I realize this topic has been dead for over two years.
I have one of (maybe two) full sets of the HR tubes made for the 9500HR, the big brother of the 8500HR.
No 9500HR was ever actually built but Tse did evaluate the tubes. They are proportionally superior to the 8" tubes, just as you would expect. Same spot size, more tube surface area to play with.
I bought this evaluation set from VDC for what I consider to be a very fair price. They are in LC chamber assemblies.
I have chosen to try to absolutely maximize their potential performance by using them in conjunction with the superbly sharp focus yokes found in a Barco 909/Cine 9. Those yokes require positioning on the tube neck that is far away from where the regular stock Thomson focus yoke would be, which is requiring me to at least attempt to modify the shorter Barco deflection yokes to run in a Marquee. Which by itself is a bit of a challenge, as I'm having to totally rework all the wiring on the deflection yoke and then have to analyze the resonant circuit values of the HDM and probably will have to modify the HDM and VDM to restore resonance to the deflection circuits.
I have (briefly) run the HR tubes with the regular Thomson magnetic stack and yes, they are sharp indeed. But I think I can make them sharper yet.
However they don't even have dag ground leads so I didn't run them for long at all. You would be stunned to find out how much voltage builds up on the ungrounded CRT bell and how quickly it builds up without ground straps! Enough to draw a 1/8" arc when shorting it to ground, in five seconds of operation.
To add to the madness, I just bought, from a surplus reseller, 13 of the "new" Tse designed neck cards, new in the boxes, still in anti-stat bags with unbroken seal stickers. They're the "retrofit" version that can be mixed and matched with the legacy cards.
The goal: If not 4K via HTPC, then at least find out what the limits are with the HR tubes and better magnetics. I predict that the GT17 lenses will reach their 12 line pairs/millimeter resolution limit at a resolution of about 3048x1714 pixels assuming that the pixels are "square", with 1000 pixels measuring the same length on the CRT face whether measured horizontally or vertically. This is a 16x9 aspect ratio raster.
In full frame, the lens resolution limit would be equivalent to 3048x2438, call it 3000x2400, near enough.
This assumes a 5" wide raster, which doesn't even come close to being a maximized raster. A 9" tube's raster
can be stretched to almost 6" wide before getting into edge of glass distortion. So...actually with rasters pushed to the ultimate limits, the lenses would permit a resolution of up to 3658x2743 (4:3) or 3658x2047 in 16:9.
Consumer 4K is 3840x1920 so from an optical perspective, we're within spitting distance of 4K, and allegedly the later Elcan manufactured HD10L lenses are "significantly sharper" than any lenses made by 3M/USPL.
So maybe they're 4K capable lenses.
I'm not going to bore you to tears with further detailed mathematical analysis of tube resolution, but I do make note of the fact that around the time Scott was doing the 8500HR evaluation and posting about it, I was able to duplicate his 2048x2048 resolution results using a green 180DMB (Or maybe it was a P16LNN, honestly I can't remember which) tube with the first "frankenyoke", the KF2211, on my Marquee 8000 testbed chassis. Though I could not resolve the pattern on the screen due to the severe limitations of the HD-8B lens, the full pattern was quite sharp on the CRT face. The improvement vs. the stock Thomson focus yoke was sufficiently dramatic that if the improvements to be made using the Barco focus yoke are comparable to that old experiment's results,
then 4K will indeed be close enough to be worth fighting for. However of course there will be bandwidth limitations that may keep true 4K forever out of reach.
That's OK, because I'll settle for 3K, which I judge to be VERY possible. 3K at 72 Hz, no less. That'll fit in 258 MHz of video bandwidth.
Come to think of it, I'm kind of confused how the neck card on my 8000 test bed was able to deliver 2048x2048 with good definition.
I know it wasn't a modified card. Maybe the refresh rate was low. I'd have to power up the same signal generator and check that pattern's parameters to get that answer. (I can do that. I used a Sencore CM2125 and today I still have one of those.)