Problems rebuilding system

[Norm Why]

Hi,

I gutted my old system because the MOBO would not boot. I have an Intel
Q9650 CPU and two sticks of DDR2 RAM, SSDs and a Zotac GTX 970 Single Fan
Graphics Card. I bought an official Intel cooler fan for the Q9650. I'm
still looking for a plastic funnel to vent the waste heat outside. To
accommodate my CPU and RAM I purchased a GIGABYTE GA-EP45-DR3LR MOBO from
Hong Kong with all the trimmings. When I went to install it in my old box, I
realized I needed a PC case for a full size ATX board. So, I ordered an
MX330-X gaming Midi tower from the Source, $40. I had fun attaching the
front panel connectors. I was able to satisfy myself the power and reset
switched were working. I have only USB2 and on the new box USB3 was not
installed on the front panel. Generally, I am satisfied with the MX330-X. I
just need more screws to secure the MOBO, so I don't get more sympathetic
(harmonic) vibrations in the future. My old 400W PSU was working fine and
seems to be working now. All power leads are attached and also to the GTX
970. When I first tried to power the system up, all fans came on except the
GTX 970 and no HDMI image on the screen. I cleared the CMOS with a shunt for
one minute and the GTX 970 fan came on but no image. I then tied 5 minutes
to clear the CMOS. BTW the GIGABYTE MOBO came with a battery but it was dead
so I used the one from the old board that checked was OK. Even after 5
minutes to clear the CMOS there is little progress. The GTX 970 fan spins
for a while then it shuts down. I understand the HALT on all errors is the
default but with no BIOS screen I can do nothing. there are many pretty
colored LEDs that say my RAM is doing something.

Any suggestions what to do next? Thanks.

Why

[Paul]

Did you get a manual ?

Did you read the section in the manual about "Clear CMOS" ?
All power must be removed when you use the shunt!

Since the battery was dead, you don't even need to do
a Clear CMOS. The settings are gone.

Page 31
https://download1.gigabyte.com/Files/Manual/motherboard_manual_ga-ep45-ds3l(r)_e.pdf

"Short: Clear CMOS Values" <=== this shorts a potentially low impedance power source.
Excess current can be drawn through a ~70mA dual diode.

"Always turn off your computer and unplug the power cord
from the power outlet before clearing the CMOS values. [ switch off, using the switch
on the back of the PSU is sufficient
to remove +5VSB. On an Asus mobo
the Green LED should be extinguished.
For Dell switch-less, unplug the PSU. ]

After clearing the CMOS values and before turning on your computer,
be sure to *remove* the jumper cap from the jumper.

Failure to do so may cause damage to the motherboard. [ Uh Oh ]

After system restart, go to BIOS Setup to load factory defaults
(select Load Optimized Defaults) or manually configure the
BIOS settings (refer to Chapter 2, "BIOS Setup," for
BIOS configurations).
"
*******

On many motherboards, the manual text section on Clear CMOS was wrong,
and participants damaged motherboards because of the errant
manual section. Some Asus motherboards had a single page "insert sheet"
containing a rewritten Clear CMOS section, with the proper
instructions on removing all power via PSU. Either by switch-off
at the back, or by pulling the power plug *before* using the
jumper. When the shorting jumper is removed again, then it is
safe to use the system.

The shorting jumper is only needed for a matter of a fraction
of a second.

So now the question is, was there an initial problem
with your board ? Or is the problem now due to the burning
of the ORing diode caused by improper usage of the CMOS jumper ?

I helped one guy repair such damage. But he was more
clever than me, and used a pair of regular silicon diodes,
instead of the Schottky I told him to buy, and the thing
still worked afterwards. (A regular silicon diode has a
higher Vf forward drop, and the CMOS battery will "appear"
to die sooner because of that. You get another two weeks
out of the CMOS battery if you use a Schottky dual-diode
as a replacement.)

For a schematic of the potentially damaged portion,
see page 76 here. This section of circuitry is common
to a large number of Southbridge and PCH chips.

https://www.intel.com/content/dam/doc/datasheet/io-controller-hub-6-datasheet.pdf

I've annotated the figure to show how it really works.
I have three motherboards with a "K45" on them. To see
the legend on the device, a magnifier is required.
It will look like other SOT23 transistors, so you
have to check the laser-printed letters on top.

https://i.postimg.cc/V6krB8tk/clear-cmos-K45-damage.gif

You would attempt to cut the legs of the device with
precision cutters (I broke my $100 precision cutter).
With the three legs cut and the device removed, you
clean up the pads a bit, then solder on the new one.
It's surface mount.

A TV repair shop could do that for you. They love
to solder stuff, and they also like to not have to
do troubleshooting to find the thing.

*******

Other possibilities:

1) BIOS not compatible with CPU. That's a reason
for a black screen on new systems. The CPU Support
table on the Gigabyte website indicates that's
just not possible for your 9650. So that isn't
the problem. The original BIOS works for 9650.

2) No Aux power to the video card. You forgot some
PCIE 2x3 or 2x4 connector. Highly unlikely.
If the CPU was running, it might beep the speaker
three times if the video was bad. On ATI cards,
they could pop up a red-decorated warning box on
the screen, that the power cable was not seated.

3) Many other problems could be detected with a
PCI Port 80 card (in slot nearest CPU). It's
possible there are PCI Express versions now, but
the topic of Port 80 cards never comes up any more.
A few expensive boards, have a two-digit Port 80
display right on the motherboard.

But that's about all that comes to mind.

Each mobo screw hole has a "keep out" zone marked in
the silk screen. No electrical components in the
motherboard design should enter the keep out circle.
As then, a screw head could touch a circuit. On a
certain Asus Nforce2 motherboard, one mounting hole
could short out an Audio channel on the sound subsystem,
leading to one dead speaker. Something was too close
to the keepout area. Design reviews are supposed to
catch shit like this.

Otherwise, the mounting holes are *designed* and
*intended* to be grounded. They help join the PCB
ground to the chassis ground. They do not need to be
insulated. On computer cases where the mounts are
made by bending metal (instead of using "posts"),
sometimes the job is so poorly done, the extra-wide
support "bump" shorts something out.

Paul

[John McGaw]

On 11/23/2019 10:31 PM, Norm Why wrote:
> Hi,
>
> I gutted my old system because the MOBO would not boot. I have an Intel
> Q9650 CPU and two sticks of DDR2 RAM, SSDs and a Zotac GTX 970 Single Fan
> Graphics Card. I bought an official Intel cooler fan for the Q9650. I'm

snip...

First things first. You say that there is no video of any sort. Have you
tried to verify that the video card you have is actually working? I would
expect to see something show up on the screen even if everything else in
the boot sequence was kaput.

Also, if the video card has multiple outputs (3 or 4 on your card I
believe) have you connected a monitor to each of them and tried the reboot
sequence? I got trapped in an "impossible" situation that sounds a bit like
yours a few years back and it turned out that everything was actually
working pretty much as intended but I had the monitor hooked to the wrong
video output and didn't see anything that was happening. Just a thought and
it should only take a few minutes to check...

[Norm Why]

"Paul" wrote

Thanks Paul for all the thoughtful and helpful suggestions.

The GIGABYTE GA-EP45-DR3LR MOBO, PDF manual is about 150 pages. At the very
bottom, troubleshooting advises getting in touch with GIGABYTE, Taiwan.
First one must register. Registration is a very good exercise where one must
provide all the info on the MOBO. One is asked what PCI VGA card is used.
Then I remembered the ten year history of my MCP73VE MOBO. It has a built-in
VGA port. Only after VGA is used to work with the BIOS program can one
download drivers for a PCIe GPU. On eBay cheap ones are available for $10.
There are two shops in town that might have a cheap one. Then after some
time I can download all the drivers from GIGABYTE. Hopefully by Monday it
should be working.

[Norm Why]

"John McGaw" <Nob...@Nowh.ere> wrote in message
news:iQwCF.137647$Ys1....@fx42.iad...

