Power Tips: How and Why to Sense Current – Part II

It is almost always necessary to measure some sort of current.  In my last post, I covered two main reasons why you should measure current and a few ways to do so using loss current sense techniques. This post will focus on loss-less current sense techniques.

Use what you already have! We are going to talk about two ways to sense current using circuit elements that are already present.  Those two methods are inductor DCR sensing and FET sensing.

Inductor DCR sensing is not super accurate, but good enough.  Typically, the DCR of an inductor is on the order of +/- 10%. Factor in temperature changes to the copper and you can get some pretty inaccurate measurements.  One saving grace is that after the DCR network, you end up with a pretty clean signal that does not have switching noise.  Figure 1 shows the network necessary to extract the current information from the DCR of the inductor.

Figure 1

The components of this network are chosen by the equation below:

There are a couple of factors that need to be taken into account when setting up the DCR network:

  1. Maximum signal size that the controller or sensing circuit can handle - it may need to be divided down.
  2. Temperature compensation - components with a negative temperature coefficient can be used to help keep the DCR constant over temperature.

Usually, DCR sensing is used in a multiphase configuration for current-mode control.  It is very easy to implement current sharing between multiple phases using this technique.

Use FET sensing, but watch out for switching noise!  When FETs switch, a lot of noise can be generated.  This noise needs to be filtered.  Figure 2 shows a FET sensing scheme and how noise can interfere with the measurement.  A couple of things can be done to mitigate this noise, but they do not come without penalty.

Figure 2

  1. Filter using an RC network.  This is okay, but it can round off the current signal and the edges are not clean.  The rounded off current sense signal can lead to jitter and other noise issues.
  2. Leading edge blanking.  This is a technique where the first part of the current sense signal is ignored.  The main issue with this is it leads to a minimum on-time, if there is an issue the duty cycle can only be reduced so much.

Figure 3 shows the penalty from the two fixes shown above.

Figure 3

Loss-less current sense techniques are not as accurate as using a precision resistor, but avoid the efficiency and power loss issues.  Usually the loss-less methods are good enough for the application that the efficiency gains outweigh the accuracy issues.  The methods for current sensing covered in the last two blogs are by no means the only ways.  Leave a comment and let me know how you do it!

Related resources: