While there may be very rare exceptions to this, nearly 100% of the time neutralization of an HF PA or power amplifier vacuum tube has virtually nothing to do with VHF or UHF oscillations. VHF oscillations are almost always caused by a high impedance (parallel resonant) path from a grid to ground. The high impedance prevents the grid from being "clamped" or held at ground potential for RF at some frequency or range of frequencies. If this high impedance resonance happens to occur at a frequency range where the anode path to ground is parallel resonant, the tube can act like a tuned-plate tuned-grid oscillator. 

Cathode Driven Power Amplifier

Many people think grounded grid HF PA's do not need neutralization. In many cases this is true, but in some cases it is not true. Tubes with low impedance grid structures and an internal shield like the 8877 have very little feedthrough capacitance. This is true all the way to UHF.

Tubes like the 3CX1200A7 or D7 have significant feedthrough capacitance, and exhibit "out of neutralization" effects on bands like fifteen or ten meters.

Tubes generally not requiring neutralization in GG HF amps are the:

8877/3CX1500A7   8873 8874 8875  3-500Z  3CX800A7  3CX1200Z7  3CX3000 series 3CX5000 series 3CX10000 series

Tubes generally benefiting from neutralization in HF GG amps are the:

810  811A  572B   304TH  3CX1200A7 and 3CX1200D7, 833

Additionally tetrodes are generally have very low feedback when the grids operate at RF ground potential, but tying a screen or control grid to the cathode can cause unwanted feedthrough that requires compensation through neutralization. Some amplifiers like the Amp Supply LA1000 or Dentron sweep tube amps were unstable on ten meters because the control grid was tied back to the cathode.      

Note that tubes not requiring neutralization in GG circuits are generally those with conical grid supports and grid connections made via a very short wide exit lead or leads. Stable tubes with low internal feedback often have compact control grid structures inside the tube.

Tubes benefiting from neutralization are those with long thin single grid leads to single pins, wide spaced grid wires, and poor or no internal shielding from anode to cathode.

Tubes with better internal construction, shorter wider leads, more compact grid structures, and closer spacings not only work better at higher frequencies, they also are significantly more stable for unwanted parasitics and they rarely require neutralization or parasitic suppression!

How Do We Neutralize a Grounded Grid Amplifier?

In the circuit above, T1 inverts phase 180 degrees. Cneu approximately equals Ckp, the cathode plate capacitance (or feedthrough capacitance) of the tube. Unwanted feedthrough capacitance, Ckp, varies widely with frequency. This capacitance is not frequency linear. It has less reactance at higher frequencies, and higher reactance at lower frequencies. The absolute equivalent value of Ckp varies more than a pure capacitor would with frequency because all stray inductances, including Lint (internal lead inductance) and Lext (external lead inductance), cause Ckp to have a reactance vs. frequency slope much more rapid than a normal fixed capacitor. This means we can really only neutralize a PA perfectly over a small range of frequencies.

In the Ameritron 811H amplifier, the neutralization is almost perfect on 15 and ten meters. It still does a good job from 7 to 45 MHz. Below 20 meters feedthrough capacitance is so low the lack of optimal neutralization doesn't matter, above 45 MHz parasitic suppressors load the circuit enough to greatly decrease gain and stabilize the stage.

If you remove the antenna or exciter from an AL 811H, key the PA without drive, and rotate the controls it will be perfectly stable and not break into oscillation on any band. If you make the same test with a Clipperton L, Yaesu FL2100, or a Collins 30L1 (all un-neutralized amplifiers) you will see most of the amplifiers (if not all) will break into oscillation on 15 and 10 meters.

This effect is common in all these amplifiers because 811 and 572 tubes have similar construction, with very poor shielding inside the tube from anode to cathode. Both tube types exhibit very high amounts of feedthrough capacitance, enough feedthrough capacitance to make amplifiers unstable on the operating frequency on higher bands like 15 and 10 meters.

Grid Driven  

Grid driven tetrodes like 6146, 807, or 4CX250's have high power gain. High gain systems require very little feedback to become unstable, so they generally neutralization. The following circuit shows a commonly used tetrode grid-driven amplifier with neutralization:

 

L1/C1 is the normal input tuning coil. Being resonant on the operating frequency, it inverts phase 180-degrees from end-to-end. C2 is a voltage divider to control the feedback voltage ratio and provide a return path for grid excitation. Cneut is adjusted so its voltage feedback equals the voltage fed through Cgp from plate to control grid inside the tube.

