Experience is relative, but I would say we have been fighting the same
issue for over a year now. between in-house burn tests and tests at
various NRTL labs, we have burned over 40 shelves varying in height from
1 RU to 6 RU. We have discussed this issue several times with Bellcore
and they concur there is a problem, but insist they are trying to
harmonize with the ANSI T1.319 1995 and therefore cannot justifiably
change the test method. Here is basically the history, the problem and
suggested solutions.
1.) Until 1995 the TR-NWT-000063 did not have a burn test designed for
shelves or small systems. The test was designed for cabinets and fully
loaded frames. As a result, Bellcore decided to incorporate a shelf
level burn test "loosely" based on ANSI T1.319 1995 into the new
GR-63-CORE. At this point they ignored a circuit pack level test which
is basically what ANSI T1.319 1995 is all about.
2.) The intention of ANSI T1.319 1995 was to obtain supplemental
information about the burning characteristics of an assembled circuit
pack that would not be evident based on a flame database, needle flame
tests, etc.. The idea was to place lots of the circuit packs in a
configuration independent of the host system and see what the affect was
of the placement and quantity of various components on the overall burn
characteristics of the cards. Placing lots of 94V-0 and 94V-1 polymeric
components together on a circuit board and applying a flame source is
totally different than doing a needle flame test on a single component.
3.) ANSI T1.319 1995 defines a flow rate and burner configuration for
applying the flame. It was based on statistical data obtained from
fully ignited 8"x12" circuit cards loaded with a variety of 94V-0 and
94V-1 components supplied by major telco providers at the time. The
study was conducted in conjunction with Underwriters Labs and lasted
about 30 months. As a result the flow rate in Table 1 and the line
burner in Section 5.2.4 of the ANSI document were modeled to duplicate
the characteristics of a fully engulfed 8"x12" loaded circuit pack. The
only pass/fail criteria was a heat release rate.
4.) Bellcore then adopted this ANSI standard and applied it in
"complete" shelves and systems less than 36" in height, and created a
several new pass/fail criteria (R4-20). Unfortunately for vendors these
days, Bellcore assumed that they were the only ones that would actually
perform testing to GR-63-CORE where the results would be reviewed by
telco providers. Bellcore provided "Technical Audits" and if they felt
a failure mode of a system was not a substantial risk then they would
make that kind of statement in their Technical Audit and the RBOC's
would not be concerned because Bellcore said it was probably OK.
5.) What Bellcore did not realize was that independent labs would start
performing these tests due to the vast number of new products, long
turnaround times and very high test costs at Bellcore. As a result,
most/many burn tests are performed by labs other than Bellcore who do
not have the leeway to use good engineering judgment when evaluating the
burn tests. They can only state that a system complies with or does not
comply with the requirements or objectives. Due to attrition, layoffs
and shortages of manpower, the RBOC's, GTE and other independents do not
have the time or desire to interpret the flame data for products as
stated in the opening paragraph of section 4.2 of GR-63-CORE. As a
result, they typically want a report that simply states "complies with
ALL requirements". If they don't get that, our experience has been that
they typically will not buy! There are some exceptions such as Bell
Atlantic North and South who have very experienced people who are
willing to work with a vendor on this subject. They realize that
mitigating certain failure modes creates bigger problems.
6.) Vendors are building smaller systems (1-2 RU high) with circuit
boards that are substantially smaller than 8"x12" with far fewer
combustible components (due to surface mount technology) and the test
labs must still apply a flame that is ridiculous based on the intent of
the ANSI T1.319 1995. As a result, the only way to pass all criteria is
to enclose the shelf with solid metal or a fine mesh perforated
material. This however causes a much worse problem since the heat from
normal operation is trapped in the shelf, which will eventually result
in premature failure of the components. In addition, the burn
characteristics of horizontal circuit boards were not studied in the
ANSI T1.319 yet Bellcore GR-63-CORE does not address this "substantial"
difference between a horizontal board and a vertical board and you must
test both the same. As a side note, horizontal boards burn much less
than vertical.
