ALC Exciter Power Overshoot |
Many exciters or transceivers have problems with leading-edge power overshoot. This overshoot can be severe, and result in increased splatter. The power overshoot causes two problems:
Power Meters To Read OvershootNormal overshoot durations are in the range of a few milliseconds and less. Overshoot can be as short as Because of the short duration, many or most peak reading wattmeters (even very good meters) will not indicate the true peak power on overshoot. I ran into this problem using standard Coaxial Dynamics and Bird peak reading power meters. Around 1990 when I was working on a pulsed medical generator, Coaxial Dynamics modified a meter to greatly improve the rising response time and extend the decay or hold time. This corrected problems reading true overshoot levels with my Coaxial Dynamics meter. Other than the modified Coaxial Dynamic meter, one of the very best meters for reading such pulses is the peak reading meter system in Ameritron ATR-30 tuners and the AWM-30 power meter. These meters are incredibly fast on rise time response. They are so fast they will indicate momentary arcs from improper relay sequencing or arcing connections in antennas that do not show on other meters. Be very careful assuming an exciter is clean based on unknown meter response times or pulse power reading characteristics! Meters with extremely fast sample and very long hold times are the best way to read peak power levels. Overshoot can appear within the first 200-300 µS (.0002-3 mS) of initially keying the exciter, and not reappear again until the next signal restart after ALC has fallen back. Oscilloscopes To Read OvershootOscilloscopes are one of the most overrated methods of measuring power levels. While oscilloscopes can be used, their accurate use is probably well beyond the capabilities. First, scopes cannot capture and store the largest very short duration peak over very long duration envelope power monitoring times. To even have a chance at measuring overshoot, it must be a storage type scope. The storage scope has to trigger on a moderate duration (perhaps 50 mS) single sweep. The trigger signal has to be whatever signal is inputted into the exciter, it cannot trigger on exciter RF output. The scope has to be recording waveform as the RF actually rises from zero until after the envelope has peaked and fallen back to a stable value. Second, oscilloscopes measure peak voltage and do so with very poor resolution and accuracy. Power is RMS voltage squared over load resistance. This means we must use the following process: 1.) Look at the display cycle-by-cycle for the first several hundred cycles upon initial activation of transmission after a total ALC drop. 2.) Find the highest peak, and determine the voltage. 3.) Multiply the voltage times .707 to find RMS voltage 4.) Square the voltage (this squares any errors) 5.) Divide the voltage square by the load resistance (now load resistance is a factor) Oscilloscope are a very poor way to read peak RF power, and even more difficult or complex to properly set up to find and record the occasional overshoot peaks. Failures Introduced by OvershootComponent DamageHeating failures are generally not an issue. Temperature rise is a function of thermal lag or "thermal inertia" of the element being heated. Since excessive voltages and current are of very short duration, the most common damage will be arcing or voltage breakdown. This is a major problem with ALC overshoot that affects all amplifiers. Splatter or ClicksSplatter is normally not a major problem. ALC overshoot normally just causes a fast wide-bandwidth click or spit at the start of each ALC rise period. Because the time duration of overshoot is short, and because the problem only repeats as ALC rises from low levels, the average power of splatter or clicks is very low. With most exciters it is more of a problem with slow speed CW or slow voice rates, where ALC is allowed to fall to low levels repeatedly. Splatter can be particularly bad with tetrode amplifiers. Tetrode amplifiers have two grids that are voltage-critical for linearity. If the grid voltage supply circuits have slow voltage recovery and stabilization times, the recovery time will extend duration of splatter or clicks beyond the normally very short initial overshoot duration. This increases energy or average power in wide bandwidth emissions. Most tetrode amplifiers have this problem! They are much less tolerant of very short duration overdrive than grounded grid amplifiers. In summary, the major problem with overshoot is voltage breakdown damage to amplifiers and components following the amplifier. This occurs in all amplifiers, from solid state to vacuum tube triodes. A secondary problem is bandwidth, with this problem mainly occurring or being most disruptive with tetrode amplifiers. Cause Of OvershootALC overshoot, or power overshoot, is caused by the basic flawed design of ALC circuits and RF power control systems. Normal ALC is like closing the barn door after the horse has escaped. The problem is rooted in group delay times as the signal makes its way through the radio to the output port. Signal is generated and applied to the IF amplifier stages. These stages have the gain control circuits that control PA drive level.
When initiated, ALC voltage is zero. The IF is full gain. This applies full unregulated drive power to the PA stages. After RF from PA stages appears at the SWR power detector, power level is detected. The resulting detected voltage is applied to the ALC system. The ALC system compares power detector voltage to a reference voltage from the power adjustment control. If the detected power voltage exceeds the power control reference voltage, after a short response delay time ALC is developed. The ALC system has a long hang time, like using slower AGC in a receiver. ALC is then applied to the IF amplifier where it reduces IF amplifier gain. After 2-10 mS delay as the signal passes through the IF filters and other IF stages, the reduced RF signal level appears at the PA stages. The detected SWR/power voltage is again compared to the reference voltage for any further required adjustments in ALC voltage. This is a continuous analog loop with time delay in the loop. The sum of loop delays are the root cause of ALC overshoot. To test for overshoot, we must capture the rising edge between 0-50 mS after initialization of the transmitter when ALC voltage is near zero, and is first being developed. This will be the leading edge of the first syllable or application of audio tone on SSB, or the leading edge of the first dot or dash with off-on carrier keying (CW). Overshoot can occur anytime within that period, and is often in a very short window of 7 mS or less. Problematic RadiosI first noticed severe ALC overshoot when I purchased a new ICOM IC775 DSP radio. I had an 8877 amplifier with fast grid protection and very heavy components, and used an antenna tuner capable of handling about 3kW or so on the worse bands. When I tuned the amplifier for 1500 watts output, it would immediately fault as a grid overload and excessive drive fault upon release and reactivation of drive. When I advanced loading to a setting where grid and drive power fault protection did not activate, my antenna tuner would arc over! I connected a fast pulse measuring meter, normally used for measuring medical pulse generator peak power. With my IC-775DSP set for 50 watts output, it produced a peak power of over 200 watts when ALC rose from zero at the start of transmissions. This triggered the grid current fault and drive power fault in my amplifier. When I advanced the amplifier loading to handle the 200+ watts leading edge drive pulse, peak amplifier output power was over 5 kW. This caused my antenna tuner to break down. Once the arc started, it continued even after drive pulled back to normal levels. This is because plasma in the arc and ionized air in the variable capacitors reduced voltage breakdown. My second experience was with an IC706. I measured my IC706 as having about 150 watts of overshoot, the exact value dependent on supply voltage and band. Sometime in the early to mid 90's Eimac asked me to referee a problem with replacement tubes in a Ten Tec 3CX800 amplifier. They were having repetitious problems with short tube life in the amplifier when driven by an IC775DSP. My conclusion was two things contributed to the premature failures of 3CX800A7 tubes. The radio had severe overshoot, and the amplifier lacked control grid overload protection. Since that time I've measured overshoot in some Yaesu radios, as well as Kenwood and Ten Tec radios. Good RadiosThe FT1000/ 1000D radios (NOT the FT1000MP series) are very clean if drive power is reduced to bring very little ALC on scale. The FT1000MP Mark-V can be corrected also if TX gain is reduced. Average power can be brought up by using speech processing, rather than depending on ALC. My K3 is very clean, although average power is not pushed up normal amounts by ALC in the K3. This can be corrected through use of the speech compressor. My older IC751A's are also reasonably clean. It is generally far better to run processing than excessive ALC. People depending on high levels of ALC to increase average power are being foolish. |