Class-E RF power amplifiers have become quite fashionable in ham radio in the last decade or two.
One of, if not the main contribution to efficiency in a Class-E RF amplifier is the operation of the active device in switching mode where it is either not conducting, or conducting hard (saturated, with very little voltage across it). Both of these are very low dissipation conditions, but in the transition between these states there is significant current and voltage present, the product of which gives significant instantaneous power… so the idea is to make this transition very fast so that the average power is low.
This article discusses effect of slowed switching times on PA efficiency.
Above is a circuit above is from (Sokal 2001) which explains the amplifier and gives guidance on selection of components.
Mosaic published a LTSPICE simulation of a 140W Class-E PA design for 3.5MHz. It appears to be a detailed and valid simulation, so let’s use it for a quick path to exploring the effect of switch rise and fall times. The simulation provides a convenient way of showing currents that can be very hard to measure in practice.
Above is an extract of the simulation schematic.
Above is a plot from the simulation showing:
- green: driver voltage;
- blue: gate voltage (slowed driver voltage);
- magenta: instantaneous FET dissipation (W);
- cyan: drain voltage; and
- red: drain current.
Above is the simulation where the series gate resistor has been increased from 5Ω to 25Ω, causing slower rise and fall time at the gate terminal.
Let’s focus on the waveforms around the transition of OFF state, it is around the middle of the charts.
Whilst the gate turn on wave (green) might seem really slow above, the FET turns on quite rapidly above about 4V, so it is the time from perhaps 4.0V to 4.5V that is critical. The drain current turn off time in the first graph is about 3.5nS, and about 10ns in the second graph.
A result of the longer fall time of the gate voltage (green) is that the transition from ON to OFF is slower, note that it takes longer for drain current (red) to fall, and during this time there is a small but significant voltage impressed on the FET, and the combination gives rise to the instantaneous power curve (magenta).
At the turn ON transition, note the narrow spike in drain current. It is due to the FET discharging C7 (Sokal’s C1). There is energy lost in the FET (see the spike in the magenta curve), but there is also energy lost in the ESR of C1 (i^2t, more so if the dielectric is lossy, eg Class 2 dielectric).
Whilst the slower turn ON results in a lesser current spike, it is longer in duration and the total energy is higher.
The slowed switching speed is likely to increase the duty cycle of the drain current waveform which may degrade performance. In this simulation, it increases from around 42% to around 50% as configured.
A poorly ‘tuned’ Class-E stage may have significantly higher drain voltage at turn ON, and more energy is stored in C7 to be converted to heat within C7 and in the FET… so bad ‘tuning’ also degraded efficiency.
In the figure above from Sokal (using BJT notation), the effect of a mistuned load shows appreciable drain voltage (and capacitor voltage) at the turn ON transition.
Realise that the turn ON and turn OFF times tend to be circuit dependent, and independent of the signal frequency, a quantum of energy is lost as heat on each cycle, so if frequency is increased, the average power lost is increased, and PA efficiency will fall.
Nothing in this discussion is to imply that the conduction angle should be 180°, nor that a conduction angle of 180° defines Class-E.
Measuring real world PAs
It is relatively easy to measure the voltage shown in the simulation results at HF, the current waveform is much harder to measure, and whilst lots of modern DSOs could calculate the v*i product, if you cannot measure i, you cannot compute and display power.
But… we can learn from the simulations:
- the drain voltage waveform is a very good indicator of proper Class-E behavior; and
- gate drive voltage waveform is interesting.
So, why do very few Class-E projects show these quantities?
Links / references
- Sokal, N. Jan 2001. Class-E RF power amplifiers In QEX Jan 2001.