Loading and Tuning linear amplifier

Related pages:

Vacuum tube amplifiers how a PA Converts DC to RF

Arcing in amplifiers

This page might be a bit long, but it explains how and why you need to adjust the amplifier in a certain way. If you find it confusing and have suggestions to improve it, please email me at this address:

Tuning Worries

Destructive things sometimes occur when amplifiers are mistuned or improperly operated. There are two overall damage mechanisms,  excessive temperature (heat) and high voltage arcing.

Excessive temperature permanently damages components. Excessive heat is the most common source of tube damage. Heat is always a time function of dissipated power. Heat damage depends on the thermal mass of the thing being heated, the initial temperature, how long heat is applied, and how much heat is being applied. Heat damage always relates to temperature rise, which is a function of heating power power over time. Time is determined by the ability of things to absorb (thermal mass) or get rid of heat (dissipate heat). Heat damage always takes some finite time, even if it is a relatively short time for small objects with low thermal mass. Heat damage is sometimes cumulative with repeated small overloads spread over time. Heat damage is generally not cumulative until a certain temperature threshold is reached, overloads below a certain temperature will not cause life-shortening in components.

For example, vacuum tube life is virtually the same with repeated severe anode dissipation rating overloads so long as the anode does not exceed a certain temperature limit. The same is true for many FET's. This is why tubes and FET's can be operated far beyond published dissipation limits with no apparent life reduction. As a matter of fact, some vacuum tubes actually have a pronounced reduction of life when operated at low temperatures! The filament of a 3CX1500A7/8877 and the anode of a 3-500Z are two examples where low temperature operation accelerates device failure. Carbon resistors, on the other hand, are damaged by small thermal overloads over long accumulated periods of time. This is why carbon resistors near hot running tubes, or carbon resistor high dissipation loads, age down in resistance over time.

Excessive voltage breaks down components. Arcing can destroy things in small fractions of a second! Arcing comes from too much voltage. Arcing causes instantaneous failures by punching a hole through insulation, very rapid failures by surface damage (such as carbon tracking a surface), or fairly quick damage by concentrating heat like a welder does. Arcing can ruin insulation or damage surfaces. 

For example, FET's, band switches, and tuning capacitors are most frequently damaged by arcing. Switch contacts can be ruined in fractions of a second, FET's almost instantaneously. Tuning capacitors, because of larger mass at arc points, are slower to damage but still might fail in a few seconds. 

Arcing is generally an instantaneous or rapid failure mode, while heat dissipation damage requires time. Arcing is from excessive voltage, and heating from excessive current. Because of the difference in cause, these two very different failure modes do not normally occur under the same operating conditions.

A Word about Tuning Pulser Systems

Tuning pulsers, or if we are into locker-room innuendo "peckers", reduce average dissipation. This means tuning pulsers or "peckers" can reduce heat damage in physically larger objects, such as vacuum tube anodes. Contrary to some claims, tuning pulsers or peckers do not lessen the chances of most arcing or voltage failures, unless the arc failure is a product of long-term heating or excessive average dissipation. Claims a pulser prevents arcs or results in optimum tuning because it "emulates speech" are just not true. Peak voltage is not directly tied to average power, so a reduction in average power will not reduce peak voltage.

By definition, peak power and peak voltage has to be the same or we cannot properly load and tune an amplifier! Tuning always has to be done at full peak power, and that means we have full peak voltage available. 

What surprises me more than anything is how or why so many fell into a trap, thinking how one pulsed single-tone (into an audio jack) emulates multiple-tone voice, while another pulsed single-tone (via the CW key jack) cannot emulate voice. There certainly are cases where otherwise knowledgeable people do not understand the difference between audio tones converted to RF in a SSB transmitter and CW signals. A single audio tone into a SSB transmitter produces exactly the same RF output spectrum and characteristics as a CW carrier, RTTY carrier, or unmodulated AM or FM carrier.   

Also there seems to be a lack of understanding of failures and stresses. The only thing pulsing does, for a fixed peak power level, is reduce heat. Pulsing will not reduce arcing, although pulsing can make the arc occur in bursts, reducing arc heating of larger metal surfaces.

Audio pulsers increase bandwidth of the tune signal, and offset the actual RF frequency from the dial setting by the pitch of the audio injection tone. This might annoy operators on adjacent SSB channels, although all over-the-air tuning can be annoying to others. Signal bandwidth of an audio tone pulser is determined by the actual RF frequency and any audio harmonics, plus shape and time of rising-and-falling tone envelope edges. Audio pulse bandwidth is limited by SSB filter and audio stage bandwidth.

