Re: How to PostMortem?
From: Jean-David Beyer (jdbeyer_at_exit109.com)
Date: 04/22/04
- Next message: Hiram Barber: "Move Redhat AS 2.1 from Machine A to Machine B"
- Previous message: Sonoman: "flash drive will not mount"
- In reply to: w_tom: "Re: How to PostMortem?"
- Next in thread: w_tom: "Re: How to PostMortem?"
- Reply: w_tom: "Re: How to PostMortem?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Wed, 21 Apr 2004 23:34:22 -0400
w_tom wrote (in part):
> If your three UPSes have serious surge protectors, then you
> can provide the numerical specifications that describe how
> well they perform for each type of surge. I keep asking for
> and no one has ever been able to provide specs to prove this
> protection. Furthermore how does that UPS earth the typically
> destructive transient when it is, virtually, not connected to
> earth.
I cannot give you the numbers you request, but the manufacturer implies
that their effectiveness is high with the following words:
"Surge energy rating 880 joules
Filtering Full time multi-pole noise filtering: 0.3% IEEE
let-through: zero clamping time: meets UL 1449"
"American Power Conversion's Equipment Protection Policy:
...
For customers that meet the qualifications and conditions set forth in
this policy, APC will provide reimbursements (cost of repair or fair
market value) during the period limits and up to the dollar limits
stated as follows:
[for the units I have]
Dollar Limit $25,000
Period Lifetime"
>
> You are confusing safety ground with earthing. To be
> effective, the UPS must make a less than 10 foot connection to
> EARTH ground.
Of course, that is impossible in a residential situation. The best that
can be managed is using the green wire both as a safety ground and an
earthing connection. The electrical code around here requires that the
power panel be connected to earth as close as possible using two
separate connections. In my case, one uses a #12 guage wire less than 5
feet long going to a stake at least 8 feet long into the ground. The
other, also #12 guage wire, is probably 10 to 15 feet long and goes to
the cold water supply pipe just as it enters the house.
The panel contains two circuit breakers that supply the two computer
power sockets. Nothing else is connected to those circuit breakers.
3-wire #12 guage cable goes from the breakers to the outlets. Each of
those cables is probably 10 to 15 feet long.
I am not as concerned with common mode noise as you seem to be, it would
be generated mainly in the connection to earth from the power panel to
the ground stake less than 5 electrical feet away. Any voltage generated
in that 5 foot wire would affect both outlets the same, so while the
computers might be going up and down in potential due to common mode
noise in the 5-foot wire, the difference between them would be negligeable.
> That means UPS must be part of or at breaker
> box. Building wide UPSes are effective because a short earth
> ground is part of the installation.
Out of the question in a residential situation.
> But plug-in UPSes are not
> located and connected less than 10 foot from earth ground.
> Earth ground - not safety ground - is essential for effective
> protection. That fault light reported a safety problem - and
> could never report the existence of earth ground. No earth
> ground means no effective protection.
>
> Your description of a green wire connection back to breaker
> box is an earthing path? But if that green wire does carry
> the destructive 'direct strike' surge, then green wire only
> induces that transient on all other adjacent wires.
The transient induced by the common-mode connection, right? And that is
the wire from the power panel through a wire less than 5 feet long from
the power panel to the earthing ground stake.
> Now we
> have additional induced transients throughout the building as
> well as the direct strike. Not only must the connection from
> each utility wire to earth ground be less than 10 feet. It
> must also have no sharp bends, no splices, and not be bundled
> with other non-earthing wires. Destructive transient must be
> dumped into earth before that transient can enter a building.
> Well proven concept from before WWII. Green wire ground
> violates all three criteria for effective protection.
>
> This problem with a green wire safety ground is why your UPS
> manufacture avoids the entire discussion.
My manufacturer does discuss this:
http://www.apcc.com/tools/mytools/index.cfm?action=search&category=whitepaper
> Notice that
> discussion about earthing never comes from the manufacture of
> ineffective protectors. But real world (serious)
> manufacturers discuss earthing extensively:
> http://www.polyphaser.com/ppc_technical.asp
> http://www.polyphaser.com/ppc_pen_home.asp
>
> Running a dedicated circuit to computers is nice. But it
> does not do anything worthwhile for noise or transient
> protection - from a computer's perspective. Nothing in the
> house creates noise or transients that adversely effect a
> computer. Computer power supplies are some of the most
> resilient devices in the house. So resilient that this UPS
> creates 120 VAC in battery backup mode by outputting two 200
> volt square waves with a 280 volt spike between those 200 volt
> waves.
My UPSs produce sine wave output.
Since the computer power supplies are switching type (the big one with
power factor control), they can probably tolerate pretty bad input
waveforms.
> An example of protection cited by an APC product:
> SURGE PROTECTION AND FILTERING
> ...
> Normal mode clamping response time 0 ns, instantaneous
> Normal mode surge voltage let through <5% of test peak voltage
> when subjected to IEEE 587 Cat. A 6kVA test
> Normal mode noise suppression Full time EMI/RFI
> filtering
> Modem/10Base-T/100Base-Tx network cable port single line
> (2 wire, RJ11) or network (UTP, RJ45) compatible jacks
>
> Where is the common mode protection? It is not even
> claimed.
>
> If your power supply is inferior, then why spend big bucks
> on a UPS? An inferior supply may also damage other computer
> components since essential functions are missing. Functions
> that were defacto standard 30 years ago and specifically
> demanded in Intel specs.
My power supply claims to follow Intel specs, in particular, SSI EPS12V
Power Supply Design Guide (mine is rated at 660 Watts output). It is not
meant to be an inferior power supply. ;-)
> Functions that a UPS does not even
> provide, but are necessary for a computer. Before even
> considering a UPS, one must first use an acceptable power
> supply. That means either a brand name computer or the clone
> assembler must have a long list of specification from the
> power supply manufacturer. UPS cannot compensate for these
> missing and essential power supply functions.
Well, I built this computer myself, since I could not find one as I
wanted it. I do have the SSI EPS12V specification that the P.S.
manufacturer claims to meet. The maker of the mother board just says the
P.S. needs to meet an earlier version of the SSI standard.
>
> Obviously, this is still well beyond the OP's problem.
Quite.
> Of
> those five basic power problems - blackouts, brownouts, noise,
> harmonics, and surges - the plug-in UPS really only claims to
> address the first two.
Actually, my UPS claims to address noise (probably including harmonics),
and surges as well. When it detects too much noise on the line, even if
the line is within voltage specification, it switches to battery.
> Yes, stepping up and down the
> transformer does put less strain on the battery and does
> provide supplementary voltage adjustments to a power supply.
> Nice. But the OP must get basic, essential tools to first
> provide basic facts - to get numbers to learn where his
> failure is.
No question about that.
> When he provides facts, only then can we even
> discuss power supply as reason for his problem. Procedures to
> diagnosis his failure provided in an earlier post.
-- .~. Jean-David Beyer Registered Linux User 85642. /V\ Registered Machine 241939. /( )\ Shrewsbury, New Jersey http://counter.li.org ^^-^^ 22:35:00 up 20 days, 18:51, 7 users, load average: 4.31, 4.14, 4.05
- Next message: Hiram Barber: "Move Redhat AS 2.1 from Machine A to Machine B"
- Previous message: Sonoman: "flash drive will not mount"
- In reply to: w_tom: "Re: How to PostMortem?"
- Next in thread: w_tom: "Re: How to PostMortem?"
- Reply: w_tom: "Re: How to PostMortem?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
