Harmonic Injection Doubles Ultra-Wideband TWT RF Output
by Dave Pfaff, Senior Engineer, dB Control
Wideband traveling wave tubes (TWTs) can fall short of an application’s power requirements due to low gain and low RF output power at the lowest frequencies. For example, an ultra-wideband TWT may have 60 dB gain at its mid-band frequencies, but only 10-20 dB at the low end. Fortunately, harmonic injection can be used to produce a new generation of ultra-wideband microwave power modules (MPMs) with twice the low-end RF output power of standalone ultra-wideband TWTs.
Using harmonic injection to improve a TWT’s output power is not new; design engineers have employed these techniques since the mid-1970s. However, using harmonic injection on an ultra-wideband TWT is a relatively new concept. In this case, the objective is to cancel the naturally-generated second (and/or third) harmonic within the TWT, and thus recover the available beam power back into the low-end fundamental frequencies. This can be achieved by driving the TWT input with an additional signal that is two times that of the fundamental drive frequency. The doubled signal is injected into the TWT at a preset amplitude and phase (relative to the fundamental signal) so that it cancels with the naturally produced second or third harmonic generated by the TWT.
Practical Advice for Harmonic Injection Schemes
Harmonic injection works well over a narrow frequency range and can be achieved at reasonable cost. Unfortunately, the power boost effect diminishes as the fundamental frequency is applied over a wider range. This is due to the gradual or rapid loss of the 180-degree phase and amplitude relationship between the injected signal and natural harmonic of the TWT. dB Control, a designer and manufacturer of high-power TWT amplifiers and MPMs for more than two decades, has developed and consistently enhanced its harmonic injection capabilities.
For example, as the fundamental frequency changes over the critical low-end frequency range, it is essential to control the injected harmonic’s phase and amplitude. Clearly, the wider the bandwidth that needs boosting, the more difficult the task and the more likely the injection circuit must be match tuned to each TWT. The task is easier if the host system can provide early information about frequency (or a narrow sub-band code), the fundamental input drive level, and be given enough lead-time to set up the injection circuit. Individual TWT characteristics can also be preloaded into the injection circuit.
Design engineers and injection circuit designer must be able to characterize and understand the TWT’s fundamental, harmonic and phase transfer curves. However, 100 percent cancellation across the entire band of interest is not practical, nor necessary. It’s all about meeting the specification. Examples of typical harmonic injection schemes include:
- Harmonic generator diode with phase shifter and TWT equalizer
- Channelized assembly with frequency multipliers, phase shifters and TWT equalizer
- Output feedback with filter and phase shifter
One reason project engineers must stay up-to-date in the design phase of the injection circuit is because the design may be handed over to a digital/analog team with little or no TWT experience. The engineer must understand and specify the unique input signals the harmonic injection system should amplify with fidelity. If left undefined, misinterpreted performance requirements may not be revealed until final system integration test.
High Helix Current: Not a Tube Killer after All?
A real concern for TWT engineers is that of high helix current which reduces TWT life and operational reliability and can be a “tube killer.” Like most TWTs, ultra wideband TWTs generally have low DC mode helix current and well-behaved RF mode helix current. But when harmonic injection is applied, the helix current can skyrocket. This forces near heroic efforts to refocus the tube while harmonic injection is applied. After much effort, the helix current in the injection band might be halved, but there are still frequencies where helix current peaks to destructive levels. Surprisingly, the TWT doesn’t die. Why?
A major TWT manufacturer confirmed its TWT design tool only predicts an ideal helix current. Frankly, this isn’t very helpful, and is completely blind to injection mode conditions. One particular TWT manufacturer that experienced this high injection mode helix current said that it was simply benign body current caused by a low potential secondary emission from the collector striking a collector pole piece. At first this seemed like a reasonable conclusion. However, with most TWTs we observed that:
- There is no significant temperature increase along the helix section of the TWT body proportional to the higher helix current
- It is unusual to see a high secondary emission without the TWT oscillating or producing spurious signals, (and in this case, there were none)
- The helix current increase is always an immediate step function when injection was applied
- The helix current increase is frequency dependent
- The power supply’s helix current fault circuit did trip correctly and turned off the high voltage
Note that the wideband TWT used for this project had a recommended helix current limit of 20mA and 25mA absolute CW. (An open helix failure under a non-injected condition later confirmed this was correct.)
Although it may be tempting to solve the problem of high helix current by resetting the current trip to a previously unacceptable level, this won’t protect the TWT at the higher non-injected frequencies, nor from the real beam intercept current also present within the injection band. In the injected band, the real helix current might increase while the benign current (explained later) may not. Even though most helix over-current faults are sudden spikes that would trip most any threshold setting, resetting the helix trip current is a risky procedure.
An estimate of RF output power can be made based on the measured harmonic and fundamental content of a TWT. The calculation focuses on the injected signal’s phase and amplitude error relative to the fundamental’s measured harmonic. While this estimating technique seems rather straight forward, the engineer must also include the injected signal’s power, which is the signal that is actually doing all the work, and express the injected signal in the equation. One solution is to add the injected power to the boosted fundamental. This produced predicted calculations that were close to the actual measurements.