Thanks John. All the parts from my old build were working. But a more
important is what the GIGABYTE MOBO expecting. Troubleshooting is done with

GIGABYTE, Taiwan. First one must register. Registration is a very good
exercise where one must provide all the info on the MOBO. One is asked what
PCI VGA card is used. Then I remembered the ten year history of my MCP73VE

MOBO. It had a built-in VGA port. Only after VGA is used to work with the

[Paul]

Norm Why wrote:

>
> Thanks Paul for all the thoughtful and helpful suggestions.
>
> The GIGABYTE GA-EP45-DR3LR MOBO, PDF manual is about 150 pages. At the very
> bottom, troubleshooting advises getting in touch with GIGABYTE, Taiwan.
> First one must register. Registration is a very good exercise where one must
> provide all the info on the MOBO. One is asked what PCI VGA card is used.
> Then I remembered the ten year history of my MCP73VE MOBO. It has a built-in
> VGA port. Only after VGA is used to work with the BIOS program can one
> download drivers for a PCIe GPU. On eBay cheap ones are available for $10.
> There are two shops in town that might have a cheap one. Then after some
> time I can download all the drivers from GIGABYTE. Hopefully by Monday it
> should be working.

Northbridge video and your 970 video, have a "compatibility mode".
It is this unaccelerated frame buffer mode, which both the
BIOS and the OS use, when no drivers are present.

The 970 will have a "declaration ROM" on board, with the
details for any softwares that load the code in the ROM.
Included, is some information that allows setting up a
frame buffer at 0xC0000 or the like.

The BIOS draws in that buffer, as if it was a pixmap.

On older Windows, the "in-box" driver was called the "VESA driver".
In Windows 10, it's called the "Microsoft Basic Display Adapter".
Once the OS is installed and the install phase finished, Windows 10
will say "searching for video driver". It's at that point, it
acquires the NVidia proprietary driver for your 970, from
Windows Update (the driver is from NVidia, but is stripped
of PHYSX and Geforce Experience).

On an older OS, the user installs the NVidia driver manually,
from the CD that came in the video card box. Or, the user
downloads the latest driver from the web site. On an older OS,
you'd be "stuck" at 800x600, until the proprietary driver
is installed. The Microsoft Basic Display Adapter runs at
1024x768, and you don't feel quite as "cramped by the VESA driver"
in that case.

Currently, NVidia ships only 64 bit drivers for its newest cards.
There are still both 32-bit and 64-bit drivers on the website,
but the last 32-bit driver is two years old or so. Maybe if you
were to buy an RTX2080 today, there'd be no 32-bit driver at all.

Summary: There's always a driver. At any point in time, you
may not have the "best driver" (accelerated, BITBLT, 3D shaders),
but you will be able to see the screen via VESA frame
buffer mode.

Paul

[John McGaw]

I could help you out if you were my neighbor -- I've been cleaning out my
closet and box room full of computer junk and have found at least six video
cards, some of them going all the way back to a Matrox standard VGA. Good luck.

[Norm Why]

"John McGaw" wrote

Thanks John,

I just phoned a neighbor who has a box full of old parts. He may have a PCI
VGA card. I got a reply from a local shop. They no longer sell PCI VGA and
if they did it would likely be as expensive as new parts. It is a seller's
market for the best and most expensive.

[Paul]

Dude. To start debugging, you don't even need a video card.

Insert CPU and RAM. Try to make sure RAM is in good shape,
like use that "small" capacity stick you don't normally use,
a stick that won't overload a Northbridge. For example,
on my first computer, I had a 64MB stick used just for "bring-up),
and the 128MB and 256MB sticks were for full time usage.

The thing is, the PC speaker (1x4 connector on mobo labeled SPKR)
is your friend.

When the BIOS POSTs, it starts simple.

1) For a while, BIOS runs *without RAM*. It uses
register oriented code. it tests stuff. It reads
the RAM SPD chip. It then programs the Northbridge
memory controller.

2) Now, it carries out a RAM test. It does read/write to
RAM. If a written value does not come back, it beeps the
speaker a couple times. There is a repeating beep code,
sorta like an SOS ("Save Our Souls").

3) If the RAM passes, now it's time to search for video
and set it up.

4) If no video is present (card not detected), there is a
slightly different beep code. An SOS, but with a different
pattern.

Make *sure* the SPKR (1x4 pin header section) is plugged to
the case speaker, so you get the benefit of the SPKR output.
The speaker wires are connected to pin 1 and pin 4.

Now, without RAM or video card in the system and
CPU installed, and *2x2* ATX12V power plugged in,
there should be SPKR beeping because of the missing
hardware. If it won't beep, then the CPU might not
have power, the CPU could be blown, the motherboard
could be bad, the VCore regulator could have failed,
the usual sorts of things.

Instead of immediately wasting time on a search for
an obscure video card, I want to hear beeps first!

If there are no beeps, you have more serious problems,
like a BIOS flash version which won't run a 9650. Now,
normally, the initial release BIOS code runs a 9650, so
it really should have run, and there isn't really an
excuse for this one. Check that you have both the
main 24 pin and the 2x2 ATX12V plugged in.

To avoid problems with the computer case itself,
I test on the kitchen table first. First
insulate the motherboard (a phone book underneath
will do). The hard part is propping up a heavy video
card so it won't fail over. If you screw the video
cable to the faceplate, that's pretty strong. Then,
take the video cable and make a "sandwich" out of
heavy objects, so the cable cannot move, and then
the cable can keep the video card upright.

But just a beep test right now would be a start.

If you burned the ORing diode for the CMOS power
path, then the system can't start without that.
If there's no voltage on VBAT, there's no way for
the processor to read out stuff in the Southbridge
CMOS well. And that will stop your beep test, too.

A PCI Port 80 card can be used for a quick test
that "she's dead Jim". If the Port card display
stays at 0x00 or 0xFF and won't change, then the
machine isn't starting to POST at all. A jammed
RESET switch could do that, as an example of
another root cause type. Swap the RESET button cable
and the POWER button cable, so the bad reset switch
can be tested by using it to power up the system .
Unplugging the RESET button is also an option, as
at this point, you don't need it particularly.

Paul

[Peter Jason]

On Sat, 23 Nov 2019 19:31:01 -0800, "Norm Why" <nob...@microsoft.com>
wrote:

>Hi,
>
>I gutted my old system because the MOBO would not boot.

Throw it all away and get one of these...
https://www.intel.com.au/content/www/au/en/products/docs/boards-kits/nuc/nuc-family-overview.html

You get the NUC computer into which you plug a powered hub.
No more SATA plugs & sockets.

I did it after my old MB trashed data on several HDDs.
There are no easy diagnostics to trace problems on old equipment.

The little NUC is faster than my old gaming Gigabyte board.

Don't keep a computer for more than 5 years.

I have survived to tell thee!

[Adrian Caspersz]

I fear for you...

"WARNING: This product can expose you to Antimony oxide (Antimony
trioxide), which is known to the State of California to cause cancer. "

--
Adrian C

[Paul]

You're not supposed to lick the printed circuit boards.
As tasty as they might be.

And if a SATA port breaks on a motherboard, you may
be able to add a plugin SATA card to replace it. If
the whole Southbridge/PCH fails because the onboard
regulator isn't sending it power, that's not going
to work. It's only if you blow out a SATA port with
ESD, that an addin card will help.

Paul

[Peter Jason]

SbO3 is a fire retardant, which may be why California is on fire at
present; like every year!

Since I have found the true path, I keep all my HDDs & SSDs in an
external housing which is connected to the powered hub with just ONE
USB3 cable.
https://www.amazon.com.au/Mediasonic-HF2-SU3S2-ProBox-Drive-Enclosure/dp/B003X26VV4

[~misfit~]

On 25/11/2019 3:22 PM, Norm Why wrote:
> "John McGaw" <Nob...@Nowh.ere> wrote in message
> news:iQwCF.137647$Ys1....@fx42.iad...
>> On 11/23/2019 10:31 PM, Norm Why wrote:
>>> Hi,
>>>
>>> I gutted my old system because the MOBO would not boot. I have an Intel
>>> Q9650 CPU and two sticks of DDR2 RAM, SSDs and a Zotac GTX 970 Single Fan
>>> Graphics Card. I bought an official Intel cooler fan for the Q9650. I'm
>> snip...
>>
>> First things first. You say that there is no video of any sort. Have you
>> tried to verify that the video card you have is actually working? I would
>> expect to see something show up on the screen even if everything else in
>> the boot sequence was kaput.
>>
>> Also, if the video card has multiple outputs (3 or 4 on your card I
>> believe) have you connected a monitor to each of them and tried the reboot
>> sequence? I got trapped in an "impossible" situation that sounds a bit
>> like yours a few years back and it turned out that everything was actually
>> working pretty much as intended but I had the monitor hooked to the wrong
>> video output and didn't see anything that was happening. Just a thought
>> and it should only take a few minutes to check...
>
> Thanks John. All the parts from my old build were working.

