Load Pull measurements report the non-linear RF performances of an active device under non-50 Ohm conditions. When driving an active device into compression and varying the load, a load pull user will get great insight about a device’s behavior and performance – in its non-linear region as well as its performance across the impedance plane. Linear models have struggled for years in predicting the true non-linear behavior of a device. Load pull is based on real measurements and provides great accuracy on how a device will behave for a given bias point and frequency.
Load pull can be as simple as a fundamental tuner being utilized for ruggedness testing, which comprises varying the load of an active device at a constant VSWR. Load pull can become more complex when an RF designer is interested in seeing the effects on linearity when changing the second harmonic impedance on the source side of the active device.
In the past passive load pull was the simplest way to get access to non-linear non 50 ohms data, but these days engineers expect more from load pull. It is not uncommon to see in the same lab a simple scalar load pull setup for quick validation and a high end vector load pull setup, which provides advance capability for multi-parameter characterization and more advanced design work. With increasing modulation bandwidths active load pull setups are in high demand to address markets like 5G.
Below we provide more information on basic scalar load pull, vector load pull, pulsed load pull, hybrid load pull as well as time domain load pull.
Scalar Load Pull
This is a classic form of loadpull where a pre-calibrated source and load tuners are connected directly at the input and output of the device. The precise moment of the pre-calibrated tuners allows generation of complex reflection factors in highly repeatable manner.
Power meters at the input of the source and output of the load tuners are connected through couplers to search for the optimal gain and output power by synthesizing the source and load impedence using the pre-calibrated tuners. The optimum load impedence extracted is the complex reflection factor where maximum output power is measured using the maximum power transfer theorem. Similarly, at the input side the optimum source impedence is the point where maximum transduces gain is measured. The optimum load and source impedence are completely dependent upon the pre-calibrated tuners repeatability.
The typical scalar load pull setup comprises a signal generator, two RF power sensors, a power meter, some DC bias networks and two fundamental tuners. The input and output passive block comprising of couplers, bias tees are also pre-calibrated, and S-Par are defined in the measurement setup.
Vector Load Pull
More advanced loadpull setup is referred as Vector loadpull or VNA assisted loadpull. In this measurement setup the forward (a1,b1) and reverse (a2,b2) travelling waves are measured using the two directional couplers connected at the input and output of the DUT. Measuring “a” and “b” waves forms allows vector loadpull to calculate the real time tuner impedence presented to the DUT and does not fully rely on tuner calibration and repeatability. However, tuner calibration is useful to be able to steer tuning into right area of the smith chart without long search.
The typical scalar loadpull comprises of two directional couplers, source and load tuners and the Vector Network Analyzer with receiver access capability. An absolute calibration at DUT reference plane is required to fully compute the 8-term error model, without leaving composite terms, as is the case for a standard small signal calibration.
The measured RF parameters include:
- Pin, Pout, Gaintrd ,Gainpwr
- PAE, ACPR
- ΓLoad, ΓIN
In this measurement setup both Power In delivered to the DUT and the PAE are calculated using the captured waveforms. All RF parameters are captured in a single shot which dramatically increases the measurement speed. All harmonic power levels are also measured by the Vector receiver.
Pulsed Load Pull
For high power non-linear devices, the self-heating and memory effects severely affect the performance when excited in the continuous wave mode. Using pulsed stimulus signals, the device can be characterized at the higher peak power levels up to saturation to which the devices will be subjected in their intended use. If the devices are excited in pulsed mode, they can be operated at higher peak power level at a reduced risk of device breakdown and a better control of operating temperature. In many applications Pulsed approach represents a realistic operation for devices, like RADARs.
Another important application of pulsed measurements is pulsed I-V and pulsed S-parameter. This approach has been widely used to extract electro thermal models of different device technology. Pulsed operation when combined with high DC voltage levels can also be very interesting for the validation of nonlinear models which considers thermal and trapping effects and can be used to test large devices that could not be tested at the same power levels under CW conditions due to excessive self-heating.
In Pulsed LP measurements both DC and RF can be pulsed. A pulsed load-pull test bench comprises of pulsed bias tees, a primary DC pulse generator, RF source synchronized to a pulse generator, and digitizing scope. The digitizing scope is used to monitor the Pulsed DC characteristics within the pulse. Focus LP uses two different options for Pulse generators, AU5 or MPIV. All passive components of the test setup (including the programmable tuners) are wideband enough to let the pulsed signal pass without any distortion.
For RF characterization there are two different scenarios
1. Scalar LP
Peak power meters synchronized to primary pulsing instrument are used to monitor the RF power within the pulses.