Demonstrating How Energy Transfers Back to Fundamental Frequency
It helps to visualize two standing waves on the helix (two forward waves and no reflection coefficient implied). One wave is the fundamental driven harmonic as it grows in amplitude along the helix, while the other is the injected signal also growing in amplitude along the helix. The phase of the injected signal was externally set to be 180 degrees out of phase with natural harmonic and occurring only at one point near the output of the TWT – the location on the helix where both amplitudes are equal and at their maximum.
Up to the point where 180-degree cancellation occurs, the two standing waves are out of phase relative to each other and produce voltage null and sum points. Both signals get amplification power from the high voltage power supply (HVPS) via the cathode DC beam. The HVPS acts as a rectifier and draws current only on the negative sum cycles through the HVPS ground return sense resistor. This is the high injection mode helix current as applied to the HVPS, not the beam interception current which would prove destructive. This phase current is produced on the helix by the voltage created by the periodic phase difference between the two standing waves and takes the same path through the helix-to-ground resistance as does the conventional beam-to-helix intercept current, but at a very low potential.
The phase current and the beam interception current are illustrated in Figure 1. Note that this is not a manufacturer’s TWT design simulation, but rather a simple demonstration illustrating the process. Standing waves produce periodic currents between the waves and along the conductor. However, the HVPS is still in the current loop where current flows through R3 and not just between R1 and R2. (This is not the same situation as when a TWT is driven by multiple randomly phased signals which are not preconditioned and working to cancel a large harmonic.) The values used in Figure 1 were selected to replicate the total helix currents as actually reported. R3 at 49.9 ohms is within a typical range for a TWT’s helix-to-ground resistance. The green current waveform is rectified and its baseline is set at a reasonable -15.4mA. This is a typical non-injected RF mode helix current for this TWT.
Figure 2 shows the current through the helix resistor R3 (and D1, in green) at -15.4mA. The natural harmonic and the injected harmonic are 180 degrees out of phase, at 100% cancellation. Signal voltages are measured at the source output.
Figure 3 illustrates the peak current through the helix resistor R3 (and D1, in green) at -27mA when the natural harmonic and the injected harmonic are 45 degrees out of phase and the -5,000 V intercept current through R3 still remains at -15.4 mA (top portion of green trace). In comparison, Figure 4 shows the current through the helix resistor R3 (and D1, in green) at -55mA when the natural harmonic and the injected harmonic are 90 degrees out of phase and the -5,000V current through R3 still remains at -15.4mA. Figure 5 shows the current through the helix resistor R3 (and D1, in green) at -80mA when the natural harmonic and the injected harmonic are in-phase and the -5,000V current through R3 still remains at -15.4mA. In all three illustrations, signal voltages were measured at the source output.
In this demonstration, the power dissipated in the helix is 77.8 watts, worst case. (Beam intercept power .0154A x 5,000V = 77 watts plus the phase delta power of, .080A x 10V = 0.8 watt). This accounts for the benign appearance of the high injection mode current. But, if calculated in the traditional manner 80mA x 5,000V = 400 watts on the helix would mean instant death for this TWT, which did not happen.
From these demonstrations it can be inferred that the injected signal is continuously in a state of conversion into DC body current along the length of the helix in order to maintain cancellation of the TWT’s natural harmonic. This is how energy is transferred back to the fundamental frequency. Amplitude of the phase current is dependent on input drive, frequency, harmonic content, relative phase and amplitude relationships.
Although the increase of the beam intercept current along with phase current was not specifically addressed, it appears it was focused down collectively in the presence of phase current (by a combination of injection tuning and TWT focusing). Using a current subtraction measurement might have resolved the difference. However, the HVPS engineer will be challenged to design a suitable helix current protection circuit. Another challenge is that the ambient temperature becomes a stronger pushing factor on total helix current, which could potentially damage the TWT. However, it has not as yet been determined if this has a significant effect on yields.
MPMs with 2X TWT Power Worth the Effort
When it comes to ultra wideband TWTs, the lack of practical technical advice and real-world examples has resulted in some expensive and poorly performing designs. It can be difficult to achieve harmonic injection over a wider frequency range because the power boost effect diminishes. Matching the injection circuit to each TWT requires greater control. There is also the likelihood of misinterpreted performance requirements due to very unique input signals that the harmonic injection system must amplify with fidelity.
As long as designers review filter theory, mind the infinitesimal delays, watch out for linearity issues, be aware of noise power density calculations, analyze how well the design handles multi-tones and noise and incorporate temperature compensation that includes the TWT, the design should not be plagued by unexpected output power dips and other issues. Despite the challenges of using harmonic injection techniques, dB Control has found that the resulting products are worth the effort and pave the way for a new generation of ultra-wideband MPMs with twice the low-end RF output power of a standalone TWT.