Does this include the CPU? Did you have that exact combination of CPU, RAM and GPU running off that
PSU? Because experience tells me that and "old 400w PSU" might struggle to run what you've got.

You said "my old 400w PSU". Older PSUs split their rated output between 3.3v, 5v and 12v rails.
Modern CPUs and GPUs want all or most of their power on the 12v rail...

I suspect that it was the PSU in the library with a candlestick...

> But a more
> important is what the GIGABYTE MOBO expecting. Troubleshooting is done with
> GIGABYTE, Taiwan. First one must register. Registration is a very good
> exercise where one must provide all the info on the MOBO. One is asked what
> PCI VGA card is used. Then I remembered the ten year history of my MCP73VE
> MOBO. It had a built-in VGA port. Only after VGA is used to work with the
> BIOS program can one download drivers for a PCIe GPU. On eBay cheap ones are
> available for $10. There are two shops in town that might have a cheap one.
> Then after some time I can download all the drivers from GIGABYTE. Hopefully
> by Monday it should be working.

--
Shaun.

"Humans will have advanced a long, long way when religious belief has a cozy little classification
in the DSM"
David Melville

This is not an email and hasn't been checked for viruses by any half-arsed self-promoting software.

[Norm Why]

[snippage]
"Paul" wrote

> Did you get a manual ?

Manual is online. Gigabyte driver siftware is online

> Did you read the section in the manual about "Clear CMOS" ?

Yes. I'm not stupid.

> All power must be removed when you use the shunt!

Yes.

> Since the battery was dead, you don't even need to do

My bad.

[snippage]
> https://www.intel.com/content/dam/doc/datasheet/io-controller-hub-6-datasheet.pdf

> Other possibilities:
>
> 1) BIOS not compatible with CPU. That's a reason
> for a black screen on new systems. The CPU Support
> table on the Gigabyte website indicates that's
> just not possible for your 9650. So that isn't
> the problem. The original BIOS works for 9650.

Not true Q9650 and BIOS program working as expected. RAM detected When I
press <DEL> BIOS keeps trying. PS2 keyboard detected.

> 2) No Aux power to the video card. You forgot some
> PCIE 2x3 or 2x4 connector. Highly unlikely.
> If the CPU was running, it might beep the speaker
> three times if the video was bad. On ATI cards,
> they could pop up a red-decorated warning box on
> the screen, that the power cable was not seated.
>
> 3) Many other problems could be detected with a
> PCI Port 80 card (in slot nearest CPU). It's
> possible there are PCI Express versions now, but
> the topic of Port 80 cards never comes up any more.
> A few expensive boards, have a two-digit Port 80
> display right on the motherboard.

Not true or not relevant.

>
> But that's about all that comes to mind.
>
> Each mobo screw hole has a "keep out" zone marked in
> the silk screen. No electrical components in the
> motherboard design should enter the keep out circle.
> As then, a screw head could touch a circuit. On a
> certain Asus Nforce2 motherboard, one mounting hole
> could short out an Audio channel on the sound subsystem,
> leading to one dead speaker. Something was too close
> to the keepout area. Design reviews are supposed to
> catch shit like this.
>
> Otherwise, the mounting holes are *designed* and
> *intended* to be grounded. They help join the PCB
> ground to the chassis ground. They do not need to be
> insulated. On computer cases where the mounts are
> made by bending metal (instead of using "posts"),
> sometimes the job is so poorly done, the extra-wide
> support "bump" shorts something out.

No Shorts. Circuit board working as expected. This is V1.0 BIOS of 2008
board. After many attempts I have not found a PCI or PCIe VGA or HDMI video
adapter that is recognized by BIOS. Consider: what's in an adapter? Answer:
I/O ports, RAM addresses and ROM. A not yet updated 2008 V1.0 BIOS knows
nothing. This board has dual BIOS waiting to be update.

Probability theory says my best bet would be to find a part number for a
2008 GIGABYTE PCI VGA adapter. This is a hard problem for GIGABYTE techs and
me. I have tried using Wayback Machine for dates 01/01/2008 to 31/12/2009.
I'm so stupid I cannot even find a webpage for Microsoft in that date range.
Please help me Paul, you know Wayback Machine. Please find GIGABYTE web page
or GIGABYTE PCI VGA in that date range.

Thanks.

[Paul]

The only motherboard I know of, with that kind of
strange requirement, is an Asrock board with VIA chipset.
It has only a PCI Express x4 wiring for the video card
slot. And Asrock listed video cards they had tested to
work with their x4 wiring. I didn't think this was really
necessary, but Asrock insisted that some video card
products would not work. This could be the "Rev1.1 problem".

But many other motherboards have no such limitations
and they have a full video card slot. The only video slot
problem, is some (again, VIA) products, which don't
tolerate Rev2 lanes versus Rev1 lanes. So some early
revisions of PCI Express do cause problems, because
the automatic hardware speed negotiation doesn't work
properly.

The chipset involved here is P45. It's not one of the
crappy chipsets. I don't see a reason for this to fail
in that way (negotiation problem at startup).

The thread here is a random sample of such info. It
suggests newer cards having a problem in Rev1.1 slots.
But yours has a Rev2 slot.

https://forums.tomshardware.com/threads/pci-express-1-1-a-problem.1833049/

The video slot on P45 is Rev.2 . There should not be
a problem either backwards or forwards with that. I
have an X48 (Rev.2) and a Q45 (Rev.2) and no problems
seen with them. And I have an instance of the Asrock
with the "potentially problematic" x4 wired x16 slot Rev1.1,
but I've never tested a PCIe card in that one. It's
a neat board that has two video card slots, an AGP and
a PCIe, and I always used the AGP on it.

I have tried my newest video card in the X48. And it
failed. But it failed for a good reason - not enough
address space for mapping (no beeps). The motherboard is not
new enough to have uncapped addressing. The video card
has too much memory on board for its own good. My
newest motherboard is interesting, in that the mapping
done gives "8GB onboard available" with a 64 bit OS,
and a "2GB onboard available" with a 32 bit OS. Which means
the address map is adapting some how so that the 32 bit OS
is not "starved out" by the video card. I don't think
the system I'm typing on, is quite that sophisticated,
which is why it black screens with the "large" video card
in it. The rest of my video cards are 1GB or smaller onboard.

So at the moment, I don't see a good match for "problematic
combo" in your case. The only thing that comes to mind
when those ideas are exhausted, is a power-related issue
(not enough juice for video card). And that's easy enough
to happen - I've had one case where a wire burned on the
PSU harness, and that's why the video card had "not enough
power" symptoms. I had to solder a cable to the video
card to replace the power connector on the card in that
case (no spares in my repair box for a proper fix).

*******

One trick with the Gigabyte web site, is the name changed.
It was gigabyte.com.tw at one time, and later became
gigabyte.com .

This should work.

https://web.archive.org/web/20080224022719/http://www.gigabyte.com.tw/

The problem is, one of the links on that page is broken,
so we can't see a list of their "New" Rev.2 boards. There's
only a list of some Rev1.1 or so, boards. The 8800GTX might
have been the last of Rev1.1, in terms of "big-name" video
cards.

*******

If you are "not rich" and "don't have enough hardware for this test",
you should see if any shop offers a "$25 diagnosis service". In
the past, there were some newsgroup participants, who were
able to find mom&pop stores offering such a test. If the
tester is good, they can give feedback on what worked or
didn't work, to make a board work.

I've had to buy double the hardware on one system, to
retest every card that goes into the system, to get
things working. That's my most expensive "home diagnosis"
experience. I don't want to send you through that,
if a "$25 diagnosis" or whatever it is today,
will achieve the same results. But such tests are
subject to local availability, and lord knows what
a diagnose costs today. With inflation, the price
now might actually be unattractively high.

The only video card slots with some sort of ESD
problem, were NVidia Northbridge chipsets. Some of
those arrived out-of-the-box with one dead video
slot of two. Yours is Intel. Intel Revision2.

The other problem way back when, was the BIOS would
not accept anything other than a video card in
slot #1. This was some sort of BIOS bug. Later systems
(yours likely included), can have USB3 cards, SATA cards,
whatever you want, in the video card slots. It was just
a couple first-release BIOS that had an issue with
card type. Video would work, but not declarations
of other card types. And a BIOS update would fix it.

