I did not say each gun handled 1.5mA. In the test, 100% white (NTSC 100%) means all three guns are in use and a specific white type of color temperature of is produced during the video portion of each line. It is during this operation that the total beam current is measured (less blanked time and those lines reserved for alternate uses). So, The currents of each gun (which are slightly different) are added to arrive at the whole.
26.5KV IIRC, x 0.0015A = 39.75W, so it is inline with your numbers for the projection CRT and with what an average color CRT of that type would dissipate. Additionally, the shadow mask carries heat to the envelope, and those CRT faces run nice and cool having the heat spead around the much larger faceplate section, unlike a faceplate of a single-phosphor projection CRT with no mask, necessary high current gun and liquid cooled screen.
Not to derail from the OP's high voltage amplifiers, I know the topic is on those of 6500V output.
- So ingnore this side comment about some weird experiment with lower voltage amps +/- 300V.
Experiments with unmodified TV set yokes pressed into vector usage have led to amplifier requirement of +/-300V and +/-3A drive to the yoke, to obtain undistorted DC-60KHz @-3dB bandwidth, which is untenable for an inexpensive deflection amplifier. There being few to no designs to study, I've destroyed a few lash-ups before stopping the waste of money on transistors. Any little thing, they will blow up very nicely.
Since, I've been experimenting with yoke rewinds on a couple of old 10" round monochrome CRTs to allow XY (vector) display for scopeclock, asteroids and other silly purposes. Rewinds and partial (1-winding) rewinds seem to be the way.
There's a nice and recent youtube video on rewinding a color TV's yoke for use with a common vector-type video game and wells-gardnet-type application. He left one winding as-is (the low inductance one I'll bet). The trick part is keeping the the purity (for color games like Tempest, space duel, etc), and the way seems to be using the yoke core (same yoke part Nr. or application) that came with the CRT. With monochrome it's not as important. I hope there is a follow up since he had not yet re-installed the ring magnets on the CRT and demonstrated a test pattern, only shown the color vector game up and running.
In this general vector display vein, I'm looking at using a small old projection CRT (VPH-1050 style) for a RF modulation monitor that would at least be bright enough to be easily seen across the room in high ambient lighting. Lower voltages as well as much lower percentage of attainable screen brightness apply for direct observation.
While it's true that a number of Tektronix CRTs are capable of high enough brightness and can be adapted to the purpose, they are not as common because people tend to retain the old Tek scpes, and therefore may be in higher demand and cost than EM deflection tubes from such old projectors, from which full brightness will never be required.
In any case, it is much better to keep the original yoke core and distribute the (fewer and heavier) new windings as per original, to preserve as much of the geometry quality as possible. As for the RF, which certainly can't drive a yoke, the idea is either to detect the envelope, use that to modulate an oscillator to 'fill' the envelope with a waveform the yoke can reproduce, - or - to downsample the RF waveform as described in a QST article, though that is more complicated due to lower EM deflection bandwidth -vs- deflection plate bandwidth.
Alas nothing impossible is easy, and nothing expensive is free, but I'm having a great deal of fun pushing the boundaries.
The ones found in vector video games run on +/- 25-30VDC, so it's the yoke all the way!