TL;DR:
PSRR filtering is becoming more and more difficult for engineers to incorporate into timing designs. In this post, we explain why this is, what PSSR is in the first place, and the best ways to avoid the dreaded "supply ripple". You'll have a strong understanding of the problems and solutions that arise from PSRR by the end of this post.
TOP STORY
Noise Gets Rejected by Power Supply
PSRR (Power Supply Rejection Ratio) is a measure of how well a circuit rejects noise of various frequencies from a particular device's power supply. More specifically, from a device's voltage input. Optimally, if the voltage input of a device changes, the output should not change. Unfortunately, this is almost always not the case.
How do you determine PSRR?
The ratio of power input to output is where we get the term PSRR. To determine the basic PSRR of a circuit, simply take the change of input voltage divided by the change of output voltage.
(ΔInput ÷ ΔOutput) x 100 = % PSRR
The ripple can be caused by multiple sources including a switching ripple from a DC/DC converter, a 50Hz/60Hz supply ripple, or a ripple caused by the sharing of an input supply between multiple circuit blocks on the board.
Wait a minute... What's this "ripple" you're talking about?
There are actually multiple terms engineer's use to describe PSRR. "Power Supply Rejection Ratio" is also referred to as "Power Supply Ripple Rejection", but in the world of timing solutions, we typically take the simple route and refer to PSRR as "supply ripple" (or even just "ripple" for us lazy folks). For the remainder of this post, we will use both terms, "PSRR" and "supply ripple", interchangeably.
Now that we cleared up any confusion, let's move you on to the good stuff...
The RF Engineer's Dilemma
Engineers and Designers are often faced with the challenge of increasing the amount of functionality of a printed circuit board (PCB) while also making them smaller and more compact. Hardware designers typically do this by using a "cleaner" power supply; however, the board space required to implement power supply filtering can increase the overall size and cost of the design. Without going into too much unnecessary detail, connecting multiple components to this clean power supply can result in direct jitter problems from just one device to all other devices. This can cause issues to arise for oscillators that didn't have ripple problems to begin with.
...Don't fix what's not broken
To make everything even more challenging, power supply noise results in additional jitter. Many frequency oscillators are tested only in low-noise environments resulting in their sensitive nodes coupling noise. That coupled noise will translate into output jitter that will oscillate at the fundamental oscillation frequency.
To learn more about output jitter, check out this Ultimate Guide to Understanding Phase Noise! It also covers the basics of phase noise... which may help you to grasp this topic even better.
Stop the Design Madness!
As long as you take the easy route, solving this PSRR design problem might be easier than you first thought.
The secret in its simplest form: Filters, filters, and more filters!
Ok...it may be a little more complex than that.
A device's supply ripple can usually be managed by applying a filter across the entire power supply. Applying such a filter will help tremendously with managing the supply ripple. Some recommended filter examples could be:
- A ferrite bead
- A discrete SMD capacitor (Low Pass Filter)
- A linear regulator (used as a high pass filter)
In some cases, both filters acting as a low and high pass filter may be needed over the entire frequency band of a given device. Check out the Top 4 Filter Types to discover which filter typology will work best for your application.
Luckily, some of this burden is now being covered by device manufacturers.
let me explain...
As a device's performance increases, the more likely expensive external circuitry will be needed. Because of this, many device engineers have started incorporating internal circuits in their designs (such as PLL's, filters, and LDO's).
So why is this a good thing?
Because such internal circuits actually aid in controlling the supply ripple of the device... taking some of that burden off the system integrators.
But beware! Every device is different. Designs by various component vendors may not always impact the observed supply ripple. Incorporating an all-encompassing filter circuit may also not be a good decision in your design. This is because the oscillator may not work well in certain conditions, including conditions under an all-encompassing filter. Tests should be done before hand to make sure your system will perform correctly under the given design conditions.
In Summary: Using various filter combinations are a great way to manage PSRR in your designs as long as they are applied correctly.
What's next?
In addition to ensuring RF systems' PSRR requirements are met, there are many other common RF design best practices that you should follow and peer review. Download this FREE checklist before you tackle your next RF design challenge!