Since you have a dual BIOS, you could switch back
and forth between them. Normally, the contents would
be identical on the two BIOS. AFAIK, the way GB works,
is there are two main code modules (one in each Flash chip),
but only one of the chips has a boot block. This leaves
a single-point-of-failure in the Gigabyte design. I'm
sure the GB Tech Support will correct me on this,
but I was reading this somewhere years ago, as to
how theirs worked.

Note: Don't forget to hook up SPKR to the header.

We need those "beep" results.

The beeps tell us what is alive. With the video card pulled,
your system should be able to beep the "video
is missing" on the computer case SPKR (or the piezo
disc which is used on OEM boards in place of SPKR).

Both "RAM Missing" and "Video Missing" beeps are
possible. Start with both RAM and video pulled
(switch off PSU before add/remove). Listen for
"RAM Missing". Say it is two beeps. Now, plug in
RAM, listen for "Video Missing". Say, that is three
beeps. The beep pattern of the two test cases should
differ. The BIOS is able to run without RAM or Video.

You only get beeps, if CPU and BIOS are available.
No beeps, means you should look elsewhere besides
the current fixation on video. On a dual BIOS, if
no beeps, flip and try again.

The reason the PC beeps 1 time at POST, is "lamp test",
a proof that SPKR is working.

Methodical testing helps triangulation...

Paul

[~misfit~]

On 11/12/2019 5:26 AM, Paul wrote:
> Norm Why wrote:
>> [snippage]
>> "Paul" wrote

<much snipping>

>> Probability theory says my best bet would be to find a part number for a 2008 GIGABYTE PCI VGA
>> adapter. This is a hard problem for GIGABYTE techs and me.

You've got Gigabyte techs helping you with an 11 year old mobo that you bought from ebay? That's
very impressive!

> So at the moment, I don't see a good match for "problematic
> combo" in your case. The only thing that comes to mind
> when those ideas are exhausted, is a power-related issue
> (not enough juice for video card).

I believe my (unanswered) input into this thread suggested that the GPU wasn't getting enough
power. IIRC the OP is using an "old 400w PSU" that worked fine with a slower CPU with fewer cores
and in an older mATX mobo (with this GPU?).

As I said in my previous post there's a good chance that the PSU simply isn't providing enough on
the +12v rails. The OP seems have a penchant for using 'old gear' so the chances are good that his
"old 400w PSU" is in fact ancient going by modern standards. Ancient enough to be putting out most
of it's rated power on the +3.3v and +5v rails and nowhere near enough on the +12v rail.

I'm not sure I'd be comfortable running a Q9650, a GTX 970 and SSD / HDD RAM etc from a *modern*
400w PSU (unless it was from a tier 1 manufacturer) yet alone a (possibly) ancient one.

I ran a Q9650 for a time and, unlike modern CPUs they don't have very low power 'idle states' so
are constantly drawing quite a few watts. A GTX 970 pulls 69 watts at idle and 256 watts under load
(according to Techspot*). All from the +12v rail.

* <https://www.techspot.com/review/885-nvidia-geforce-gtx-970-gtx-980/page7.html>

Most hardware sites recommend using at least a minimum 500w PSU with a GTX 970 and quad-core CPU.
And that's a modern (as in >75% of rated power on +12v rail) PSU.

Just for completeness' sake I just grabbed a PSU from my 'still works just fine but obsolete' pile.
It's a Thermaltake Xaser Purepower 480W with an 04/10 date code. It's rated 18A on the +12v rail.
That's 216 watts according to my calculations. Not enough for just that GPU under load yet alone
the rest of the PC.

[naugl...@gmail.com]

Thanks Paul, I did not find my GA-EP45-DS3L until Dec. 16, 2008. Back then they recommended ATI. I used WBM to find www.ati.com. That was worse. ATI was acquired by AMD so MBW was very confused.

One kind gentle whom I may have unfairly roasted suggested I check power. I checked the ATX power cable and it was not tight, so I tightened it. I can check +5, +12 and -12 (if present) on a SATA power cable. It is hard, and I haven't gotten there yet, but it should be good, given a new 400W PSU. This is the fourth PSU. The 3rd I sold because I falsely diagnosed a sympathetic (harmonic) vibration. NOT the PSU but a bad bearing in the CPU fan. It is good to sell good parts, one finds new friends. The first 250W PSU failed in the manner described by kind gentleman I roasted. Using a multimeter I showed the +12 rail had drooped. Then went to 400W. Now I will go backwards and recheck the various boards offered me.

[Norm Why]

Thanks Paul, I did not find my GA-EP45-DS3L until Dec. 16, 2008. Back then
they recommended ATI. I used WBM to find www.ati.com. That was worse. ATI
was acquired by AMD so MBW was very confused.

One kind gentle whom I may have unfairly roasted suggested I check power. I
checked the ATX power cable and it was not tight, so I tightened it. I can
check +5, +12 and -12 (if present) on a SATA power cable. It is hard, and I
haven't gotten there yet, but it should be good, given a new 400W PSU. This
is the fourth PSU. The 3rd I sold because I falsely diagnosed a sympathetic
(harmonic) vibration. NOT the PSU but a bad bearing in the CPU fan. It is
good to sell good parts, one finds new friends. The first 250W PSU failed in
the manner described by kind gentleman I roasted. Using a multimeter I
showed the +12 rail had drooped. Then went to 400W. Now I will go backwards
and recheck the various boards offered me.

---------------------------------------------------------------------------

I was not able to test voltages on SATA power cable, but did so on PCI
connector. Of academic interest: I have an USB/SATA adapter. With 3.5 HDDs
solution to drooping power was to buy new PSU. For 2.5 laptop HHDs my
USB/SATA adapter delivered +5V to power drive. Drill is: put non spinning
drive in freezer. Warm. Then use USB/SATA adapter to recover files. Discard
drive. +5V is sufficient power. However this trick does not work with SSD
drives. Requirement of +12V may be security measure. Hence: 1. this is one
more example of hardware specific hardware and 2. this is a reason I need to
make my new build work.

PCI voltages: Only +5V is present . I was able to find +5V and COM on some
pins shared on side A and B. Using probes on two PCI slots, I confirmed +5V.
First it peaks and then falls back, consistent with smart power management.
I may be able to confirm +5V on an old Fax/modem adapter but that would
require solder leads onto board. Useless idea.

Gigabyte does not make video adapters but does recommend ATI, now AMD. I
found that in 2008, ATI released ATI Wonder in a decades long series. In
2008 ATI still supported AGP. ATI Wonder evolved through CGA, MGA and VGA
for IBM compatible PCs. Apple is different. EVGA is now current. In 2008,
ATI even offered an ATI Wonder VGA with composite video for TV. Long ago I
may have owned an ATI Wonder, they were common.

Google search revealed such prices through the roof. Why would a low tech,
decrepit adapter appreciate in value? Supply and demand and hardware
specific hardware. There are many people trying to rescue old
hardware.Plug-n-Play only works in Microsoft Windows.

My task now is to fund a 2008 ATI Wonder PCI VGA adapter nearby and not to
pay hundred$ of dollars.

Ideas? eBay and Amazon are clip joints. Does any reader of this newsgroup
have a 2008 ATI Wonder PCI VGA adapter?

Thanks.

[Paul]

The difference between an ATI Wonder and an ATI card,
is the Wonder has a TV tuner. This does not materially
affect the problems you're having. You don't need the
Wonder part to solve a technical problem. Specifying
Wonder just adds to the card price and that's it.

Any old PCI based card would do, if you want to get
video without a hassle. PCI is 33MHz 32 bits wide on
the average desktop, and this is not exactly a high
speed interface. It's about as challenging as
making an IDE ribbon cable work.

AGP slots go up to 8X, and the interface has a
clock and can do eight bus transfers within one
clock tick (of the lower speed clock). There were
all sorts of problems over the years with this.
One USENET poster seems to have had some
first hand engineering experience with AGP,
and only had swear words to offer.

PCI Express is similar, in that like SATA, there
are attempts to be backward compatible, that don't
always work out. PCI Express 2 cards in PCI Express 1.1
motherboards were a problem in a few cases. Some of the
PCI Express 2 cards, the manufacturer offered a
video flash upgrade that would cause the card to
have the interface set at the PCI Express 1.1 rate
instead (so the card would refuse to negotiate
and screw up, and would always use the correct
speed without any assistance). Maybe you could
get a video flash upgrade for the 64KB flash chip
on the 8800GTX. That was somewhere in that era.

Even if you acquired a PCI 6200, that would be
good enough to get a BIOS screen working. Most Windows
OSes would run that screen at 800x600 without a
driver, and Windows 10 could manage 1024x768.

