Dealing with Line Injection Issues (J2120A, J2123A)
Here is some information on the Power Supply Rejection Ratio (‘PSRR’) measurement setup using the J2120A line injector.
Depending on the load current and impedance you are trying to drive, the output of the injector may drop several volts.
In the vast majority of cases, the line injector is almost certainly working correctly. It is often the case that the test setup impacts the results and that a large voltage drop at its output is seen. We do get many questions about the J2120A and its operation.
After reading the following if you are still having issues, please let us know and we will be happy to have a call with you to discuss your particular measurement and test setup.
The J2120A is being asked to supply a certain amount of current (over frequency). If the impedance is too low, the J2120A runs out of steam basically. Why is this the case?
The amount of load current (out of the J2120A) is FREQUENCY DEPENDENT because the input impedance of the load (power supply input) varies with frequency and can fall if there is an input capacitor on the input of the power supply circuit. This is usually the case in normal designs.
In general, the power supply voltage + noise that is being supplied to your DUT is driving some amount of impedance that varies over frequency. You can actually test this with the J2121A (https://www.picotest.com/downloads/AppNoteDCDCconverterInputImpedanceVer03Final.pdf ) and your VNA. There is usually one or more input capacitors on the input to the DUT. The impedance of a capacitor is too low, in general, and especially at its resonance point where it can be very low (milli-ohms) that will cause the DUT to draw too much AC current through the J2120A. And its output voltage will drop as per the graphs below.
Please see this page https://www.picotest.com/products_J2120A.html. Look at the Detailed Specifications tab. You will see the reason for the issue more clearly.
The more current, the more voltage drop. What you have with the J2120A and an input capacitor to your circuit is a very effective low-pass filter. As shown on the https://www.picotest.com/products_J2120A.html page in the Detailed Specifications Tab, you will further see the reason for the issue. The more current, the more drop from the input of the J2120A to the output.
The solution is to remove the capacitor or replace it with a smaller cap if you can so that the impedance is not too low at higher frequencies.
Other Usage Notes
Keep Interconnect as short as possible to limit cable inductance: If you can only get to a few kHz of bandwidth that might mean that the input impedance is still in the way of the injection somehow. Likely the interconnect cables from the J2120A to the DUT are the issue. The cable impedance of your interconnects will FURTHER affect the drop. Keep it as short as possible.
Have some current load on the J2120A/Input to your circuit: It should be noted that the J2120A needs some output current in order to function properly. You MUST have operating current to achieve a low output resistance from the J2120A as described by the Shockley equation response in the graphs above. See this video for an example: J2120A voltage drop and output resistance. If you are operating at near zero current (and have very little board capacitance) you should increase your input voltage to achieve the required output voltage and add load current, externally (external resistance on Vin at the DUT) if your board requires very low input current, and then you will get the J2102A to have a low resistance drive signal (allows greater modulation signal to reach the DUT).
The impedance of the probes use matter: Some PSRR articles below use high impedance probes, some use the PML 111O 1MHz probe (mid-range impedance), some have direct soldered cables, and others use 50 ohm probes. Theoretically speaking, the probes should all give similar results. But the lower the probe impedance, the lower the probe noise. This might not be an issue when measuring -50dB but it can be when trying to measure -100dB. Separate the DC loading from AC loading. DC loading changes the operating point and a DC Block can fix this. For AC loading, the probe should be at least 5-10X the impedance of the DUT. In most cases, 50 ohms meets these criteria providing lower noise. 50 Ohms is always lower noise, but can it tolerate the DC Voltage is the question. Amplifiers add noise but are helpful when the noise limit is the instrument and not the DUT. See https://www.signalintegrityjournal.com/blogs/15-extreme-measurements/post/2484-the-ultimate-power-rail-noise-measurement
Ground loops can get in the way: It may be a cause for error depending on the magnitudes you are trying to measure. It’s always there, though it’s not always the case that it will impact the measurement. See Tektronix Innovation Forum Webinar: Ground Loops - What They Are, Why We Care, and How to Fix Them, 2023
Solving the Voltage Drop: A good solution to the voltage drop is to implement a remote sense option on your main bus power supply, to adjust the input to the J2120A up to compensate for the voltage drop. A remote sense board, setup images, and an app note are available to guide you through this.
