1
\$\begingroup\$

I'm trying to understand the effect of ripple rejection in linear regulators since I'm using 2 LDO's in designing a linear power supply, however I'm unsure about something. A lot of datasheets have values that range from 50dB to 90dB, however they're dependent on the ripple frequency, load current, and other variables.

For example the LT3090 http://cds.linear.com/docs/en/datasheet/3090fa.pdf has the following Ripple Rejection characteristics:

So going by this, my understanding is that the Ripple Rejection capability of an LDO decreases with an increasing ripple frequency, and increasing load current?

Does anyone know how much of a significant impact a 50dB ripple rejection would have on the operation of an LDO? Are there 'common' or 'typical' ripple rejection values that are acceptable? My linear power supply should be able to deliver maximum of ~700mA, so I need to consider the ripple. It's the first time I'm dealing with this so I'm unsure.

For the record, the power supply I'm designing will have a 21.6V Li-Ion battery as input, which will then power a dual rail supply, so I can get anywhere between +/-8V - +/-10V. I'm currently going with the LM337 and LM317 regulators, which both have ripple rejection values of 75/77dB, but they're also rated at 120Hz.

Improve this question
\$\endgroup\$
2
  • \$\begingroup\$ 700 mA sounds like a lot for a linear regulator, and apparently out of spec for the one in your link. What is your input voltage, and what are you going to drive with it? \$\endgroup\$ – Dampmaskin Feb 28 '17 at 0:27
  • \$\begingroup\$ It's just the maximum figure I'm working with, I don't expect it to reach 700mA, but it's the limit for now. I've updated my original post, but the input voltage will be from a 21.6V Li-Ion battery, which will drive a dual rail power supply into a PCB with a 32bit MCU, some filters, ADC, bluetooth, and ultimately a coil antenna. \$\endgroup\$ – deki Feb 28 '17 at 0:37
2
\$\begingroup\$

All LDOs, like all opamps have gain that falls with frequency, however in your case I would note that with a Lithium pack feeding the thing, significant ripple is unlikely.

50dB is saying that for an input ripple of 1V you get ~3mV of ripple at the output.

20dB is saying 100mV at the output for 1V at the input, but note that is at 1MHz, this is basically warning you about possible fun when following a switched mode supply with one of these without doing some L/C filtering first...

Only you can decide what ripple is acceptable, it depends hugely on your application, sometimes a hundred mV or even a volt is a 'who cares' thing, sometimes 100uV is too much, it just depends on the application.

Improve this answer
\$\endgroup\$
2
  • \$\begingroup\$ Linear regulators have a parasitic capacitance across the transistor that dissipates heat. Assume 1,000pF between RAW DC and the REGULATED OUTPUT. Even if the internal servo amplifier has excellent bandwidth, that 1,000pF IN-OUT link is troublesome. \$\endgroup\$ – analogsystemsrf Feb 28 '17 at 6:50
  • \$\begingroup\$ The cap doesn't dissipate heat obviously, but ANY linear reg dissipates at least (Vout - Vin) * Iout in the pass transistor. The real gotcha is more of a loop bandwidth thing then directly a parasitic cap thing, at least at any frequency where the LDO has any worthwhile PSRR at all (It would be a HUGE LDO to have a nF input to output!). Like any component you just have to understand the implications of the datasheet and design around them. \$\endgroup\$ – Dan Mills Feb 28 '17 at 10:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.