CW pulse bandwidth is determined by CW keying system rise and fall characteristics. CW bandwidth is free from audio tone harmonics and generally much narrower than SSB. CW pulsed tuning can be just as effective as a single-tone SSB pulse, but because of slower rise and fall and elimination of harmonic distortion, is almost always less disturbing off-frequency.

Final tuning results are identical for CW pulses (ditter) or SSB pulses (pecker), while both have identical potential problems. Potential problems include meter response, making sure full peak exciter power is produced in the tuning process, and ensuring the amplifier is not saturating or non-linear. Neither method produces better end results, since both methods are equally critical for pulse shape, pulse level, and pulse rate.

Pulse tuning is not always a final step. No matter which pulse generation system is used, steady-carrier grid current should be checked as a final step at maximum drive. This can even be as crude as a "whistle" on SSB.

Both pulse creation methods, when applied correctly, provide exactly the same results. Overall, except for bandwidth, the two systems are generally identical in results. Neither has a consistent advantage in tuning results. Depending on meter response and equipment, some system will require 100% duty carrier, while other systems might get away with few percent duty cycle. Any pulse duration and injection method will produce identical results providing transmitter response limits are not reached, and if meters indicate RF pulse peaks correctly.

Pulser Duty and Repetition Rate

Any pulser system has two critical adjustment parameters. The critical characteristics are pulse repetition rate and pulse duty cycle. Optimum duty cycle and pulse repetition rates are defined by system response to pulses. Optimum duty cycle is NOT defined by some imagined "voice duty cycle" emulation. A single-frequency pulsed audio tone is no different than a pulsed dot on CW, for the same pulse (dot) repetition rate and pulse (dot) duration.

Emulating speech requires at least three test tones of syllabic, lower voice tones, and upper voice tones, and is really only especially useful if we look at output on an expensive and complicated spectrum analysis or frequency domain device. Speech emulation is a lab-type performance proof procedure, not a tuning aid. Two-tone tests are most commonly used, but do not load the system at a syllabic or speech-pause rate. Two-tone tests, and even notched noise tests, fail to show many power supply and bias regulation problems. 

An audio injected pulsed single tone, other than rise-and-fall rates which create "click" sidebands and wide bandwidth) and harmonic distortion, is not any different than just running a keyer on dots. Once any pulsed system is inside the bandwidth of the modulation system and meters, results are identical. Staying within limits of power detector meter response and the modulation system is critical, as is making sure the exciter reaches full peak power long enough for accurate meter readings.

An ideal pulser would allow adjustment of pulse rise and fall times, pulse repetition rate, and pulse duty cycle. It would always force output level to the absolute maximum PEP power from the transmitter on CW or SSB (whichever is highest). By allowing adjustment of pulse rise and fall, pulse duration, and pulse rate, the pulser could be set within metering and ALC system limits.

It cannot be stressed enough, peak power from a pulser should always be slightly greater than maximum PEP power ever expected on SSB (or CW). Any critical meters would need to fully respond to the peaks.

other than heat, the most reliable tuning method is a steady carrier. With a steady carrier, meter response is no longer an issue. Meters will always work. This is why we should always do a final confirmation check with a short carrier at full peak power.  Remember, peak envelope power equals average power with a carrier.
 

Basic Operational Theory (this section is optional reading SKIPDOWN)

The output device in your amplifier has a certain optimum available voltage swing, and has limited current available. It is important that load impedance presented to the output device matches the optimum values of available RF voltage and current from that device. When we adjust the tank circuits (or auto-tuners) in our power amplifiers (PA), we are really setting or adjusting the load impedance presented to the output device. Here's what happens when we tune:

1.) If load impedance presented to the output device is too low, current is excessive and efficiency suffers. This is also called over-coupling. This causes too much heat. Heat is a long term problem that takes a finite time to cause damage. It is generally NOT instantaneous damage, although tube anodes or transistor junctions can be overheated to the point of damage in a matter of 15-30 seconds in some cases. This is the case where we use too little loading capacitance.

One good thing about over-coupling is screen or control grid current is reduced, and this protects the most sensitive and easily damaged parts of vacuum tubes. Another advantage in tube amplifiers is linearity generally is a bit better with slight over-coupling. There is slightly less splatter or distortion.

2.) If load impedance presented to the output device is too high,  current is reduced but voltage will be too high. This is called under-coupling. This is the case where we use too much loading capacitance. Efficiency is normally very good, heat is reduced or remains in a normal range for the level of output power produced. Voltage can increase well above the supply voltage limits, up to several times the dc supply voltage in extreme cases. This is the worse scenario because severe damage can be instantly caused by arcing or voltage breakdown of components, and damage can be instantaneous even with very slight over-voltage. Worse yet, once an arc starts, it causes a dielectric failure. The dielectric failure destroys insulation, creates sharp points or surface irregularities that reduce voltage breakdown, or the arc ionizes air or creates a plasma. All of this works to sustain the arc even after voltage is reduced to safe levels.

