Measuring LDO Noise Spectral Density for a Novice

If you recall, I mentioned back in my blog IMS2016: Preview and Preparation in May of this year that I would be transitioning from supporting ADCs to supporting space products at Analog Devices. I have spent the last few months covering some topics that I had previously committed to for ADCs and now am moving on to discuss topics more relative to my current work in my new group. There are many additional things to think about when working with space products that aren’t necessary to consider for commercial products. One example would be harsh radiation environments that are present in space. For a little more insight into that topic you can check out a blog from one of my colleagues at Jupiter: The IC Danger Zone, Part 1. Another example is more stringent and specific test requirements. Keep in mind that systems designed to go into space must be essentially failure proof. It is reasonably simple to climb a radio tower and replace a defective or damaged radio for a cellular network for example. However, it is not feasible to fly up to a satellite in orbit and replace a defective or damaged component. The system design must be robust and is generally set up with redundancy so that the design should not fail. As you can imagine this translates back to the system components which must be tolerant of radiation in space, operate over temperature extremes, and must be hermetically sealed to name just a few requirements.

In this vein one of my recent tasks when evaluating a product for space operation was to characterize the noise spectral density (NSD) of an LDO, otherwise known as a low dropout voltage regulator. I have performed many different measurements in my ten plus years in the industry, but I had yet to attempt to make such a measurement as this. When performing an NSD measurement on a VCO (voltage controlled oscillator) or on an amplifier the setup is generally pretty simple using a signal source analyzer. It is relatively easy to measure noise at close in offsets of 10 Hz or 100 Hz from the carrier when the carrier frequency is at 100 MHz or 1 GHz, however, it is an entirely different task when the carrier frequency is at DC in the case of an LDO. I researched this topic and talked to many different folks within Analog Devices before finding the answer. I offer this information to you as the reader so you can avoid the lengthy time that I spent attempting to figure out what turned out to be a relatively simple measurement. As you may have heard before it generally is all about having the right tool and I had to find the right tool and get a little expertise from some folks on how to set up and use this tool.

In order to perform an NSD measurement on an LDO there are two pieces of equipment that are necessary. The first is a DC power supply to power the LDO input voltage. In this case, a Keysight (Agilent) E3631A DC power supply such as the one pictured below is used. The second and most important piece of equipment is the Keysight E5052B Signal Source Analyzer with baseband input option. The key here is to have the baseband input option which is an option that includes a special port on the analyzer that can accept a DC input. Obviously, the output of the LDO is DC and if we are going to be able to measure noise at close in offsets of 10 Hz, then an analyzer that accepts DC is the ideal solution.

The test setup as I mentioned is actually pretty simple. The DC power supply should be connected to the DC supply input of the LDO. The LDO output is then connected to the baseband port of the Keysight E5052B Signal Source Analyzer as shown. Note that the standard connection port on the E5052B will not work for this measurement since it does not accept a DC input nor does its frequency response go below 10 MHz. The E5052B must have the baseband input port option.

For this particular case the LDO under test requires an input voltage of 5.5V and has outputs of varying voltages from 3.0V to 5.0V. Banana to clip lead cables are used to connect the E3631A DC power supply to the VDD voltage input of the LDO evaluation board. The LDO evaluation board has SMA connectors on all the voltage outputs which allow for easy connection to the BNC baseband port input of the E5052B via an SMA to BNC adapter. A BNC to SMA cable can alternatively be used. I have not included the data here but I did compare the noise measurement with the BNC to SMA adapter and with the 24-inch BNC to SMA cable. I found no appreciable difference in the noise measurement.

Once all the connections are made the E5052B must be set up to properly measure the NSD of the LDO. There are a few simple steps to set up the E5052B to make this measurement. Below is a list of settings that I used, but some of these could be different depending on the specific application requirements. These settings should at least be a good place to start.

On the Keysight E5052 Signal Source Analyzer configure the following settings:

  1. Press Meas/View
    1. Select Baseband Noise
    2. Show Window – Turn off all except Baseband Noise
  2. Press Format
    1. Select Volt/Hz
  3. Press Trigger
    1. Select Trigger to Baseband Noise
  4. Press Start/Center
    1. Select 10 Hz (may select different start frequency if desired)
  5. Press Stop/Span
    1. Select 40 MHz (may select different stop frequency if desired)
  6. Press Trace/View
    1. Select Spurious
      1. Select Omit (spurious can optionally be included)
  7. Press Setup
    1. Select IF Gain
      1. Select 40 dB (set appropriately to achieve desired noise floor)
    2. Select and set desired markers by pressing Marker and entering desired frequencies
  8. Collect data for the NSD performance of the LDO.
    1. The screen image can be saved from the System menu (Press System key)
    2. The trace data can be saved from the Save menu (Press Save/Recall key)

    Once the E5052B set up is complete the NSD measurement can be made. I would suggest first taking a measurement with the LDO under test disconnected from the E5052B to gauge the noise floor of the instrument. This trace can be saved to memory and displayed along with the current measured data. As an example, there is a typical NSD plot included below that shows the noise floor of the E5052B plotted along with the measured NSD of the LDO under test.

    One of the things I am learning in my role in the space products group is that there is a good variety of products to support including devices such as LDOs. I had previously viewed an LDO as a fairly simple component, but in learning to measure the NSD of an LDO I found it interesting the many things to think about that are different from a typical RF component. It is quite a bit different to look at measurements so close to DC. In some ways it almost seems easier to measure parameters in the RF domain, but perhaps that is because of what I learned early in my career since I started in the RF field. Stay tuned as I look at measuring PSRR (power supply rejection ratio) of an LDO in my next blog post.

0 comments on “Measuring LDO Noise Spectral Density for a Novice

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.