PCI cards "don't need to match era". No need to
aim for the year 2008. A 6200 PCI offers PCI without
bridge chips. An HD3450 would offer PCI by bridging
PCI Express to PCI with a separate chip. And a
card which delivered an AGP connector to the motherboard,
it would use a bridge chip like ATI Rialto to convert
from PCI Express (on the GPU) to AGP (at the motherboard
connector). Those are some examples of how
it's done. Today, you could take practically
any PCI Express card, and with a good bridge
chip, run it in a PCI slot by using a PCI
to PCI Express bridge card.

That being said, the market doesn't have much to offer
in the way of PCI video cards. Even though bridging
new kit is easy. There's a lot of used cards on
Newegg for $50, but that's a bit too high. For that
price, you would expect a new card.

This is for situations where you have a low profile
card you'd like to use. The problem with this card,
is the end away from us in the picture, needs
the end of the connector sawn off (to allow the usage
of x1, x4, x8, x16 cards in the adapter). This
one uses a *closed* x1 connector. The second link shows
the kind of connector Startech *should* have used.

https://www.newegg.com/startech-com-model-pci1pex1-pci-to-pci-express/p/N82E16815158190

https://en.wikipedia.org/wiki/PCI_Express#/media/File:PCIe_J1900_SoC_ITX_Mainboard_IMG_1820.JPG

Now, this adapter plugs into the previous adapter,
giving us overall, a PCI Express x16 to PCI adapter.
It solves the closed x1 connector problem.
The card is passive, just a slab of FR4. There is
no mechanical support to speak of. The box contents
don't show a "stubby" faceplace metal piece to be
affixed to the FR4 drill holes.

https://www.newegg.com/startech-com-model-pex1to162-pci-express-to-pci-card/p/N82E16815158223

So again, a swing and a miss. I'm sure with a little
shopping, a better adapter is out there. Just a matter
of a crappy Google search coughing the thing up.

If I needed a PCI video card today, I have one surplus
place I could try that has "old new stock". I got my
IDE DVD writer there. Other than that, I've have to
go to the recyclers and get one.

Paul

[Norm Why]

> > Thanks Paul, I did not find my GA-EP45-DS3L until Dec. 16, 2008. Back then
> > they recommended ATI. I used WBM to find www.ati.com. That was worse. ATI
> > was acquired by AMD so MBW was very confused.
> >

> > One kind gentleman whom I may have unfairly roasted suggested I check power. I

Thanks again Paul. After completing all my legal and medical business today I returned to search the web. It should be noted that it is best to find part numbers to distinguish different features. For "ATI Wonder" here are two:
MPN: 1090005910
100-703271
If I had a complete list it would be long and modify your view. However, 'search and ye shall find'. I found a museum at YorkU:

http://www.cse.yorku.ca/museum/collections/ATI/ATI.htm

Here is a bigger list of 12 ATI cards from that era:

ATI VIP graphics card (1988)
ATI VGA 1024 graphics card, v4-01 (1989)
ATI VGA 1024 graphics card, V60M-1.03 (1990)
ATI VGA WONDER+ graphics card (1990)
ATI 2400etc/e modem (1990)
ATI 28300 SA Graphics Adapter (1991)
ATI ATi Graphics Vantage card (1991)
VGAWonder XL24, ver. 4.1, ATI (1992)
ATI 14.4I/R.1.625 board (1993)
ATI VGAWONDER GT graphics card (1993)
ATI All-in-Wonder prototype, PCI bus, multimedia board (1994?)
ATI PCI MARCH64 video card (1996)

The one species of cards that we had fixated on is not there but WONDER+ (1990) is. I am going to write to the CompSci YorkU Prof for his help.

Merry Christmas

[Paul]

Norm Why wrote:

> Thanks again Paul. After completing all my legal and medical business today I returned to search the web. It should be noted that it is best to find part numbers to distinguish different features. For "ATI Wonder" here are two:
> MPN: 1090005910
> 100-703271
> If I had a complete list it would be long and modify your view. However, 'search and ye shall find'. I found a museum at YorkU:
>
> http://www.cse.yorku.ca/museum/collections/ATI/ATI.htm
>
> Here is a bigger list of 12 ATI cards from that era:
>
> ATI VIP graphics card (1988)
> ATI VGA 1024 graphics card, v4-01 (1989)
> ATI VGA 1024 graphics card, V60M-1.03 (1990)
> ATI VGA WONDER+ graphics card (1990)
> ATI 2400etc/e modem (1990)
> ATI 28300 SA Graphics Adapter (1991)
> ATI ATi Graphics Vantage card (1991)
> VGAWonder XL24, ver. 4.1, ATI (1992)
> ATI 14.4I/R.1.625 board (1993)
> ATI VGAWONDER GT graphics card (1993)
> ATI All-in-Wonder prototype, PCI bus, multimedia board (1994?)
> ATI PCI MARCH64 video card (1996)
>
> The one species of cards that we had fixated on is not there but WONDER+ (1990) is. I am going to write to the CompSci YorkU Prof for his help.
>
> Merry Christmas

The All-In-Wonder are the ones with TV Tuners onboard.
That stopped in the year 2008 or so.

https://en.wikipedia.org/wiki/All-in-Wonder

Apparently ATI Wonder is just a flavor of video card marketing.

ATI Theater chips came after 2008, and were sold as
separate cards. Diamond Multimedia had an ATI Theater
capture card, for example. In this example, you can
just barely see the ATI branding on it. This is a
TV tuner capture card. With a TV and an FM 75 ohm input.

https://images.hothardware.com/static/articleimages/Item720/card.jpg

ATI stopped this business, because the software development
was costing too much. That's why they made the TV tuner card companies
provide their own capture software (and much of that
capture software sucked). This kinda crushed the business
and made it unattractive.

After that ATI/AMD stuck with video card functions, to make a buck.
They got out of that distracting business and concentrated
on their strengths.

*******

Try to concentrate on the bus compatibility issue here.
I don't think the words "All-In-Wonder" help in that regard.

I don't want to see you wasting money on garbage.
There should at least be a reasonable plan for what
any item like this, hopes to achieve.

The purpose of finding a PCI video card, is just so
you can see the screen. This assumes the rest of the
motherboard/RAM/CPU are working.

This is why I want you do those damn beep tests first!
Fuck the video card. You need to test that mobo/CPU
work with no RAM and no video in place. Add RAM back,
observe that "missing video" beep happens. These
are basic tests to prevent you from wasting your
time on acquiring a video card.

If the appropriate beeps come out of the computer SPKR,
*then* you can settle in searching for some sort of
magical video card.

If the machine won't beep... a video card will *not* help.

Asus made one motherboard, where due to some BIOS issues,
the SPKR will not beep at all. I'm not aware of any
Gigabyte boards suffering from this sort of issue.
They wouldn't react the way Asus did at the time.

Various "Port 80 POST cards" exist. Some enthusiast motherboards
have this two digit display right on the motherboard, as an
aid to debugging. If you already own one of these, you could
use it. I do not recommend buying these, as the benefits
they product are not great enough to justify the purchase
price. The value in these is "Go/No Go". If the display
shows a value other than 0xFF or 0x00 or the display
goes blank (when the system starts to boot), fine.
The status LEDs that indicate system power rails are
pretty to look at (one poster was able to detect
a power failure by looking at the four LEDs). But
the values on the digits are almost worthless in
figuring out root causes of motherboard failures.
Consequently I don't recommend these for general usage.
The PC SPKR by comparison, is a cheap cheap way to test.
It's a kind of GO/No Go test too. But you don't have to
buy anything (as long as the computer case has the standard
SPKR and two wire cable with the 1x4 connector on the end).

https://www.computerhope.com/jargon/p/postcard.jpg

Paul

[Norm Why]

Yes. I owned an ATI All-In-Wonder. It turned my PC monitor into a TV. In the
Apple era folks turned their TV into a monitor.

> https://en.wikipedia.org/wiki/All-in-Wonder
>
> Apparently ATI Wonder is just a flavor of video card marketing.

Their are different flavors of ATI Wonder. You need to distinguish by part
number. This is a general rule. Only with part numbers can with distiguish
specs of similar products. Names like 'ATI Wonder' are just for marketing.
From WBM in late 2008, when my GA-EP45-DR3LR MOBO was found Gigabyte ATI
Wonder was adfvertised by Gigabyte. People resell old parts. ATI Wonder is
no where to be found. Strange, what does that tell you. Gigabyte is not the
only MOBO maker. If hardware specific hardware was the industry norm, that
would tell you someting about resale market in subsequent decades.

