But last week, I found another Trinitron, this time a Dell P1130, that someone was giving away for free! Now with two, I figured I should adjust both of the monitors the proper way.
The best resources were the two mega-threads at HardOCP (hi spacediver!) and Icrontic:
http://hardforum.com/showthread.php?t=952788
http://icrontic.com/discussion/19549/de ... too-bright
But I also used:
http://www.piclist.com/images/com/geoci ... ua/windas/
http://dor-lomin.com/images/forums/hardocp/windas-conv/
The only hardware needed is a USB to TTL adapter and a USB extension cable. I used the Kootek PL2303 because it was cheap on Amazon, but I'm not sure I can recommend it (I'll explain why later).

The modified WinDAS worked fine (mostly) on Windows 7:
http://www.mediafire.com/download/yyt5p ... -+Copy.zip
With that out of the way, I started with the Dell just in case I broke something. The P1130 is actually the newer of the two monitors, manufactured in 2003. The IBM was built 2000, but I know it works.
Taking the case off of the Dell requires removing two screws and flipping two plastic tabs on the sides of the monitor. I couldn't get it off at first, so I thought there were screws hidden underneath the back sticker, but I just wasn't giving it enough force.

The ECS port is easily accessible:

From left to right, that's TXD, RXD, +5V, and GND.
When connecting the adapter, the TXD and RXD wires need to be reversed. The following picture has them connected incorrectly:

When I first connected the Kootek to my computer, Windows installed the drivers automatically, and it showed up in Device Manager as:

In WinDAS, I selected the model (F2) DELL P1130, and set the configuration (F3) to the COM port above and SG Name to MANUAL.
At this point, WinDAS could finally see the monitor, so I selected Save Data to File, the monitor turned off, the progress bar moved a bit, and then it failed:
ECS2 Error Code :FFFFF001
Check RS232C Line
This would end up happening a lot. I turned the monitor back on, disconnected the adapter, closed and reopened WinDAS, and tried again. After many attempts, it finally worked, and I was able to read the G2 level: 154. It was easy enough to find in the .dat file by looking at the surrounding values. Notepad++ is safe to read from and write to the .dat files.
The Dell was pretty bright, so I changed the value to 134, saved to a new file, and selected Load Data to Set.
"Hold on to your butts."
The monitor turned off, the progress bar crept along for a moment, and then it failed. Totally dejected, I thought, "Well, I've ruined this one." But after turning the monitor off and back on, everything functioned normally. As I mentioned earlier, this failure kept happening, but as I slowly realized it wasn't catastrophic, I became less and less apprehensive each time.
After five or six more tries, it finally completed. The G2 value of 134 ended up being too dark, so I made a few more adjustments to hone in on a proper number. From 134 to 144 and finally to 141, this took more than twenty attempts. I didn't know which was the problem: the monitor or the Kootek adapter. I was hoping to find out when I tried the IBM. Anyway, from 154 to 141:

Much better.
WinDAS is quite powerful, and there's a lot to do in the Adjustment menu. I used the dynamic convergence to finely adjust what I couldn't correct with the OSD menu. But again, this crashed or lost connection a lot. After I finished all my adjustments, I forgot to complete the final setting procedure, so I got locked out of the OSD the next time I turned the monitor on.
Feeling confident, I went on to the IBM P260. The outer casing is much easier to remove than the Dell.

Before adjusting, I needed to undo my previous resistor mod. Right underneath that top shielding is this:

That 6.8M Ohm resistor needed to come off, and I needed to re-solder the old resistor on the backside of the PCB. Restored to its original condition, I was curious to see how bright the monitor was going to be.

Yikes. Visible retrace lines and super bright.
The ECS port on the IBM is a little harder to access, and it's quite hard to read the labels:

But it's the same as the Dell: TXD, RXD, +5V, and GND.
Following the exact same procedure that I used for the Dell, I only encountered one failure with the IBM. Everything went much smoother and was much less frustrating. I suspect the Kootek adapter was not the problem. The starting G2 value was 161! I had to drop this all the way down to 100 to get perfect blacks.
Dell on the left, IBM on the right:

It's hard to color match exactly, as one is on the DVI output of an NVIDIA card, and one is connected via VGA from Intel integrated graphics. And they need a little help from software gamma sliders to keep the blacks as black, but I'm happy with the results.
The Dell P1130 is actually the better monitor. It's newer, has less image warble when transitioning from dark to light scenes, and has a higher max horizontal scan frequency (130 kHz vs 122 kHz on the IBM). That higher scan rate means a max refresh rate of 170 instead of 160, and at 1600x1200, 102 Hz vs 96Hz. Not huge, but I'll take it.
The only problem is the noise, and that's where I'm hoping someone can help. Both monitors have the CRT buzz, but the Dell has a high frequency whine that comes and goes depending on the picture content. It's driving me crazy. Before I slather epoxy over every inductor inside the monitor, are there better solutions?
Also, if you haven't had enough pictures just yet, here are some really interesting high resolution shots of the interior of the Dell:
http://chattypics.com/files/DellP1130_1_4b94gemyas.jpg
http://chattypics.com/files/DellP1130_2_8ofni21pdg.jpg
http://chattypics.com/files/DellP1130_3_vgoubcztuz.jpg
http://chattypics.com/files/DellP1130_4_0rh8a74lpu.jpg
Most of the noise is coming from the area in the last image. It seems like they kind of made an effort to epoxy the inductors.