DSU/CSU Electrical Protection

Hello
We are having problems with the internal cisco DSU/CSU cards on cisco
1721s as they are dying. I'm guessing this is due to energy surges,
such as lighting and induced voltages.

I cannot seem to find any optical isolation equipment, which would
work. Opto isolation is the best, in my experience. Any
recommendations on opto isolation are appreciated.

What I have found, at www.edcosurge.com/products/telecom/lcpd.asp
is RJ45 surge protection. It operates (clamps) at 8, 15 and 30 volts.
Any recommendations on which volatage to select? Other circuits or
devices for DSU/CSU protection?

The circuits are Verizon 56kbps Frame Relay, if that matters....

Any discussion or recommendations are most welcome.


James

 

If your problem is due to environmental transients, then
nothing is going to stop block, or absorb those transients.
However we routinely avoid such damage by doing what Ben
Franklin did in 1752 to protect church steeples. Concepts are
that simple and that well proven.

Destructive transients are current sources. Therefore
voltage will increase, as necessary, to maintain that current
flow. Miles of non-conductive air did not stop that
transient. Is a silly quarter inch of opto-isolation going to
do what miles of air could not? That is exactly what myth
purveyors would have you believe when they promote their
ineffective and grossly overpriced protectors.

The concepts are posted in previous discussions:
"RJ-11 line protection?" on 31 Dec 2003 in pdx.computing at
http://tinyurl.com/2hl53
"strange problem after power surge/thunderstorm" in
comp.dcom.modems on 31 Mar 2003 at
http://tinyurl.com/2gumt
"surge protectors trip Ground Fault Breaker" in
alt.engineering.electrical on 28 Aug 2003 at
http://tinyurl.com/238bb

If running a serious communication facility, then your
building should be properly wired for effective transient
protection. That means earthing all incoming wires per post
1990 NEC requirements - all wires connect to single point
earth ground either by direct hardwire or via a protector.
That also means exceeding those NEC requirements. Concepts of
earthing discussed among engineers in the newsgroup
misc.rural:
Storm and Lightning damage in the country 28 Jul 2002
Lightning Nightmares!! 10 Aug 2002
http://tinyurl.com/ghgv and http://tinyurl.com/ghgm

Destructive transients occur typically once every eight
years. The frequency of your problems suggest that your
building even violates NEC requirements for human safety - a
problem more serious than lightning. Transistors being
damaged because the building even may have human safety
issues. You have much to read. Both human safety and
transistor safety issues involve the single point earth
ground.

 

Fast acting devices, in the order of 10s or hundreds of nano-seconds,
sense voltage increase to trigger. By the time the increase current is
actually flowing at a given point, it's too lated to protect the
communications interfaces, even thougth the duration of the increase
current flow is of the order of nano seconds.

No body is talking about stopping the brunt of a lighting induced surge.

What we are talking about is filtering, clamping and diverting the
increased energy under the curve for nano seconds that is at a higer
voltage. If the voltage never surges, then the device only draws the
current it needs or can handle.


Um, we seem to be speaking different dialects here. Our facilities
are all code compliant, and the incomming signal wire is burried
(mostly) but it is the property of Verizon.
I would only use opto isolation inside where the engergy surges are
actually occuring in and on the skin of the wire (i.e. inside of the
plastic insulation that wraps the wire.

I do appreciate your links, but, I assure you code compliance, even
the useful things in NEC 2004 are not at issue. We are in the
lightning capital of the world (Tampabay) and our customer has many
facilities, some that criss cross swamps and forest, so they are
natural settings for all sorts of induced energy surges. And yes we
have areas where the ground is a very poor conductor, but, we 'ohn'
out all of our ground resistances, and install mitigation equipment,
like triads, as necessary.

This would be useful, except the 'gas-discharge' surge protectors used
by telcos are too slow, the clamp in the order of milli seconds.

This article is irrelevant. The POTS (plain old telephone service)
allows 90 volts to activate the ringer, so if you clamped at 95 volts
you still allow a large amount of energy into the POTS interface.
The 4 wire or 2 wire copper circuit thats in terminated in the Telcos
RJ4ks connector surely does not swing 90Vp-p.

Any idea what the maximum voltage swing is on a Frame RElay circuit
on 4 wire copper? That would be useful information.

OK, now you are talking about energy surges on the power input feed.
This is not the problem. The cisco routers are fine, we keep loosing
the internal 56K dsu/csu cards. Therefore the surges are comming in on
the 4-wire copper data circuit. Besides the routers are behing UPSes
that have 'ferro-resonant' transformers. Ferro-Resonant tranformers do
indeed block the energy surges on power feeds. You should read up on
those devices.......

When I build datacenters, all power feeds are behind 1:1
ferro-resonant transformers for protections. This is not the problem.