Under-coupling, or having the loading capacitor closed too far for the load impedance and/or drive power, increases grid current and splatter. It creates a very hard form of non-linearity where the device switches into non-linearity very quickly, and the sharp transition into non-linearity or gain reduction creates a very wide bandwidth splatter.

If we have a coupling error we would like it to be slight over-coupling in the PA output device. It is better to see a little too much device plate, drain, or collector current than too much voltage at reduced supply current. We also do not want excessive grid current in vacuum tubes.

For this reason, almost all "pre-tuned" solid state amplifiers are over-coupled to the load. They are actually optimized for a higher than normal load impedance by slightly over-coupling the output devices to the load.

Improper and Proper Loading of Amplifier (read this section)

There is very little difference between excessive drive power, antenna system faults or failures, or grossly improper adjustment of loading. All can be equally bad.

Improper tank adjustment, antenna system failures, and excessive drive are equally harmful to component life. Improper tank adjustment, antenna system failures, and excessive drive either create splatter (and in extreme cases cause keyclicks) on adjacent frequencies, or they cause excessive heat in the output devices or components in the system. Regardless of the reason for them, amplifiers are damaged by excessive tank voltages or device currents caused by improper adjustments that prevent proper energy transfer to a load.

In some cases, particularly on the lower end of the lowest frequency bands, proper loading cannot be achieved.

Signs of UNDER-coupling

When the output capacitor (load capacitor) is meshed too far (too much capacitance), especially at high drive power levels, the amplifier will be under-coupled. Under-coupling is the very worse thing to do to any amplifier because failures can occur in a matter of seconds! There are several signs of under-coupling in a grid-driven tetrode or grounded-grid amplifier. Watch closely for the following:

1.) When the drive power, using a steady carrier, is slowly increased the grid current (either screen or control grid) will at some drive level suddenly rapidly increase. The sudden rapid grid current increase will be disproportionate to the plate current or drive power increase! DO NOT go past the point where grid current starts to rapidly increase with small changes in drive power level.

2.) Too much grid current, either screen or control grid, is a clear sign you have the loading control too far meshed or closed.

In a grounded-grid amplifier or a grid driven tetrode amplifier, the grid current meter (control grid in the triode, screen grid in the tetrode) is the most reliable indicator of improper loading and/or tuning. Be especially watchful of disproportionately high grid currents compared to anode currents or drive power, or a rapid increase in grid current with a modest increase in drive power.

Never tune, peak, or dip the amplifier at reduced drive power, and then attempt to operate or attempt to suddenly apply full drive! If you are going to make a mistake, make the mistake by having the loading control too far open or unmeshed...not too far closed or meshed! At least with the loading control too far open, you will not cause an arc, blow out a bandswitch, or damage a tube grid. You have slightly more time for mistakes and corrections when the loading capacitor is open too far than too far closed.

Most Common Tuning Error 

Too much grid current is almost always a sign of a loading control that is meshed or closed too far for the amount of drive power. This is hard to see on SSB, and best to view on CW.

The most common amplifier tuning or loading error is adjusting an amplifier at low or reduced drive power as a last amplifier tuning step. When we load a radio or amplifier at reduced drive as a last tuning step, we establish that power level as the absolute ceiling for drive and output power. Final loading at reduced drive results in a loading control too-far meshed. This can cause arcing, splatter, and excessive grid current.  

Ideally (if possible) we should make the final tuning and loading adjustments at or near maximum exciter drive power. Some amplifiers drive too easy to do this, so we should always pay attention to factory instructions and avoid exceeding factory amplifier tuning current limits, especially for control and screen grids. Grid current is especially important to watch because grids often do not have sufficient thermal mass to absorb large overloads even for short time periods. Excessive grid current in metal oxide cathode tubes (ceramic tubes with indirectly heated filaments) like the 8877 and 3CX800A7 can damage tubes in less than a few seconds; whereas most anodes will tolerate severe overloads for 15 seconds and longer. It is better to let the large anode or plate in a tube take the brunt of any mistuning heat, which means with any mistake it will be better to over-couple or have the load control capacitance slightly lower than optimum.

The last few tuning steps should always be:

• Load the amplifier to maximum obtainable output at full exciter drive (without exceeding amplifier short term overload ratings)
• After that, advance the loading control very slightly beyond that point (towards less capacitance). 

ALWAYS load your amplifier for maximum obtainable power, and reduce drive to rated, safe, or desired operating power levels!  This ensures minimum voltage and current in the tank and maximum possible linearity (best signal quality). High grid current is a strong indicator of excessively light loading in grounded grid amplifiers.

