by Harley Berman
Over the past few weeks, we have spent a considerable amount of time here at Marki Microwave studying the performance of our limiters. Our goal has been to understand how our customers typically use limiters, what problems they typically run into when integrating limiters into their systems, and whether we are able to solve any these problems for them. In our first publication on limiters, we investigated ‘The Self Jamming Problem’ which occurs when limiters are placed in radar systems. In summary, the problem is that when a limiter receives a high-power pulse, there is a time delay between when the pulse turns off and when the limiter can receive small signal information again. This is called the recovery time and the shorter the recovery time, the less time your receiver spends in the dark. We showed experimentally that the Marki Microwave HLM and DLM limiters have a nearly instantaneous recovery times making them ideal candidates for systems that require extremely fast recovery times.
How Do Marki Limiters Respond to Pulses?
This result got us thinking about what implications this has on the performance of our limiters in pulsed applications beyond recovery time. In the past, we have had customers request more detailed information on the output power versus input power, and relationship when the input is pulsed with various duty cycles. Using the results of the recovery time experiments from the tech note linked above, one can reason to themselves that since our limiters respond to a pulse instantaneously that the output power versus input power would be independent of pulsed or CW inputs.
How can we reason this? Well, we must start with an understanding of what would cause an input versus output power relationship to be dependent on duty cycle in a limiter, and then look at the results of our investigation of recovery time to come to this conclusion. In all limiters, there is a performance metric called spike leakage. This describes how much energy of a pulse is transmitted through a limiter before the limiter starts to limit the signal, and it is directly related to the time delay in turning on the diodes.
If the diodes can respond to a pulse instantaneously and begin limiting the instant a pulse is incident on the input, then there will be virtually zero spike leakage since the limiter hits its flat leakage instantaneously, even if the duty cycle is very low. In limiter technologies such as PIN diode limiters which have long recovery times, the diodes do not react instantaneously to their stimulus, and they exhibit high spike leakage. This has the effect of causing the pulsed characteristics of PIN diode limiters to differ from their CW characteristics especially at low duty cycles. This is a disadvantageous feature of PIN diode limiters for system designers since they would prefer the performance of the limiter to remain constant across input powers and duty cycles.
If a limiter can react to a pulse instantaneously, by both limiting immediately upon an incident pulse (low spike leakage) and relaxing to a low insertion loss state immediately after a pulse turns off (low recovery time), then it will have input versus output power characteristics which are completely independent of the duty cycle of the input signal. In a way, the spike leakage is just an inverse of recovery time since recovery time is how long it takes a limiter to turn off, and spike leakage is a measure of how long it takes to turn on with the caveat that recovery time is a measurement of time and spike leakage is a measurement of energy (power x time). We showed in our last tech note on the recovery time of our limiters that they can relax to a low insertion loss state virtually instantaneously, exhibiting their extremely low time delay in turning off the diodes. This result makes it a good candidate for a limiter with very low spike leakage and thus would exhibit identical input versus output power characteristics across duty cycles. While this is a nice thought experiment, it doesn’t mean anything without measured data to back it up.
Pulsed Measurements: Experimental Setup
The measurement setup used to study the pulsed output power versus input power relationship is quite simple. High-power couplers are placed on the input and output of the limiter with power meters attached to the coupled ports to monitor the power entering and leaving the limiter. A high-power load is placed on the output of the final coupler and a high-power pulsed input is injected incident on the input of the first coupler. The pulsed input is then slowly ramped up in power while the corresponding output power is monitored. Below is a block diagram of this experimental setup which was created in our lab to characterize this behavior.
For this experiment the pulsed signal was generated by our signal generator and was amplified by one of our benchtop power amplifiers. The couplers used were two C20-0R620 directional couplers and the limiter DUT for this measurement was an EVAL-HLM-20P. Two peak power sensors were attached to the coupled ports of the couplers to monitor input and output power levels. The input power was slowly ramped up to as close to the absolute maximum peak power level we could reach in each of the experiments conducted.
Results of Output Power Versus Input Power Testing
The first set of measurements on this setup aimed to investigate any trends between the pulsed and CW input vs output power relationships. We expected that since these limiters have a near instantaneous recovery time that the CW and pulsed measurements would overlap each other. Using a 1% duty cycle, the input versus output power relationship was determined across three frequency points within the devices operating range and it was found that they do in fact overlap the CW measurements, just as expected.
HLM-20PSM’s Nearly Zero Spike Leakage
With our assumptions confirmed about the input versus output power relation, the next set of experiments aimed to take this to a more extreme limit. In all the plots shown above, the duty cycle is very low at only 1% but our customers will often use our limiters with much longer duty cycles. The next set of experiments aimed to investigate the performance of our limiters over the various duty cycles our customers may use. This was done by conducting the same experiment outlined above, only slowly increasing the duty cycle of the pulse each time the test is conducted. The plots below show these results over three frequency points and three duty cycles of 5%, 15%, and 30%.
The results are as expected and show a near perfect overlap of response over various duty cycles as well! This happens because our limiters are responding to stimuli virtually instantaneously and begin limiting the instant a pulse hits the diodes, highlighting the near zero spike leakage of our limiters. This result makes even more sense when you consider that the GaAs Schottky diodes in our limiters are created on a THz process allowing them to switch on and off at incredibly high speeds. This combined with the novel circuit designs our designers have implemented in our limiter products, allows us to offer devices with virtually zero spike leakage and nearly immediate recovery times making these devices the best choice for high linearity, wide bandwidth, pulsed applications.