Dude, the problem is not on the power feed. I routinely consult to
power utilities on power quality issues.....

This statistic is utter BULLSHIT. I can show you facilities that are
lightning proned, multiple times a year, no matter what you do. Other
facilities less than a mile away, never get hit. It's due to a complex
set of environmental issue, mostly realted to conductivity of the
soils and the surrounding structures and trees...

You, are full of shit!

 

You are making wild assumptions. For example code
compliance means the building does not kill humans. Code
compliant building can still kill transistors. Furthermore,
it does not matter whether wire is buried or overhead - as you
should have well known. Either way, incoming wire can conduct
destructive surges directly into equipment as was well
understood long before WWII.

Your technical background is so based upon urban myths that
you even believed that outright distortion about gas discharge
tubes (GDTs) taking milliseconds to response. I thought that
fairy tale had died long ago. If true, then those 30
microsecond transients were never diverted by the GDT. So how
did GDTs provide protection routinely so many decades ago?
Funny thing about numbers. They expose junk science such as
GDTs take milliseconds to respond.

Of course, you did not suffer damage because a transient
incoming on telephone line had no outgoing path - therefore
damage could not exist. You do remember your elementary
school science lesson. Without both an incoming and outgoing
path, then no electric current - and therefore no damage. So
what was the incoming and outgoing electrical path so
necessary for damage to occur? Did you first ask that
question? Or did you assume electricity comes in on phone
line, damages line card, then stops? In the meantime, telco
would have already installed effective protection on its
incoming wires. (Verizon does not install wires in Tampa Bay
- which you obviously knew). Protection that would be only as
effective as the earth ground your building has provided.
Earthing that is well beyond what any code requirement calls
for. You did know about this single point earth ground
requirement - right? This is also why electronics connected
to phone lines are so easily damaged by AC electric
transients.

Then there is the silly myth that ferro-reasonant UPSes will
stop transients. Did you notice the breakdown voltage?
Cisco power supply already provides protection meaning a
plug-in UPS provided no additional protection - once basic
electric principles and the numerical specifications are
learned.

You have damage because you have not learned about effective
protection. You actually think a filter is going to 'blunt'
the energy of a transient? Only in your dreams. First apply
real world numbers. Its called breakdown voltage which makes
your speculations irrelevant. Anything that the optocoupler
or filter was going to accomplish is available in that Cisco
interface card. Why do you think the protectors you seek are
not readily available?

You want better protection? Put knots in the wires. That
too is a classic urban myth.

A plug-in UPS only provides the transient with more
destructive paths through your Cisco. That UPS does not even
claim to protect from the type of transient that typically
damages electronics. You did not bother to read what has been
proven by generations of experience, is well published in IEEE
publications, and would not be understood by the mentally meek
among us who use "Dude" to mask their ignorance. This
sentence completely justified by your insults as well as your
responses that are so classic of scam artist promotions.

If you want help - and your technically naive responses
explain why you had damage - then start by being civil and by
learning about things you only have "Dude" knowledge of -
better known as junk science. Yes, in your responses were a
few technically correct gems alongside old wive's tales.

You are trained by so many myths as to even think telcos
used gas discharge tubes? GDTs were obsoleted by a low
capacitance type of MOV which is turn was obsoleted by a
semiconductor type device. Anyone with telco knowledge would
have known this. You have much to learn and have been brain
washed by too many junk science legends.

You are having problems with the Cisco because you do not
have effective protection - a problem directly traceable to
human ignorance and some mythical belief that those plug-in
UPSes provide that protection. Even worse, you actually think
that a code compliant building provides effective transistor
protection? Where did that wild assumption come from? We still
build new buildings as if the transistor did not exist.

You cannot find that protector because it does not exist for
good reasons. Same protection already exists inside the
Cisco. You have even fallen for the classic myth that a
plug-in UPS provided protection - when even the manufacturer
does not make that claim. Post the long list of
manufacturer's numerical specs if you think otherwise. But
good luck finding those specs. Only one of us does not
worship urban myths and then insult the other. You want
help. Then learn how much you don't know and then be civil.
Even the typical average of one destructive transient every
eight years was accurate - had you only read the cited
material rather than declare you have all the answers.

 

w_tom wrote:

Tom,

You have a really bad attitude. I'm looking for help,
not somebody who still thinks the problem in in the power circuitry.
It's not in the power circuits, it's in the data circuits or the
DSU/CSU design that cisco uses on the internal cards.
If it was on the power feed circuits, we'd have problems with the
cisco routers, there is no problem there. We have put power quality
meters all over the power feeds... NO problems.