Thanks. I'll try that. But the beeper is in the MX-330-X chassis. I need to
check that electrically first so as to not get a false diagnosis. When I did
not see the POWER LED, I needed to switch the polarity. The front panel
connectors are tiny. Child size finger would help. I have not yet seen disc
activity LED, since not yet booted. When booted and not seen, I'll need to
switch the polarity. The + / - marks on leads are tiny. The colored LED's on
the MOBO near RAM say RAM is good and recognized.

Thanks for the fine diagnostic advice. YorkU ComSci/EE has bad telecom. They
are closed to the outside world. I have tried an ATI Rage card. A local has
offered an ATI Radeon VGA card that I have not tried yet. I think I owned
one once. But I recycle on Craigslist and don't have a box of old parts.

From manual, here are the beep codes:

Q: What do the beeps emitted during the POST mean?
A: The following Award BIOS beep code descriptions may help you identify
possible computer problems.
(For reference only.)
1 short: System boots successfully
2 short: CMOS setting error
1 long, 1 short: Memory or motherboard error
1 long, 2 short: Monitor or graphics card error
1 long, 3 short: Keyboard error
1 long, 9 short: BIOS ROM error
Continuous long beeps: Graphics card not inserted properly
Continuous short beeps: Power error

Since I have not heard any beeps when the table says I should, that would
imply a problem with the front panel connecter. I'll be back after Xmas
duties.

[Paul]

Norm Why wrote:

> From manual, here are the beep codes:
>
> Q: What do the beeps emitted during the POST mean?
> A: The following Award BIOS beep code descriptions may help you identify
> possible computer problems.
> (For reference only.)
> 1 short: System boots successfully
> 2 short: CMOS setting error
> 1 long, 1 short: Memory or motherboard error
> 1 long, 2 short: Monitor or graphics card error
> 1 long, 3 short: Keyboard error
> 1 long, 9 short: BIOS ROM error
> Continuous long beeps: Graphics card not inserted properly
> Continuous short beeps: Power error
>
> Since I have not heard any beeps when the table says I should, that would
> imply a problem with the front panel connecter. I'll be back after Xmas
> duties.

It requires a working motherboard+CPU combo.
Then you'll get beep codes on SPKR.

SPKR is not polarized. Just make sure the span
is correct for the connector (some OEM cases might
have a 1x2 connector, while some retail motherboards
use a 1x4 pattern). You can use connector strips and
lift the tab on each, to shift wires within shells
and make the correct connector for the job.

The SPKR body should be electrically isolated from
the two wire leads. The wire leads go to the speaker
coil, and the coil moves in and out with the cone.
Even so, a few computer cases here use plastic
mounts for SPKR.

The front panel header is reasonably bulletproof.

RESET and POWER are SPST momentary contact switches, normally open.
You push the button, and the switch closes for a moment.

The LEDs are polarized as you say. The LED is rated
for 5V PIV (Peak Inverse Volts) so cannot be harmed
by reversing the leads. Each driving circuit has
a series resistor to limit current flow. If they don't
light, you reverse the 1x2 connector and try again.

Normally the spacing and span of various signal
assignments on the Front Panel connector, avoids
the possibility of shorting two power pins placed
next to one another. If there is a SPKR section
of a Front Panel connector, the "hot" end should
not be near a GND pin. The ones I've looked at,
generally have fairly safe signal assignments.
While there are connectors on a computer that get
crushed and you cannot visually check them,
the Front Panel connector isn't one of the ones
where people habitually do bad things. I don't
think I've had any reports of anyone managing to
start a fire using nothing but a Front Panel header
problem :-) (If you pinch the "hot" wire of SPKR
between the computer case door and the computer
case, which is GND, then the wire will get smoked.
And that has happened. Normally those wires aren't
sitting near the door.)

You can get a "no beep" condition, by using a
reset button crushed in the ON position. Usually
OEM computer cases are the ones with sufficiently
cheesy buttons on the front of the computer, to
make incidents like this possible. I have one
computer here, where the buttons are such,
I know that some day that's how those buttons
will fail. The buttons speak of cheapness.

Paul

[Norm Why]

[snippage]

The GA-EP45-DR3LR has dual BIOS, two tiny surface mount chips near the
battery and CL_BIOS pins. It is designed for safe upgrade, with networked OS
software. Gigabyte gives recommendations on what software to use. BIOS is
V1.0 It is an unused MOBO, never sold. It came in a brown cardboard box with
the IO shield attached with a plastic strap. It is a virgin. It had been in
storage so long the battery was dead. It has nice features, like dual
channel memory. At one time, high end RAM was advertised as 'dual channel;.
It is not a property of RAM but of the MOBO. With dual channel memory and a
3 GHz 4-core Q9650 with 12MB cache, why worry? The advantages of 8-core may
be exaggerated. A GTX 970 GPU gives it more horsepower. Its contribution
showed up in "Task Manager".

What I am saying is that Big Bucks Boyz may get better economy with an old
system, by trading labor for capital.

[Norm Why]

Thanks Paul, I'll look into it.

[Norm Why]

[snippage]

> You can get a "no beep" condition, by using a
> reset button crushed in the ON position. Usually
> OEM computer cases are the ones with sufficiently
> cheesy buttons on the front of the computer, to
> make incidents like this possible. I have one
> computer here, where the buttons are such,
> I know that some day that's how those buttons
> will fail. The buttons speak of cheapness.
>
> Paul

Paul, I may have found the problem, no beep. On the one hand the
GA-EP45-DS3L provides a 10 pin keyed audio header while the MX330-X provides
a compatible plug. The mike and line-out go to connectors on the IO shield.
I have headphone-mike connected but that is not where the beep goes.
GA-EP45-DS3L provides a 20 pin front panel header. I have insured that
polarity of POWER, RESET and POWER LED are correct. I do not yet know about
HDD Activity LED. But when I look closely at GA-EP45-DS3L manual, polarized
Speaker Leads are connected to pins 20 and 14. I suppose that is a good
idea. However, the MX330-X front panel header ribbon cable does not provide
these, none that I see. I guess I need to send an email to Cougar customer
support. Odd, cannot see this problem flagged in any review. Maybe I need
to dig into the MX330-X with a flashlight and magnifying glass? Remember the
pin connectors are real tiny. The MX330-X manual is one page, Mickey Mouse.
I need better documentation.

The Power Switch is confirmed.
The Power LED is confirmed.
The Reset Switch is confirmed.

Maybe the HDD Activity LED is in fact polarized Speaker Leads. These pins
are so tiny it is hard to tell. But it is only these that need closer
inspection.

[Norm Why]

> [snippage]

> The Power Switch is confirmed.
> The Power LED is confirmed.
> The Reset Switch is confirmed.
>
> Maybe the HDD Activity LED is in fact polarized Speaker Leads. These pins
> are so tiny it is hard to tell. But it is only these that need closer
> inspection.

This is a URL to an image of the MX330-X front panel connector:
"How-to-connect-case-cables-system-panel."
https://assets.rockpapershotgun.com/images//2018/06/How-to-connect-case-cables-system-panel.jpg

Note there is no HDD Activity LED between pins 1, 2. They belong elsewhere.
On the GA-EP45-DS3L they belong on pins 20, 14. In this image they appear
on pins 18, 14.

Now to sleep.

[Norm Why]

[correction, even and odd # pins are on opposite sides of header]

Houston, we have a problem. HDD Activity LED pins 1, 3 and MX330-X
connection is fused and clearly marked HDD Activity LED & polarity. While
Power LED MX330-X connectors are not fused, polarity marked and might be a
candidate to connect to pins 20, 14, they are confirmed as Power LED. When
ON the Power LED glows steady and does not pulse as expected for front-panel
speaker beep. Maybe an email to Cougar is needed. Strange, this problem was
not flagged in reviews.

[John McGaw]

On 12/13/2019 6:11 PM, Paul wrote:
> You can get a "no beep" condition, by using a
> reset button crushed in the ON position. Usually
> OEM computer cases are the ones with sufficiently
> cheesy buttons on the front of the computer, to
> make incidents like this possible. I have one
> computer here, where the buttons are such,
> I know that some day that's how those buttons
> will fail. The buttons speak of cheapness.