Note that this system depends heavily on L1/C1 being resonant at the operating frequency. This proves the tube is only neutralized at the frequency where C1/L1 is set. It does not stabilize the tube on any frequency except where L1/C1 is resonant.

Lp,Lsc,Lk, and Lg are inductances of leads inside the tube. Lp1,Lg1,Lk1, and Lsc1 are lead and component inductances that occur outside the tube.

While the feedback adjustment setting of Cneut holds true for multiple bands near the initial adjustment frequency, it only actually neutralizes the tube on the band in use at any moment of time!

In a 160-10 meter PA, Cneut generally only works properly over two or three bands. It is usually set near 15 meters so it has the most effect where it is needed most. By the time we get down to 40 meters and lower, feedback voltage through Cgp is generally through such a high reactance that the lack of proper balancing is meaningless.

Neutralization

Neutralization generally only affects operation near or at the desired operating frequencies. Neutralization is normally optimized near the upper frequency end of operation, perhaps between 15 and 30 MHz in a 1.8-30 MHz transmitter or amplifier.

Neutralization is sometimes needed because tubes have unwanted internal capacitances. The capacitance between the output element and the input element inside the tube will cause the output circuit to couple back to the input. If large enough, this regenerative feedback could cause a loss of efficiency. It might cause the output maximum to occur off the plate current dip, reducing efficiency. It might increase IM distortion or in rare severe cases may cause the amplifier to oscillate someplace the operating frequency. (This problem is common with grounded grid amplifiers using 572B's like the Dentron Clipperton L,  or quads of 811A's, like the Collins 30L1. Yaesu has this problem is some FL2100's.)

While a need to neutralize does occur in some HF grounded grid amplifiers, it is more common in very high gain grid-driven amplifiers.

Neutralization Adjustment Methods

Neutralization is generally accomplished by adding an external capacitance that is excited exactly 180 degrees out-of-phase with the feedthrough capacitance. One typical adjust procedure is to disable the PA stage by removing screen or filament voltage. A sensitive RF detector is connected to the transmitter output.

Normal drive is applied, and the neutralizing capacitor is adjusted until feedthrough power is minimum. The tuning controls are continually peaked for maximum power on the sensitive detector throughout the process.

A second less accurate method is to watch the plate current dip in a properly tuned normally operating transmitter. The neutralization capacitor is adjusted until maximum power output and minimum plate current occur simultaneously as the plate capacitor is tuned.

The best method  varies with the PA design, but in general the most accurate method is by applying drive to a cold PA stage (generally either screen or filament power is removed) and feedthrough power is measured with a sensitive detector.

What Happens If We Don't Neutralize a New Tube?

Many times nothing noticeable occurs if we don't neutralize a PA. The results really depend on how much different the internal capacitance is in the new tube(s) when compared to the capacitance of the tube(s) being replaced.

If the PA requires neutralization and we don't neutralize or re-neutralize it, we could find IM distortion higher. We would probably find maximum output power occurs well-off the plate current dip. The un-neutralized stage, in severe cases, might oscillate somewhere near the operating frequency under certain conditions of tuning and loading.

Neutralization is generally only accurate over a limited range of frequencies, but fortunately it is almost always at the higher frequency end of the operating range where the PA needs neutralized. The manufacturer probably knows what the optimum adjustment point is.

Unrelated Problems are sometimes Blamed on Neutralization

Since neutralization is the canceling of feedthrough capacitance, and since that doesn't change over the life of a tube (or even much from a hot tube to a cold tube), neutralization won't "drift out" with certain tube types.

Any tube will either neutralize right from the start, or it won't. If it appears to drift out of adjustment something other than the neutralization is at fault. Tubes in HF PA's cannot drift in and out of neutralization because the capacitance is set by the tube's physical construction... not by emission, age, or any other time-variable parameter.

People sometimes blame neutralization for problems when they really have gassy or defective emission. Gassy tubes can go into current runaway or even flash over inside. Doing this for 35 years for a living, I've never yet seen a tube in a HF or lower VHF amplifier  "drift" or age out of neutralization.

The capacitance is for the most part related only to the physical characteristics of the tube, like internal lead length, size of the elements, and spacing of the elements. That's why it is perfectly acceptable to neutralize a cold tube (no filament voltage). The change in feedthrough is very small when the tube is operating compared to when it is cold.