Overmodulation Indicator and Transformer Protector Years ago, dating back before I had an oscilloscope to monitor modulation, I used an 866A as a negative peak overmodulation flasher. Using a 2.5 volt filament transformer insulated for high voltage, the midtap of the filament winding is connected directly to the modulated HV plate lead to the final. The 866A plate is connected to ground. As long as the plate of the final remains at positive potential with respect to ground or at zero volts, the 866 will not conduct. As soon as the final amplifier tube plate is driven negative (the condition with negative peak overmodulation), the 866 conducts and flashes with the characteristic blue glow. I mounted the 866(A) inside a small box with the interior painted flat black, with a small window for viewing the plate area of the tube. The blue flashes were highly visible and this made a very effective overmodulation indicator. As a refinement, I later replaced the 866A with an older version, the original type 866 without the "A" suffix. This version of the tube is rated at only 7500 PIV instead of 10K, but was not problematic with my 2000v to 2500v nominal plate voltage. The advantage of the older version of the tube (if you can find a good one) is that the filament is fully exposed to view, lacking the shield that surrounds the filament in the 866A. This makes the blue flash even more visible to the eye. Because of the lower PIV, I would recommend using only the "A" version at plate voltages above 3000 volts, especially if positive modulation peaks extend very far above 100%. Although the circuit described above works as an overmodulation indicator, it puts a momentary dead short across the modulation transformer whenever the 866(A) conducts. It occurred to me that this could conceivably generate unwanted transients or even damage the transformer. This was corrected by placing a small wirewound power resistor equal to the modulating impedance (PA plate volts ÷ plate current in amps) in series with the 866A, from the rectifier plate to ground. This will maintain the proper constant load on the modulation transformer for the duration of any negative peak that exceeds 100%. With the resistor in series with the diode, the modulation transformer sees normal load impedance throughout the entire audio cycle, even while the final amplifier plate is being driven negative. Since this peak occurs over a very small portion of the audio cycle, a small resistor, 10% of the nominal modulator power or less, is sufficient. For example, a 50-watt resistor would be more than sufficient when modulating a full kilowatt DC input to the final. Besides serving as an overmodulation indicator and maintaining a constant load on the modulator, this circuit offers some protection to the modulation transformer. Normally, whenever a plate modulated final is overmodulated in the negative direction, the modulation transformer operates momentarily without a load for the duration of the overmodulation peak, which could conceivably damage the transformer if an unexpected transient spike happened to occur during this peak while the final amplifier tube plate is negative and not conducting. Unlike other schemes such as "negative cycle loading" and "ultramodulation", this circuit allows protection for the modulation transformer without introducing audio waveform distortion and does not waste a significant amount of audio power in a resistor. Due to inherent characteristics of MV rectifiers, the 866(A) plate must be driven approximately 15 volts positive before the tube conducts, so this circuit doesn't actually flash until slightly over 100% negative modulation. As a further refinement, the plate of the 866A may be biased slightly positive, using a small power supply rated at a few milliamps, to make it flash at or just before 100% modulation. With the 866(A) plate biased approximately 15 volts positive, the flash occurs at exactly 100% modulation. Further increasing this bias causes the 866(A) to flash just before 100%. For example, your final runs 2000 volts on the plate. Bias the 866(A) plate 50 volts positive, and the 866(A) will flash at 97.5% negative modulation. The positive bias may be set to cause the tube flash at considerably less than 100% if desired. However, at percentages between the flasher threshold and 100% negative modulation, the resistor is effectively in parallel with with the class-C modulating impedance, reducing the load on the modulation transformer to half its normal value, much in the same manner as negative cycle loading. This may introduce unwanted distortion and splatter at negative modulation percentages between the flasher threshold and 100%, as the modulator tubes abruptly look into half their plate-to-plate load over a substantial portion of the audio cycle. This may be avoided by increasing the resistor to approximately 10 times the modulating impedance. The 866(A) would still flash and the resistor would maintain some load on the transformer and still offer a degree of protection during negative overmodulation peaks, but the circuit no longer provides a constant load to the modulation transformer throughout the entire audio cycle. See attached schematic.