5F6-A based build B+ voltage falling

[CascoSieg]

I've searched for info on this, but haven't found anything yet. I'm probably missing something basic - apologies in advance if this is a dumb question.
Recently completed my first build - 5F6A-based design. Followed generally accepted start-up procedures mentioned here and in a number of kit build guides I've looked at, and after fixing the "squeal" problem (swapping the OT leads on the power tube sockets) and confirming basic functionality with guitar and speaker (nothing sounds wrong with it, though most of my recent experience is with solid state amps, so nuances may be lost on me), I went back to document actual voltages everywhere.
Prior to installing the tubes I noted B+ A voltage around 480V, but didn't check this again until after playing through it, and noticed B+ showed to be about 120V lower. This doesn't seem right.
So I cycled the power (allowing the caps to drain each time, through the 200K parallel cap-drain resistor) and watched the voltage change as it powered up for both diode rectified (SS) and tube (5R4) rectified modes. The problem is the same in each case, and also with the 5R4 removed entirely, albeit with slightly different final voltages, as I think would be expected.
After switching the power on, with all tubes in sockets and rectifier set to SS, the B+ A rises to 465V over about 10 seconds, hovers there for a few seconds, then begins to fall, slowly at first then much quicker, leveling off at 366V after another 10 seconds or so. Remembering that the B+ was stable without tubes, I repeated this process removing one tube at a time starting with pre-amp section, and noted the same behavior until I removed the power tubes (6L6GC's), and the voltage stayed up at 480V. I reinstalled all but the power tubes and noted a final stable B+ A voltage of 475V.
Some additional data:
- I confirmed the PT HT voltage is 348V AC to CT for each lead. (PT is MM P45RS)
- I'm using a load box @ 16 Ohms for these tests. (OT is MM O45RS-L16)
- My 200w load limiter bulb glows orange dimmly throughout these tests and changes slightly when switching between rectification modes, and when voltages change, but is never all that bright.
- power tubes are biased on the cool side
- Pin 3 of the power tube sockets reads 450V without the tubes, 360V with tubes installed.
- I swapped in a pair of 5881's and the same happens.
- With power tubes installed, tracking both the power tube and preamp tube heater voltages shows similar pattern - they surge to 6.2 VAC on start and after 10 or so seconds fall off to about 5.7 VAC. Without the power tubes in there, all heaters are ~6.4 VAC.
- The load limiter does glow brighter as the B+ voltage falls.

Seems like a short somewhere, but inspection of sockets looks fine. Any ideas on what to check next?
Thanks!

[sluckey]

- The load limiter does glow brighter as the B+ voltage falls.

Is this a current limiter light bulb? If so, don't use it when checking voltages. Plug the amp's power cord directly into a 120VAC convenience outlet. Recheck voltages.

[PRR]

A peak and sag at turn-on IS normal.

120V drop seems a lot.

Ah, yes, that's the lamp-limiter which is only for initial smoke-test, to avoid burning down the house. Once you know no major mistake, listen to Sluckey, feed it raw 120VAC like you will play it.

[CascoSieg]

Ha! Removing the lightbulb load limiter certainly changed things.  Duh - Won't make that mistake again.

Now with SS rectifier, I've got the following values on the B+ rail (tho it bounces around by +/- 5 volts). Do these values seem within the range of standard operating?

B+ A: 454
B+ B: 452
B+ C: 410
B+ D: 360

Power tube(s) Pin 3: 451
Power tubes Pin 4: 451

V1 pin 1: 247
V1 pin 6: 265

V2 pin 1: 201
V2 pin 8: 202

V3 (PI) pin 1: 271
V3 (PI) pin 6: 256

Thank-you  :)  !

[sluckey]

Looks fine to me.

[CascoSieg]

So grateful for your your help!