Remote Sensing App Note https://www.picotest.com/images/download/AppNoteRemoteSenseVer06Final.pdf If you need a power supply that can handle the voltage drop, we have that too. The P9610: https://www.picotest.com/products_P9610A_11A.html.
ALWAYS FIRST measure something you know and of the SAME magnitude.
Higher Power, Higher Bandwidth PSRR Testing
Lastly, if you need higher signal amplitude and bandwidth, please see the P2124A Noise Immunity probe. It has a number of advantages that overcome the impedance issues: https://www.picotest.com/products_P2124A.html
Additional Videos and References
Power Supply Measurement Techniques Instructional Seminar https://www.youtube.com/watch?v=yA5XRg3tWm8&t=1s
Measuring Power Supply Rejection Ratio (PSRR)
https://www.youtube.com/watch?v=8i8Pb3wY7xo&t=13s
Measuring the PSRR of a Power Supply
https://www.youtube.com/watch?v=75fEwjXspbA (Robert Bolanos)
Power Supply Rejection Ratio (PSRR) Measurements using an Oscilloscope (Keysight)
https://www.youtube.com/results?search_query=J2120A
Articles
Techniques for Accurate PSRR Measurements
https://www.ti.com/lit/an/slyt547/slyt547.pdf
PSRR App Note
https://www.omicron-lab.com/fileadmin/assets/Bode_100/ApplicationNotes/PSRR/App_Note_PSRR_2_0.pdf
Application Examples
Dealing with input capacitance and current into your converter
We need to perform the following transient input at the DCDC converter when loaded at the output.
Conditions & Question
Rise from 3.2V to 4.0V @100mV/us,
Stay there for 500us
Fall to 3.2V @100mV/us
When the VRM is loaded, the input current is around 600mA and the input capacitance is ~20uF (While it is helpful to remove input capacitance we prefer not to reduce the cap value). So, we would need 2A to charge the input capacitor at that rate of rise + the DC input current and then, to sink about the same amount of current.
Solution
The J2120A should be ok for this application. The J2120A output resistance vs current is shown above, but just choosing some round numbers, let’s assume that the output resistance of the J2120A at 600mA is about 250 milliOhms. Ignoring the interconnect cables, this forms a low pass filter with your input capacitor at 32kHz (higher if the capacitor is ceramic as the LPF frequency depends on the cap ESR to some extent). So, this should meet your slew rate easily given the J2120A bandwidth capability. Then you want it to “stay there” for 500uS, this is 1kHz equivalent and so there might be a little droop in the voltage, but probably ok. The J2120A is a SOURCE ONLY modulator, the sink is the VRM (your power supply). You can “fool" this piece (the lack of sinking current) by adding a resistive load to GND at the output of the J2120A. You don’t want the current to drop to zero.
We could also make a custom J2120A line injector for you that can sink and source. There are many customized versions available. Please just ask. We don’t recommend amplifiers, since the 20uF will likely cause an amplifier to be unstable, which is the reason J2120A is an open loop style injector.
Note that the 20uF cap will also provide a return current through the cable, creating a ground loop (similar to the 2-port impedance measurement). So if you want to measure from the same GND as you are injecting, you will need a GND isolator and since you need 1kHz, this would require a J2113A Differential amplifier (https://www.picotest.com/products_J2113A.html) on the receiver side, but only if you measure from the same ground you inject from. This is independent of what device you use to inject the signal.
Please see this video presentation about ground loops for the Tektronix Innovation Forum (Link coming shortly after 7/1/2023)