TPS7A30

[graeme1]

Thanks again.

That TI part won't do -12V, but I see others (Micrel) that say suitable for all types of capacitor. MIC5270/71. I will look in more detail at those.