Core power rails are now routinely in the 1 kAmp to 2 kAmp range, requiring a power distribution network (PDN) impedance in the range of 20 µΩ to 30 µΩ.  

I am known for saying “Measure something you know before measuring something you do not; and of the same order of magnitude.” This makes sense. It provides a validation of your measurement setup, and it assures that the measurements you are going to make are within the dynamic range of the setup. If you wanted to measure 100 µΩ, proving that you get the correct answer when measuring 1 Ω  would be less than adequate. It does not prove that 100 µΩ is within the dynamic range of your setup.

This results in a new challenge. Where do we find a 20 µΩ known measurement standard? Resistors are available off-the-shelf at 250 µΩ, but I have not seen any usable test standards in the 20 µΩ range. With a significant number of customers now looking for these standards, and Picotest’s goal of helping our customers with their power integrity measurement challenges, we have created resistor standards.  

Creating such standards is difficult. Using the same 4-wire Kelvin mount knowledge we have developed for our other products, we created standards by soldering copper slabs to our component mounts. These allow measurement connections using either SMA cables or the P2102A 2-port probe, using universal Kelvin probe pads on the back side of the mount. Using different copper thicknesses and some machining, we can create resistors in a wide range of accurately known values. A range of resistors from 1.5 µΩ to 125 µΩ, created on our 4040 mount, are shown in Figure 1.

Figure 1. Resistor standards from 1.5 µΩ to 125 µΩ , created on a 4040 mount.

The next question is how to validate the measurement of the resistors to use them as measurement standards. The resistors are measured using a NIST-traceable DC measurement. This can be achieved using a constant current power supply and a digital multimeter (DMM) or an integrated source measure unit (SMU). The first method is shown in Figure 2. If the constant current power supply is accurate and the DMM is accurate, this will provide a reasonable measurement within the accuracy of the power supply and the DMM.

Figure 2. Using a 1 Amp constant current power supply and a 6.5-digit DMM to measure a 40 µΩ resistor standard.

There are a few tricks to performing this measurement accurately. First, set the DMM to the slowest, highest resolution measurement. With Picotest’s M3500A meter, this is the 6.5-digit slow. Second, we use measurement averaging to minimize noise. Here, we are using 100 count average. Finally, we use delta-offset methodology. This involves nulling the DMM at one current and measuring at a delta current. For example, we can have the power supply ON and set it for zero Amps. We then null the meter in this state and immediately increase the power supply to 1 Amp. The difference in voltage between 1 Amp and zero Amps is the value of the resistor, shown here as 39.6 µΩ. This is accurate to the combined accuracy of the DMM and the power supply.

Achieving a more accurate result requires more precise equipment. For that, we turned to Keithley. Combining their SMU, which can source up to 7 Amps, and a separate 7.5-digit DMM provides the maximum accuracy and a quick measurement. Keithley was kind enough to measure two of our resistors using their setup. They used a delta offset method, with averaging using a 5 Amp source. The 5 Amp source increases the signal level. With the noise level essentially fixed, this improves the signal-to-noise ratio (SNR) by 14 dB compared with a 1 Amp source. The 7.5-digit DMM offers a higher resolution measurement as well. The results of 200 µΩ and a 33 µΩ nominal resistors are shown in Figure 3.

Figure 3. Measuring 20 µΩ and 33 µΩ with a Keithley SMU and DMM.

Using appropriate adapters, this equipment can be connected to the device under test via either SMAs or the P2102A 2-port probe. Note that the 5 Amp current is a bit high for the probe and even the SMAs, so it is better to measure at 1 Amp, if possible. If not, you can try to get the measurement quickly so that the current is not flowing long enough to significantly heat up the connector or probe pins. The Keithley SMUs can perform pulsed measurement to accommodate this.

Both measurements were performed at DC. With sufficient isolation (common-mode rejection ratio or CMRR) between the source and measure instruments, there is no ground loop error and the measured value is correct. How much isolation is sufficient? For the answer to that question, see this article in Signal integrity Journal “The Challenge of Measuring a 40 µΩ, 2000 Amp PDN with a 2-Port Probe: How Much CMRR is Needed?”

Want to know how to validate the CMRR of your ground loop isolator? Let us know on the Picotest Online Forum, tell us in the comments below, or email us at info@picotest.com.