1. Home
  2. Opinion
  3. To De-embed or Not?

To De-embed or Not?

To De-embed or Not?

by Alan Ake

Vice President of Applications Engineering/Technical Marketing

Guerrilla RF

As applications engineers frequently supporting ultra-low noise amplifiers (LNAs), questions often arise regarding the presentation of noise figure (NF) data to customers. Clearly, de-embedding NF to the input and output pins of the amplifier results in the most impressive numbers being put forth into the marketplace. However, how valid is this approach? How can the customer best evaluate the performance he or she is likely to observe in his system?

Losses associated with external matching, sub-miniature version A (SMA) connectors and input traces are usually minimal at lower frequencies such as VHF and UHF. However, these become more and more relevant as the device operating frequencies rise to 900 MHz and higher. The absolute level of the NF in question is also an important consideration since an embedded vs. de-embedded measurement difference of 0.1 dB is much more significant to the 0.3 dB NF of an infrastructure-class LNA than it is to the 2.5 dB NF of a general-purpose gain block.

Most high-frequency, ultra-LNAs require some simple external input matching to include discrete inductors, capacitors and short lengths of transmission line, and there is no justification for subtracting these losses from the reported measurements since they are required for successful implementation of the device. It could be argued that the losses of the input SMA connector and the short 50 ohm transmission line from the SMA connector to the first matching element could be legitimately subtracted. However, these losses are typically << 0.1 dB at frequencies below 2.7 GHz. Therefore, taking the trouble to de-embed these small losses is of little value.

The argument changes for devices and modules, which are fully internally matched since any length of transmission line connecting to the device input is arbitrary. In these situations, a device manufacturer is justified in de-embedding NF and other device measurements to the input and output pins.

In summary, the end goal is to provide design engineers with the information they need to accurately assess the performance of a particular device in their system. Parameters such as NF should always include losses that would be unavoidable with a practical application of the device. If an LNA requires matching on the input to achieve a stated level of performance, then any losses associated with that matching network must be included in the reported NF. Again, modules and other internally matched devices are fully justified in presenting de-embedded NF values since their implementation requires no associated overhead which would degrade the device performance in a customer application.

This is another example of the value of integrated matching with its associated ease of use and small application footprint. Guerrilla RF is continually striving to push the envelope in terms of both performance and ease of use. Coming soon, our new Guerrilla Bloc™ family of fully integrated modules will remove all customer requirements for layout and external component placements for a wide range of LNAs and power devices. www.guerrilla-rf.com