7.) If I were Bellcore, I would address this issue five ways:
a) Make the requirements in R4-20 objectives and add wording stating
that these criteria should be used carefully by the lab and customer to
analyze the overall flammability risk, and that shelf and circuit pack
size should be considered while making this assessment based on the
application of the EUT and the resultant affects of mitigating a
specific failure mode.
An example would be a 2 RU high shelf that is densely populated with
2"x 8" circuit cards. If flames extend beyond the vertical confines for
more than 30 seconds after the 3.5 minute point, that is technically a
failure. However, the flame from the burner is still roughly 8 " tall.
The only way to meet this criteria is to place a solid top or mesh
screen over the shelf to smash the flame horizontally inside the shelf.
The problem is that this substantially reduces airflow and increases
internal system temperatures (reduces reliability).
The flame used is far too much since the methane flow rate in the
standard is based on a 8"x12" card (96 square inches) and the card under
test is 16 square inches which is roughly 18% of the flame mass. It
also increases the cost of the system to the end user as mesh materials
are fairly expensive. And lastly by spreading the flame horizontally,
the equipment becomes much hotter during the burn test, the system
basically self consumes due to the heat pocket and creates several
orders of magnitude more smoke which is more harmful in a C.O than the
fire itself. So by mitigating an unrealistic failure mode, reliability
during normal operation is reduced and if a fire did occur the damage
would be far worse due to smoke.
b) I would provide an option that stated that if a system is Listed to
UL 1459 or UL 1950 3rd Edition and it meets the criteria stated in the
material analysis section 4.2.3 of GR-63-CORE and ANSI T1.307 1997 then
requirement R4-20 is exempt.
c) I would prorate the flow rate of the methane burner to the
proportion of the circuit board which is being simulated compared to a
96 square inch circuit pack.
d) I would incorporate the ANSI T1.319 1997 document as a way to show
compliance of circuit packs that are provided for existing systems.
e) I would make provisions for different line burner lengths so that
circuit board/ system formats that have depths less than 6.5 inches
(such T400 or T200) do not have a flame burning in front of the
faceplate due to the line burner.
Jim
James Wiese
Regulatory Compliance Engineer
ADTRAN, Inc.
205-963-8431
205-963-8250 FAX
[email protected]
>----------
>From: Dwight Hunnicutt[SMTP:[email protected]]
>Sent: Friday, March 13, 1998 6:35 PM
>To: [email protected]
>Subject: NEBS 5.2.3 Fire Test
>
>
> NEBS Group-
>
>I'm wondering if any of you have experience with
>the fire test of GR-63-CORE 5.2.3, and the grisly
>outcomes of misapplying the test to certain equipment.
>
>Specifically, I question the applicability of this
>test to a "pizza-box" type unit (1 RU height), that
>can be rack, wall, or desktop mounted. Assume the
>unit is CPE(customer premises equipment),and Listed
>to UL1950. It would not (normally) go in a C.O., but
>only an equipment closet at a business.
>
>My understanding is that the fire test of 5.2.3 was
>designed for large cabinet racks of cards, and tries
>to determine the relationship of one's equipment to
>the outcome of a severe fire in an equipment room or
>C.O., whether or not the fire started in one's own
>equipment.
>
>For a unit with top openings complying with UL1950,
>the flames from the 5.2.3 test burner would just shoot
>up through these top openings and burn the two test
>cards in the test setup, thus failing the fire test.
>Same or worse results should occur if the unit was
>wall-mounted vertically for the test, since the top
>openings would now be side openings, and the flames
>would just lick up the sides.
>
>Question: Is there a consensus that this 5.2.3 fire
>test does not apply to the "pizza-box" CPE type unit?
>Or can the "shelf-level" test be interpreted to apply
>to this type of unit?
>
>Any of you have any success convincing a customer of
>the "misapplicability" of this test to your equipment,
>even though the customer insists it does?
>
>If it does apply, do any of you have experience to say
>that you agree that TOP openings complying with UL1950
>BOTTOM opening requirements would be small enough to
>contain the test burner flames (sufficient to pass the
>fire test)?
>
>Thanks in advance for your comments.
>
>
> RCIC - http://www.rcic.com
> Regulatory Compliance Information Center
>
>
>
>
