There are many things that can go wrong when designing with a hot swap controller. For example, the hot swap can trip at an unexpected current value or a current monitor may report inaccurate measurements. As a result, the integrity of a system which relies on protection by the hot swap can now be at risk. Optimizing sense resistor layout by using four pads can help avoid failures and create a robust hot swap design.
It is important to understand that sense resistors can be very sensitive. Factors such as size, type and manufacturer can lead to varied results. The idea of using four pads with a two terminal resistor to optimize measurement accuracy can hold true for some resistors but not for others. Ensure your design is operating as expected by testing it with your sense resistor and layout in the prototyping phase of your development. If you are looking for a tested design to quick start your development, simply copy the BOM and layout from one of TI's Hot Swap Controller EVMs such as the TPS2490, LM5067 and the LM25066I!
The sense resistor, RSENSE, is a key component for sensing current with a hot swap controller. As power requirements in server and telecom applications increases, sense resistors as low as 300µΩ are becoming more common. At these levels, effects such as solder resistance can play a significant role in throwing off your measurement. Let's take a look at a simple two pad current sense:
Since we are measuring the voltage at Sense + and Sense -, we must look to the golden rule V = IR. Since current is constant throughout RSOLDER+ RSENSE + RSOLDER, the current sensed, ISENSE = (VSENSE + - VSENSE -) / (RSENSE + 2RSOLDER). For example, if RSENSE = 300µΩ and RSOLDER = 15µΩ, then your current measurement will be off by a factor of 10%. Avoid these inaccuracies by adopting four pad sensing.
There are several layouts you may choose from for your sense resistor. We will discuss three of the common choices, but first let’s understand how four pad sensing works.
The layout is broken out into four pads. Two of these pads (RSOLDER) carry the high current flow while the two remaining pads (RSOLDER_S) carry very little current, virtually 0A. This eliminates RSOLDER in the equation and the result is ISENSE = (VSENSE + - VSENSE -) / RSENSE. Now the accuracy of the current measurement is purely dictated by the accuracy of RSENSE and the measurement of VSENSE .
A four pad solution is not always the answer; the three figures below outline the pros and cons of three common layouts:
Figure 1 is a simple two pad approach, but note that the sense lines are taken from the inner center of the chip. The high currents flowing on the sides of the connection are balanced, allowing a good measurement of VSENSE + and VSENSE - . This type of layout is simple to route and is easy to assemble on a PCB. This layout is recommended for designs where the value of RSENSE is relatively large compared to RSOLDER (100:1).
Figure 2 is typically the best layout for accuracy as the sense lines route to the inner center of the chip and uses four pad sensing. The drawback to this approach is that the pads have a higher risk of getting shorted in assembly. If this occurs, you may not see the short since the resistor can be covering the pads; this would result in a two pad sense.
Figure 3 is another four pad sense layout, but the sense pads are at the bottom. This layout is easier to assemble than Figure 2 and you can easily see if there are any shorts between the pads.
Each of the three layouts have been used and tested on TI hot swap EVMs. Figure 1 and Figure 2 layouts are recommended for sense resistors in package type 2512 or smaller. Figure 3 layout is recommended for larger or specialty sense resistors which may incorporate cutouts such as the 3921 size resistor used on the LM25066I EVM.
Thanks! -Alex