Anything mechanical is always suspect. I've been using a couple of
identical Antec cases, not normally noted for being cheap, for computers
now residing in the basement "computer room". Not too long after purchase
(but outside any hope of warranty) the power switch failed on one of them
and drove me crazy for a while. Rather than give up I swapped the
connections between "power" and "reset" on the MB and it has worked fine
for years now. Only real problem is that I can almost never remember which
of the cases has the swapped switches. If I had any sense I'd apply some
masking tape over the dead switch to remind me...

[Paul]

ga-ep45-ds3l(r)_e.pdf

Page 27

FPANEL PWR PWR
LED Switch SPKR
+ - + - + -

X X X X X - - X
X X X X X

+ - - +
HDD RST
LED Switch

Polarity of RST Switch, PWR Switch, SPKR, does not matter.
All assemblies are electrically floating.

PWR LED and HDD LED can be reversed if they refuse to light.

The nine pin section does not need to be
completely wired up while testing. The two
LED indicators are superfluous, until final
build time. Just the PWR switch can be connected
for the style of testing to come.

What is not optional at this point, is the SPKR connection.
You need the SPKR connection to verify that the "missing RAM"
and "missing video" functions work properly, which tells
you that the motherboard/CPU is working, and that when
you plug the RAM in, the RAM is detected.

Once you have completed that phase of testing (successfully),
then you can go back to solving the video card issue,
whatever it is. Since the chipset is P45, I can't see a
reason for the motherboard to be screwing this up. You
should have good luck with PCI Express compatibility
with such a motherboard.

If the card has too much onboard GPU RAM, then it's possible
the OS won't start (address space issue). When doing
the "RAM detected, video not detected" test, you'll be
using a single stick of RAM. Use one of your
*lower capacity* sticks. The smallest I have is
probably 1GB for this particular test you're doing.
In years past (year 2000), you could get 64MB as
a stick, but the "smallest" has gone up a lot since then.

Your GPU has 3.5GB of onboard RAM (GTX970). But the BIOS
screen should always come up, unless the "max address space"
is exceeded. Let's say that is 8GB, just pulling the wrong
number out of the air. I could use a 2GB single stick
of RAM plus that video card, for 5.5GB total. I could
be reasonably assured that the BIOS screen would then come
up.

Once you can see the screen, *future* tests can
use your bigger DIMMs. The objective is to
get to see the BIOS screen first. Once you prove
there isn't a motherboard hardware failure, then
lots of other test cases will make sense to do later.

Windows 10 can start with as little as 256MB of system
memory (tested in a VM environment, to see if it was
possible). While the stated minimum is 1GB, that's
the stated "comfortable" minimum, not the "absolute"
minimum. If you run Windows 10 with 256MB, a task
called the "Memory Compressor" runs almost continuously,
as a measure of how "annoyed" the OS is about this
development :-) When you use 1GB of RAM, the
memory compressor almost comes to a stop.

Paul

[Norm Why]

"John McGaw" wrote

Thanks John. Maybe reassignment of connections is the answer.

Because of propagation delays, the group seems not to have seen my latest
post or

> "How-to-connect-case-cables-system-panel."
> https://assets.rockpapershotgun.com/images//2018/06/How-to-connect-case-cables-system-panel.jpg

[Paul]

No, Mr.Engineer, we don't have a problem.

https://i.postimg.cc/JnXsBwGc/LED-driver-circuit.gif

LED driver circuits tend to have current limiting
resistors in series with the power source. This
limits current flow to a safe value. In the
above picture, the rail was 5 volts, the resistor
330 ohms, and the series current flow cannot be
greater than 5V/330ohms = 15mA. Commonly available
indicator LEDs (T13/4 size) are rated for 20mA or more.
(I have one here rated for 100mA but that's more unusual.)

Fuses on motherboards, they cost money, and they're only
fitted in certain places. For example, industry practice
is to *not* put a fuse in the fan power path. On occasion,
the fan power path is burned out because of this.

Places that fuses are used:

1) Power pin on the parallel port connector.
2) Power pin on the PS/2 keyboard connectors.
3) Pairs of USB headers have a fuse (each 2x5 gets a fuse).

The very cheapest of motherboards, use a single fuse running
to 1,2,3. You might find mainstream motherboards with
five fuses or so.

It's also possible to use silicon fuses. These are an
eight pin DIP, sufficient to police current flow in
two circuits. When you're counting fuses, those would
be easy to miss in the analysis. A silicon fuse uses
a MOSFET to interrupt measured current flow. Silicon fuses
are a preferred solution inside a laptop (where there is
poor cooling available).

*******

There are various ways to do LED drivers.
"Series" or "Shunt".

This is series drive. The bipolar transistor can be a 2N2222
or the circuit can be driven by an open collector
74F06 or a 74F07 (as examples).

+ -
+5V ---Rlimit--- Anode Cathode -- C E ---- GND
LED B
|
transistor drive

In shunt drive, the circuit looks like this.

+ -
+5V ---Rlimit--+-- Anode Cathode ---- GND
| LED
+-----C E ------------ GND
B
|
transistor drive

In the shunt case, the Vce of the transistor, when
it is on, must be lower than Vfb of the LED. The LED
might be a 1.8V to 2.2V indicator. The transistor
Vce might be 0.4V. When the transistor in the shunt
case turns on, it shunts all the current (the 15mA)
to Ground. And since the transistor conducts until
Vce is 0.4V, the 0.4V across the LED is not sufficient
to cause it to light up. When the shunting path is ON,
the LED is OFF.

In the series case, the transistor Vce of 0.4V causes
minimal loss to the 5V supply. Now, the circuit sees
4.6V. It's delivered through the 330 ohm resistor, and
well less than 15mA flows through the LED. (That's
since the LED drops some voltage too, causing even
less current to flow through the 330ohm.)

The circuit above is generous enough to work
with red,yellow,green,blue LEDs. If the supply voltage
drops to 3.3V for some reason, then you might have a
problem getting a blue (or white) LED to light. The
ones I own at the moment run from 2.5V. 3.3V - 2.5 - 0.4
leaves only 0.4V across the 330 ohm resistor, or 1.2mA,
which doesn't always yield a good lighting experience.

The engineer designing the board, would not normally
assume blue LEDs are being used, as back when this
nonsense started, the LEDs were always red in color.

The design is resistant to most "expected" failure
modes. If the LED fails short, there's no fire.

It is possible, with enough shenanigans by the computer
builder, to connect the wrong things together and
damage something. The motherboard designer is under
no requirement to protect against "stupid stuff". All parts
cost money, and the motherboard makers are so profit-incented,
they will remove useful gold pins costing only a penny
or two apiece, to make more profit. And that's part of
the engineers job, is cost-reducing the product, balancing
"reasonable" feature sets against BOM (Bill Of Materials)
costs.

When they put "toys" on a motherboard, the marketing
department made them do that. Things like Port 80 displays,
Bitchin Betty (Winbond acoustic POST output chip), those
are gimmicks to sell more motherboards. And not essential
to running an OS with the door closed on the computer case.

Paul

[Paul]

What is that a picture of ?

It doesn't seem to be your GA-EP45-DS3L.

And the black wiring depicted, is a mistake when
doing builder work. You want classically colored
twisted pairs for each 2-pin thing needing
connections. It makes it easier to get the polarity
right on the first try (for LEDs).

No harm will come from hooking up a two pin pair the
wrong way around. I prefer the color code, because
it tells me "what they had in mind" when they wired it.

Antec in particular, has made mistakes on front panel
wiring more than once. The part I consider weird
about this, is when they do make a mistake, it's *never*
with the VCC pin, never a risk of smoke or fire.
How do they make these mistakes, without ever
involving a "hot" pin ? I could never figure that
part out. If you bring me an Antec case today,
I'll get out my ohmmeter and buzz out the wiring
harness, before using it.

Paul

[Norm Why]

It is not a picture of my GA-EP45-DS3L. It is other unnamed MOBO. It is a
picture my MX330-X front panel connections. It shows speaker(beep)
connections to pins 18, 14 on unknown MOBO. It would make a beep. But it
still does not solve my problem. I am preparing a customer care support
email to Cougar.

[Paul]

The usual miserable documentation. It says on the page
that they're Compucase, and this is a subbrand of theirs.

https://cougargaming.com/products/cases2/mx330/

https://cougargaming.com/fileadmin/downloads/USERS_MANUAL/MX330_X_usermanual.pdf

Doesn't have wiring harness info in there that I can see.

It's quite common, to not document the wiring actually.