Voltage Sag

Voltage sag, unless accompanied by significant conduction angle changes, does not affect loading setting. Voltage sag will not cause mistuning.

Voltage sag does not cause mistuning because voltage and current decrease at about the same rate. While sag does reduce power, it does not normally affect optimum tuning position. Even drastic changes in voltage, such as going from CW operating voltage to SSB voltage in a Heath SB220, has only a slight effect on optimum tuning point. If properly loaded on CW at maximum available drive, the amplifier will remain acceptably tuned at SSB voltages.   

Exciter Transients or Power Overshoot

Maximum available carrier drive might not result in sufficient drive for tuning. This is especially true when an exciter has transients or power overshoot from marginal ALC response.

Transients or overshoot appear on the leading edge of the RF envelope, on the leading edge of speech or CW transmissions. This is the time when the transmitter is going from zero power towards full power. Since the ALC circuit has no stored voltage at this moment, the exciter runs full throttle for an instant. This effect is missed by most power meters.

Once the ALC comes up, the hang time of the ALC will hold the exciter gain back. Transients and/or overshoot will generally disappear.

Transients and overshoot, being of short duration and infrequently occurring, make it impossible to tune correctly at maximum drive. With transients or ALC overshoot, it is impossible to tune your amplifier properly by simply tuning for maximum output with a carrier, a tuning-pulser, a whistle, or normal speech. We cannot just tune for maximum output and expect the amplifier to be properly loaded when the exciter has leading edge ALC transients!

Let's assume the exciter is rated to deliver 100 watts, but has momentary peaks or transients of 160 watts while the ALC or power control loop "takes hold". Power surges of 160 watts, too short to register on normal power meters, occur at the start of every transmission.  Of course, if we don't run the exciter wide open and reduce power to 50 watts the problem actually gets worse! In this example the transient peak would still reach nearly to the same 160 watts, but the amplifier would be tuned for 50 watts drive! This is bad news for splatter and for components in the amplifier.   

This is why the maximum power setting of the exciter should generally be used while tuning. If the exciter has far too much drive for the amplifier, we need an attenuator or an amplifier better matched to the exciter.

The loading control should always be advanced a reasonable amount beyond (further open) the actual maximum output power setting. This will allow the amplifier tank system to handle transients without arcing or component failure.

Easy-to-Drive Linear Amplifiers

Some hobbyists and manufacturers tout "very low drive" as an advantage, claiming it offers "cleaner signals". Nothing is further from the truth. 

Exciters almost always provide the best IM performance when operated at a time-averaged peak power a reasonable amount below full output, rather than very low levels. At low power levels, exciter performance is dominated by cross-over distortion. This is where bias non-linearity or device input threshold induces distortion. The ALC system also adds cutoff bias to early stages. This bias increases distortion in ALC controlled stages. At very high levels, gain compression or negative bias shift becomes an issue. Exciters typically do best when operated in the area of 60-80% of rated power.

Worse yet, low drive amplifiers are especially susceptible to damage from exciter overshoot or transient problems. Transients and overshoot peak power remains almost the same level regardless of exciter power control settings. As exciter operating power levels are reduced, the percent of power overshoot becomes worse.      

The most undesirable situations are those where exciter power greatly exceeds (by more than twice) an amplifier's normal drive power limit. Not only does this reduce system IM performance, amplifier drive transients are aggravated. Amplifiers should be designed or selected to match the exciter's maximum power output, or an external attenuator used to bring the amplifier's drive requirement up to the exciter's full power level. Low drive amplifiers are, as a general rule, bad news.

Amplifiers Without Enough Loading Capacitance

Some amplifiers do not have enough loading capacitance. The loading or antenna coupling control is all the way at maximum (capacitor fully meshed) for maximum output power, making it impossible to "peak" the output. Opening the loading capacitor up more just reduces the output power, no matter what the drive level. This is over-coupling that cannot be corrected. It can be caused by several things:

1.) The loading capacitance is inadequate through bad or improper design. This is common on the lower end of the lower bands in some amplifiers. For example the Kenwood TL-922 (which works better in the old Japanese segment of 160 meters, above 1900 kHz). Another amplifier that had poor tuning range on 160 and 80 meters was the original AL80, the Dentron Clipperton, and several Amp Supply amplifiers.

2.) The output power level you are tuning at is lower than the design target. As power is decreased, the maximum-power-output loading capacitance setting always increases. In other words as drive is reduced and we re-tune, the output power "peaks" with more and more loading capacitance.

3.) A padding capacitor has opened up.

4.) A tank inductor has shorted between turns or does not have enough turns. (Common in Dentron amplifiers on lower bands, where loaded Q is often 20 or more.)

5.) Antenna system impedance at the amplifier is too low, or is slightly inductive rather than being resistive or capacitive.

since July 2004