2. Vector LP
Measurements are made using the Vector Network Analyzer receivers which are synchronized with the primary pulsing instrument.
Active Load Pull
Active load pull tuning is, because of limited tuning range of passive tuners, additionally reduced by fixture and probe insertion loss, the only method allowing test engineers to reach up to |Γload|=1 (RDUT=0Ω) to match any DUT at its reference plane. In closed loop active tuning, part of the extracted RF power from the DUT is amplified and fed back into the DUT output port, creating a virtual load. Since the re-injected power can be higher than the original extracted power, Γ can be >1, meaning that insertion loss can be compensated. Using two (or more) synchronized (coherent) sources allows open-loop active (and harmonic) tuning; in fact the loop here is also closed, but, instead at RF, it is closed at IF synchronizing reference.
WideBand Impedance Tuning
As communication standards require more and more channel bandwidth the need for wideband tuning is increasing. Wideband impedance tuning is now possible as RAPID’s active loop has 100MHz of instantaneous bandwidth allowing users to perform real time modulated measurements. Many spectrum analysis features are also available, such as ACPR, EVM, CCDF, Spectrum mask for advanced modulation standards like LTE and 802.11a/b/g/n/ac.
Hybrid Load Pull
Vector receiver based loadpull setup can be upgraded to a hybrid setup. As per its name a hybrid load pull system includes both an active loop as well as the passive tuners. The hybrid system has all the advantages of speed and tuning range of an active system as well as the power handling of a passive system.
Mechanical tuners due to their inherent passive nature have limitations in terms of impedence synthesis for devices with very low impedence, meaning realization of impedence at the edge of the smith chart. These limitations add up when any loss between tuner and DUT reference plane is introduces as in case of On-wafer vector loadpull where a coupler and RF probes are added between the tuner and DUT reference.
The principle of operation for hybrid LP is that the passive wideband (fundamental) tuner creates a reflection factor near the optimum load impedance (which cannot be reached using the passive tuner alone) and the active power injection creates the additional reflection factor needed to reach the border of the Smith chart (Γ=1). Because the mismatch conditions are better, the required power to be injected is, typically, lower than in the case of full Active LP.
To increase the reflection factor at the probe tip (DUT) and minimize the power loss we must maximize S21 and minimize S11. Any mismatch loss must be compensated by additional injected power in a hybrid (active/passive) tuner.
Hybrid tuning is not a panacea. Whereas it allows high VSWR at DUT reference plane, it still remains a rather complex test system with feedback power amplifiers and, often, a second, synchronized, signal source, plus the requirement for in-situ vector power wave measurement, possible through directional couplers inserted between the DUT and the tuner; this on the other hand reduces the tuning range and increases the need for even higher power amplifiers. Passive pre-matching tuning in hybrid systems reduces the requirement for high power from the feedback amplifiers, but only to some extent: passive tuners are not lossless. Tuner loss increases rapidly with reflection factor and so does the power requirement. The critical quantity in tuner loss calculations is “mismatch loss”.
Mismatch loss is S212/(1-S112)
For high S11 values, as needed to pre-match for enhancing the passive reflection factor with active injection in a hybrid configuration, it happens that any increase in insertion loss S21 (due to cables, adapters etc. between tuner and DUT) is multiplied by a factor M=1/(1-S112).
Typical values of the multiplication factor:
S11=0.9 (VSWR=19:1) -> M=5.3
S11=0.96 (VSWR=50:1) -> M=13.
Time Domain Load Pull
The Vector loadpull setup can be extended to fully calibrated time domain measurements by adding a phase reference. A phase reference is a device that has been pre-characterized to uncover the phase relationships between the fundamental and harmonic frequencies.
Time domain characterization technique is aimed at high power, non-linear devices that allows insight in process characteristics, comparisons to knee and breakdown voltages, and to have the ability to place the device into a known class of operation and view the resultant waveform. The shaping of current and voltage waveforms real time in the measurement cycle by accurately terminating the fundamental and harmonic frequencies enables the designer to choose the correct PA mode of operation and how to achieve it.
First step in time domain measurement characterization is the standard small signal calibration, based on established techniques e.g. TRL which are Ratio measurements which ends up with composite error terms. The absolute calibration involves reference to known standards in terms of power and phase. Absolute power calibration is achieved with a calibrated power meter. For relative phase (time) a pre-calibrated phase reference HPR is used.
The measured RF parameters include:
- Pin, Pout, Gaintrd ,Gainpwr
- PAE, ACPR
- ΓLoad, ΓIN