*******

https://cougargaming.com/global/img/products/cases/mx330/06.png

2x5 - dual USB2 cable, hopefully shielded, industry standard pinout?
2x10 - dual USB3 cable, industry standard pinout?
2x5 - HDAudio for headphone and microphone jacks

HDD_LED, PWR_LED, PWR_SW, RST_SW, SPKRpair (five 1x2 or similar)
(may have labels on box-pins)

It's not clear from the photos, whether there is a SPKR.

I looked at the pictures of the case and cannot see any sign of it.

*******

USB2 cable - usually not a problem, follows industry standard with one keying
pin missing to help with keying it correctly. 9 working pins.
USB3 cable - usually not a problem, follows industry standard.
One keying pin missing, 19 working pins for two 9 pin interfaces
HDAudio - Can be a dogs breakfast of seemingly random wiring
and what might look like a jumper wire of some sort.
- computer cases with jack-sensing side contacts are
proper HDAudio setups. They use AC97 audio wiring
when they don't have the side contacts.

Front Panel - the pins are likely labeled
- if SPKR is a 1x4 and has lift-able tabs, you can
move the pins in the shell if the motherboard uses
a 1x2 header pair for SPKR.

HTH,
Paul

[Norm Why]

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.

[Norm Why]

[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.

[Paul]

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

[Norm Why]

Thanks Paul,

> 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"

PC knows what it is doing, it recycles in short time. Power comes on, fans
turn, RAM phase lights glow. The PC shuts down. A moment later, PC comes
back to life and cycles ad finitum.

This is an improvement over two RAM DIMMs inserted when cycles ends after ~3
tries. These tries take long because CPU is confused.

> 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.

Yep, it is

On page 48 of manual it says PCI graphics is the default. Without PCI
graphics one will never see the BIOS screen to change the default.

Wrote to Cougar.

[Norm Why]

>> From manual, here are the beep codes:
>>
>> Q: What do the beeps emitted during the POST mean?
>> A: The following Award BIOS beep code descriptions may help you identify
>> possible computer problems.
>> (For reference only.)
>> 1 short: System boots successfully
>> 2 short: CMOS setting error
>> 1 long, 1 short: Memory or motherboard error
>> 1 long, 2 short: Monitor or graphics card error
>> 1 long, 3 short: Keyboard error
>> 1 long, 9 short: BIOS ROM error
>> Continuous long beeps: Graphics card not inserted properly
>> Continuous short beeps: Power error

I switched the split connectors and still no beeps. The PC continues to
reboot ad nauseum with one RAM DIMM.
The split connectors are probably as labeled POWER LED+/-

How useful are these beep codes compared to bare metal diagnosis? Not much.
Given observations the beep codes can be predicted. The biggest issue is
RAM. Now it is less an issue.

On page 48 of manual it says PCI graphics is the default. Without PCI
graphics one will never see the BIOS screen to change the default.

I'll get a $5 PCI VGA adapter Monday and try again.

I cannot find any reference to beep/beeper with Cougar. Maybe it is
considered obsolete, and not installed

[Paul]

Norm Why wrote:

>> I need more info on what you mean by "Continuing CPU reboot cycle"
>
> PC knows what it is doing, it recycles in short time. Power comes on, fans
> turn, RAM phase lights glow. The PC shuts down. A moment later, PC comes
> back to life and cycles ad finitum.
>
> This is an improvement over two RAM DIMMs inserted when cycles ends after ~3
> tries. These tries take long because CPU is confused.

Normally, when a PC shuts down like that, it's caused by THERMTRIP.

That's a signal from the CPU, that it's past 100C or so. It's
supposed to cause the PC to go off and stay off. Clearance of
such a fault requires removal of +5VSB (switch off at the back
of the PC).

You could also cause a shutoff, by overloading +5VSB. The +5VSB
runs the USB ports and it's only rated at 2.5A or 3A or so. But
I would expect that fault to latch off too. Only if the BIOS
was set to a particular "recovery policy when power comes back"
in the BIOS, might such a flaw result in a loop.

I don't have a good explanation that precisely fits the
symptoms. There are things that can cause it, but generally
the loop should be broken.

Discovery that the processor is not one the BIOS supports,
might do that, but I don't recollect any recent results
to refer to. My P2B-S for example (year 2000), didn't know what a
Tualatin was, but I fixed that by loading a different
BIOS and doing a microcode patch. The "ID string" on the
BIOS screen was still wrong (called it a Pentium II or
something), but at least the CPU was "accepted" and boot
would finish.

>
> On page 48 of manual it says PCI graphics is the default. Without PCI
> graphics one will never see the BIOS screen to change the default.
>
> Wrote to Cougar.

That setting allows you to select which bus to examine first
for a video card. That motherboard has only PCI bus and PCI Express
bus, and has no AGP bus. And since it's a P45, at a guess
there are no graphics. P45, the P likely stands for Performance,
and a gamer would use their own video card. Other letters
they might use, would include G45 or Q45, and SKUs like
that would have a GPU in the Northbridge. Q45 is a designation
used for boards including the Intel Management Engine (ME)
for remote control of the PC via NIC packets. (The refurb Optiplex
I've got, has a Q45.)

Even without using the setting, if there is no PCI Express card,
it should still examine the PCI bus when it gets around to it.

At the current time, with the looping behavior you've got, I
don't think it is getting near enough to the conditional beep
codes, to issue any beep codes. Something is wrong in the
(register-only based) BIOS code. And whether a hardware signal
is taking it out of action, or the BIOS code itself is having
a wobble for itself and resetting, we're just not getting to
the beep code section.

Even if you had a PCI Port 80 card, the Port 80 display might not
update with new digit values before the looping hits it. Being
able to monitor the voltages, might give a hint if the problem
is power related, but you need the right electronics trash
to catch stuff like that.

*******

You can see there is something that isn't quite right
about these boards. But, what is it ? For one guy, it was
the switch to a PCI graphics card that allowed him to see
the screen and get on with it. Any old PCI graphics card
would do for this - there is no need for anything fancy.
like a 6200 PCI or a FX5200 PCI would be candidates.
The 6200 and the FX5200 PCI were available for quite
a while before they stopped making them. (The 6200 was
notable, in that it had native support for more
than one bus standard, which made it cheaper to make
for the manufacturers. It didn't have to be as expensive
as a bridged board.)

https://forums.tweaktown.com/gigabyte/28049-gigabyte-motherboard-ga-ep45-ds3-reboot-loop-2.html

Does yours loop with both RAM and video removed ?

Don't forget to remove all power, before adding
or removing those...

Paul

[Paul]

Norm Why wrote:

>
> I switched the split connectors and still no beeps. The PC continues to
> reboot ad nauseum with one RAM DIMM.
> The split connectors are probably as labeled POWER LED+/-
>
> How useful are these beep codes compared to bare metal diagnosis? Not much.
> Given observations the beep codes can be predicted. The biggest issue is
> RAM. Now it is less an issue.
>
> On page 48 of manual it says PCI graphics is the default. Without PCI
> graphics one will never see the BIOS screen to change the default.
>
> I'll get a $5 PCI VGA adapter Monday and try again.
>
> I cannot find any reference to beep/beeper with Cougar. Maybe it is
> considered obsolete, and not installed

There are some Antecs that are "too fancy to have speakers".
And they are missing the SPKR used for beeping.

You have to talk to a local computer shop, to see if they
stock either a piezo (one inch black disc with two wires),
or have a SPKR (2.5" speaker perhaps) available. Some small
shops carry things such as temporary reset and power buttons,
and also carry something you can use as a SPKR when your case
is missing one.

The reviews for this one are not good, so this link is
purely for the picture of the item.

https://www.amazon.ca/Baoblaze-Motherboard-Internal-Speaker-Buzzer/dp/B0798PXK3L

This is the more conventional kind.

https://en.wikipedia.org/wiki/PC_speaker#/media/File:PC-Speaker_IMG_8161_%28cropped%29.JPG

So far, all my cases here have the SPKR kind with the magnet.

My Optiplex refurb is neat, in that it has a tiny speaker
which also plays regular LineOut audio. But the sound
is tinny. It helps for situations where you forgot
to plug in computer speakers to LineOut, and then the
tiny speaker assembly just above the hard drives,
makes noise. Rather than being a nice 2W analog power
amp driving the speaker, the speaker is more likely
to be a 32 ohm unamplified speaker, and the sound level
is equivalent to "headphones" in magnitude. You can
barely hear it, and it's basically a reminder to
plug in the real audio output speakers.

Paul