I not looking for a lecture (and you are wrong and awefully
presumptive, that why I said you are full of shit, you presume to know
the environment and you don't. You have not listend to the statements
either. Verizon has had equipment blowouts too, even with the precious
gas-discharge device you think are wonderful and fast acting. I' sure
there are gas discharge protection that is better, than the vintage
wwII stuff you seem to have kindred emotions with, but, it does not
clamp, act, divert or filter in a fast enough time to be useful for
many transients. You need devices that act between 10 and 100 nano
seconds. I use devices such as spectrum and network analzers for Rf
problems. I was hoping somebody would suggest a test set, such as
ameritech, that would catch this problem, so I do not have to lugg
100,000.00 worth of lab gear out into the field....

If you are not going to make useful suggestions, then please find some
body else to harrass. POWER QUALITY IS NOT THE PROBOLEM! It's
intermittant energy on the copper 4-wire circuit. Also, I have already
found lots of information that suggests the problem is indeed in the
cisco dsu/csu design on those cards. This is reinforced by the fact
that now, the dsu/csu devices are locking up at other facilities that
are not on extended demarcs. These internal dsu/csu that cisco has on
the card are indeed a poor design, and probable left over form one of
the many companies that cisco has pruchased.

I now have intermittant failure, the circuit quits working and the rx
light goes out at 4 out of 22 sites with cisco 1721 all with internal
dsu/csu cards from cisco. As soon as a technician cycles power on the
cisco 1721, it starts woring again. One dsu/csu card has failed
completely. 2/4 intermittant sites have the VERISON smart jack within
our facility, so the problem(s) are at least partially due to
something not related to the extended demarc issue.

Please try to focus on what I'm saying and be helpful,
non-presumptive, and polite, or go away. Your choice.

Your wasting a lot of bandwidth on power feed issues that are
irrelevant to by issues.....

PS you are right about one thing, I did not discuss filter capacitors
with ferro-resonant transformers, as they are usually understood to be
part of circuit protection, that is built into the better ups. And
that's the point, WE should not assume to know about thing that the
author has not written about in the email. It's not a power quality
problem. It's a problem with trasients on the copper data circuit,
and a poor internal dsu/csu design/implementation on cisco's part.

What I was looking for from the readers of this list is helpful
suggestions, such as:
http://www.bomara.com/Telebyte/catalog/products/22t1.htm

Or any other protection/monitoring/logging equipment to document the
problem. OR better yet IOS commands with will capture, log, and
forwared electrical signals or layer 1 problems so as to document the
short comings of Verizon.

It's been suggested that Paradyne's top of the line external dsu/csu
has some excellent circuitry for capturing data on transient
electrical problems...

PS, I have friends in cisco engineering that will eventually come
clean on these issues, after I email then waveforms of the culprit...

Peace,
James

 

The tone of my post changed only after you starting
insulting AND posting classic urban myths as if fact. Your
ferro-reasonant, plug-in UPSes as effective transient
protection is classic myth. If the problem is on all Cisco
cards, then you have plenty of 'dead bodies' to identify the
incoming and outgoing path of failure. We learned thirty
years ago why phone line damage is most often incoming on AC
mains. But then as engineers, we first had to see facts
opinions could exist.

You started the attitude "Dude". Now decide whether you
want learn or want to promote more urban myths. Line cards
must be designed to withstand 1000 volt transients. That is
even FCC requirement (I believe Part 68). Those Cisco cards
should not be damaged IF the building is properly constructed
and wired.

Meanwhile, describe how the extended demarcs are installed -
location, distance from single point earth ground, how and
what earthing wire is routed through, distance to Cisco line
cards, etc. Everything, including the wire, is an electronic
component of that circuit.

First place one starts when suffering transient damage is
the single point earth ground. Measurements are not
sufficient. Visual inspection is required. As Polyphaser
demonstrates, if all incoming utilities don't share the same
single point earth ground, then damage results to electronics
with sufficient internal protection. The electronics becomes
a preferred path of those transients. Again, read the
citations before responding. This Polyphaser app note
demonstrates how transients enter on buried phone line.
Remember that buried line that you claimed would not carry
destructive transients? Experts say otherwise:
http://www.polyphaser.com/ppc_PEN1028.asp

All equipment connected to phone lines has onboard
protection. Protection that is sufficient but will be
overwhelmed when a building is not constructed / enhanced to
meet minimally acceptable requirements. Those Cisco cards are
not routinely failing for other customers. Building wide
problems may be insolvable without first learning about what
was posted in those previous posts. That means you cannot
dismiss the concepts until you have first read what was
learned and demonstrated by generations of theory, papers, and
experience. Polyphaser also demonstrates the problem in
simple terms:
http://www.polyphaser.com/ppc_PEN1024.asp

If you thought a code compliant building accomplished this,
they you are now ready for an expensive lesson in transient
protection and the all so critical single point earth ground.

