Power Supply Ripple

Hey Santosh,

What frequency (or frequencies) is the ripple occurring at?

~ Aaron

Hi Aaron,

Here I have attached the images captured for peak to peak voltage and frequencies also.

Santosh,

If the ripple is propagating through the line, it is likely that it is not being attenuated enough by the LM78 or LM79 family of linear regulators. As you can see in this PSRR curve for LM78, PSRR starts dropping off significantly at the 10kHz mark by about ~20dB per decade. At 1.5MHz, the PSRR will likely be at or close to 0dB. As a result, the ripple at this frequency is unaccounted for. As for the 133kHz ripple, you're getting about ~40dB worth of attentuation through the regulator, which may or may not be enough.

PSRR is defined as: 

I'd advise trying an LDO with a wider bandwidth. Take a look at TPS7A4901. Here's the PSRR curve:

Of course, there's always the possibility that the linear regulator output is picking up the ripple via the air. 

Let me know what you think.

~ Aaron

Hi Aaron,

I'm already using TPS7A4901 LDO in some other circuit which is of 150mA. Here I want high current rating linear regulators of around 1A to 2A. Can I use some LC low pass filter stages before LM78/79 series regulators to reduce these ripple to some extent?

Santosh,

Sure, you could. Here's an article going over the considerations for such a filter.

However, I believe that you could use another LDO like TPS7A4701 to eliminate the ripple.

What's the input voltage to the linear regulator?

~ Aaron

Hi Aaron,

I will check whether I can use TPS7A4701 for my application. The input voltages to linear regulators are +12v.

Santosh,

I think that the attenuation will be much better than that associated with the LM78 device.

Please don't hesitate to contact us if the problem persists.

~ Aaron

Hi Aaron,

Every time when  I combine analog & digital grounds at one point, I found the noise from digital ground spreads to the analog ground. They are good when isolated, but bad when combined. I usually use ferrite beads between them, but still the result is not that good. So my query is how can I well combine these two grounds?

Thanking you for your previous replies,

Santosh

Santosh,

Which ferrite bead are you using? Also, which frequency is the noise occurring at? (Is it the same as before?)

Here is a good article on the issue you are experiencing.

~ Aaron

Hi Aaron,

The ferrite bead I am using is 120 ohm @ 100 MHz with DCR = 30 mOhm. The frequency of occurance is same as before. I will see the article too.

Thank you,

Santosh

Santosh,

It sounds like the impedance at the frequency of interest (133kHz and 1.5MHz) might not be high enough. You probably need to size up the inductor. The other option is to create a virtual ground by using a resistor divider between the outputs of a positive and negative LDO (ex. TPS7A47 & TPS7A33). Since the outputs are regulated, you won't encounter the bouncing ground problem.

Let me know what you think.

~ Aaron

Hi Aaron,

I designed new power supply board using TPS7A47 and TPS7A33 but still the noise is not reduced. However it is better than LM7XXX series. The noise immunity is just 5% to 10% better. Do I need to use any filters? Then how to select the values of the components?

Santosh

Santosh,

Have you determined that the noise is propagating through the supply or is it possible that the output of the LDO is picking up the emissions from the air? LC filters might need to be added in either case. Also, if the noise is being radiated from your switcher through the air, it might be beneficial to shield the inductor.

~ Aaron