Norm Why wrote:
[snippage]
Thanks Paul. I removed one stick of Samsung 4GB DDR2 RAM and there is
progress. The BIOS never stops, it keeps trying repeatedly. I'm still
trying the get a better handle of front panel wiring. Beep codes would
give more debug info. Now the spinning of the CPU fan is concurrent with
GTX 970 fan. But still no output on HDMI. I may be able to get an early
2000's ATI Radeon VGA card locally for $5. The BIOS screen might be seen
without solving beep diagnostics. Maybe not. So I'll need to get BIOS to
recognize GTX 970. It worked on my old build.

It is said that when confused sleep. This confusion makes me very sleepy.
After sleep I arrived at some working hypothesis:

1. The front panel connectors are meant to be dynamic not static, so I moved
split POWER LEDs to SPEAKER+ and Minus.

2. If one RAM stick is good zero is better.

Observations: Still no beeps. Fancy color RAM LEDs still glowing. Continuing
CPU reboot cycle. Still no HDMI BIOS page from PCIe GTX970.

Working conclusions:

1. Zero or one RAM sticks makes no difference. Might as well use one.

2. Don't know function of RAM LEDs.

3. I still don't know operation of SPEAKER+ and SPEAKER-. No beeps. Does it
matter? Every time I use recommended polarity, I fail.

4. A PCI VGA adapter would be nice.

5. PCIe GTX970 will never be recognized until I boot Wn10 OS.

6. I will reinstall one RAM stick and wait for further sleep.

The manual says the four LEDs near the DIMMs are "phase LEDs"
and "The number of lighted LEDs indicates the CPU loading".
Your guess is as good as mine as to how the hardware drives
those LEDs, whether they're an actual current measurement
done with LEDs, or on the other hand, the VCore phases may have
enable signals, and as the BIOS SMM changes the number of
phases enabled many times a second, the LEDs are a side effect
of how many phases are being used.

Since the function is tied to "Dynamic Energy Saver" and runs
from the OS, a wild guess would be that all phases stay turned
on during BIOS POST (as nothing has attempted to request a
different phase setting just yet). This means, at a guess,
the LEDs aren't helping us right now. Unless they
flicker or indicate some sort of issue with VCore itself.

Since the CPU tends to rail on one core in the BIOS, you
would expect the CPU to be in a relatively high power
state during the POST sequence. I can watch this on
my current computer, the Test Machine, by using the
Kill-O-Watt power meter and monitoring power usage.
My machine draws 180W with CPU fully railed (Prime95),
120W on one core, 96W on idle, and POST shows 120W values
typically.

*******

I need more info on what you mean by "Continuing CPU reboot cycle".
This implies that perhaps there is still a power issue
in the picture, which could be related to the ATX power supply,
or it could be an issue with the VCore circuit. The 9650 is
in the CPU compatibility chart, so VCore should be beefy
enough to drive it.

To detect whether the board is coming out of RESET OK,
you can try monitoring the LEDs on the NIC interface. If
you have a router, run a cable from the router to the
NIC RJ45 and plug in. When the PHY comes out of RESET,
the LEDs will auto-negotiate (needs no CPU) and indicate GbE,
or 100BT or whatever. Not all motherboard products have LEDs
on the NIC connector area. Some will have two, like a yellow and
a green LED. On my GbE NIC, in GbE mode, the Yellow and
Green are both lit. If stuck in RESET, the LEDs are likely
to be OFF. If they stay OFF, and you know there is a good
eight wire Ethernet cable in place to your (running) router,
then the board could be staying in RESET and that's why it is
not POSTing and beeping like it should.

The ATX12V 2x2 should be plugged in. The cable has two yellow
wires (+12) and two black wires (GND). It can provide up to
144W of 12V to run VCore with. The connectors have a latch,
and with the latch engaged, the connector cannot "walk out"
due to thermal stress. It's the same latch that the
main 24 pin connector uses.

You can use 20 pin power on the 24 pin motherboard power input,
and pin 1 aligns with pin 1 in that case, leaving four motherboard
pins exposed and unused. The additional pins give slightly more
ampacity for three of the major rails. The most important contribution
there, is the yellow 12V wire in that section, as it provides
sufficient power to run a 6600 PCI Express video card (which needs
four amps). PCI Express video cards with no power connector
on the end, can draw up to 4.5A from the slot. High end cards
like your GTX970, draw only 2A from the slot, and lots
more current from the one or more power connectors on
the end of the card.

You know that video cards, high end ones, have PCIE connectors
on the end. The video card is unlikely to go to a high power
state during BIOS POST. If they didn't follow that rule,
my ATX supply would probably be flat by now :-) It's the
CPU that likes to spool up, while the video cards are
mostly well behaved during POST.

And since we're doing the "beep tests", one of the bonuses
of having no video card, is no video card electrical load.
So we know for sure the video card is not contributing to
the reboot loop or whatever it is you're seeing.

*******

If the ATX power supply fan is going "On and Off"
repetitively, then it could be an ATX power supply problem!
It can also be a problem with the PS_ON# signal sent
by the motherboard. There are a surprisingly large
number of PS_ON# signal failures, which has never
made sense to me. The pin only draws maybe 2mA,
yet even on boards with beefy drive (24mA? 48mA?),
the damn thing can fail to work properly. Part of it
has to do with the fact the motherboard input "PS_ON#"
is not a conventional logic signal.

While we might call that a "reboot loop", it's a
power issue of sorts. Modern supplies do not do this,
as they tend to latch off under fault conditions and
require the operator to "cycle" the rear power switch
to remove the fault state. Only a really cheesy supply
could go ON and OFF on its own, for fun. My first power
supply could do that, especially after the 12V became
so weak, it could only supply 0.2 amps before tipping
over.

Paul