If the problem is incoming on copper data circuits, then
where is the outgoing path? No outgoing path means no damage.
And how is that transient being completely ignored by the
'telco provided' protection that exists where phone lines
enter your building? Why could you have destructive
transients on phone line if those lines already have telco
installed protection? It was answered previously. Maybe you
could provide a better reason. Visual inspection provides
important facts.

 

The usual advice from experience (boilerplate with slight editing for
your situation):
- Consult Verizon to run full diagnostics/ send a tech. There may be a
wiring fault which is killing the equipment.
- Consult Verizon to guarantee that there's no chance of ringing voltage
on the pair(s).
- Consult Verizon to guarantee that they're not performing ANY overnight
automatic diagnostic testing on the pair(s).
- Consult your network installation contractors about this issue. They
may be able to test for a wiring fault sooner than Verizon.
- Consult Cisco about Cisco-approved DSU/CSU protection products.
- Make sure Cisco is replacing the failing equipment under warranty, and
with "advance RMA" until the problem is resolved.
- Test for voltage & current transients, ground faults and AC
overvoltage, especially during building heating/ventilation/air
conditioning cycling and lighting transitions.
- Verify the 1721 power supply is solid (low noise, not fading, spiking,
no leakage current, no imposed AC or switching noise).
- Verify the wiring in the building, and in the equipment closet (no
hot/neutral swap, valid ground, <0.1V between ground points throughout
building).
- If your equipment is plugged into a GFI/GFCI, try a non-GFI outlet.
GFIs fail over time, I found out.
- Try an Edco device. Consult Verizon to specify 8V, 15V, or 30V
protection, or just measure the voltage across your telco pair(s), and
get the Edco device with the next higher voltage rating. If you measure
6V, get the 8V protection. Make very sure this device is grounded to a
valid non-structural building ground (all copper water pipe), same as
all the other network equipment.
- Make sure the cleaning crew isn't plugging their 12-horsepower vacuums
and floor polishers into the same circuit as your network gear.

Normally, telco-connected gear is well protected against the usual
insults.

And finally, don't exclude the possibility that Cisco may have a chronic
problem with this module, especially if it's a low-demand device.

Good luck.
L

 

They are getting the package distroyed in one location, twice by
lightning. They are trying,but I'm going to get some test gear and do
some monitoring on the circuits myself. Any recommendations on
test_set? Ameritech seems to be the most recommended for deep analysis..
????

All of these are Verizons domain. I'm looking for tests and test
equipment I can use, in addition to Verizon.

power feed transients are not the problem. The routers stay up
and never fail. After a storm the internal dsu (WIC-1dsu-56k4) says
that serial is up, line protocol is down. Cycle power on the router
(and hence on the internal dsu/csu) and serial comes up, line protocol
{Frame Relay} comes up, and everything is fine. It could be the
embedded RTOS (i.e. ios) filling up a error file buffer, but, cisco
uses circuit writes to avoid locking up a device on error messages....

Any IOS suggestions or things to check?
Searching the net for "WIC-1dsu-56k4" reveals lots of refurbished
cards. NOt many used cards. This and other data lead me to believe
it a less than robust internal dsu/csu implementation. They have
deprecated the sister t1 card, which it's replacement now does
single channel 64/56k with the internal dsu/csu.....

Power is not the problem and I have put power quality metering/logging
equipment on site for days, and during lightning storms. No problems,
transients or other isses that the equipment does not handle.

Comm circuits in on a separate grounding system. It does not make much
sense to ground signaling devices to the same grounds that 480v 3
phase systems use, in my experience...... 3 phase power protections
schemes are much more robust than 110V single phase stuff most
networkers uses......

Now you are talking. We have devices coming form LEA, Edco, and
Telebyte... Any other suggestions, before I desing something myself?

see above

Yep, as soon as I have some log files and waveforms, I'm going to
email them to some 'assholes' I know personnally inside of cisco.
Cisco use to be a premiere technical company. Now they are like
microsoft, not owning up to technical flaws.

Oh well, my customer is now letting me design networking system from
the ground floor up. Cisco was pushed on us, an industrial
environment, by the MIS dir at a gov agnecy. Now I have about 15
different issue that let me plan on system enchancements without using
cisco gear.....

Now I'm happy, and did not even have to get into extended temperature
issue, as those are coming.....

sincerly,
James

 

Since you mention actual lightning damage (that wasn't clear before),
the best approach is a spark gap+snubber+fuse chain with a good ground.
It once was usual for the telco to spark-gap and fuse every pair
coming into a building, to a really good ground. This should be
inspected and tested. This knocks down the likely voltage excursion to
200-1000 volts (ringing voltage is still 90-140 pulsed DC), leaving the
end equipment to fend for itself.
The Edco devices sound good, if they recover without shorting or
degrading their signal characteristics.

But if there's no upstream protection, the Edco probably won't be able
to sustain the damage or protect the DSU/CSU.

Sorry, no advice to offer on brands/models of testing gear.