[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference 7.286::ethernet

Title:	Ethernet Volume 3

Moderator:	UPSAR::THOMAS

Created:	Thu Oct 08 1992
Last Modified:	Thu May 22 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	914
Total number of notes:	3047

901.0. "H4005 problems" by CHEFS::KERRELLD (To infinity and beyond...) Tue Feb 11 1997 02:55

Can anyone help with this problem a UK DECUS member has with H4005s failing 
:-

"Thanks for your reply.  Our H4005s are all late 1992 vintage, so I guess we
can asssume it's not a batch problem.  Given that all our failures are recent,
and seem to be getting closer together, we're working on the possibilities of
electrical problems on our network. 

Since the terminators have not been damaged we assume the AUI side of the
transceiver is what's failing and the transceiver isolates the thickwire
cable.  The first one to go did actually catch fire (burst electrolytic
capacitor, charred transformer board) so whatever's happenning seems to be
dramatic, but somehow the driving devices (2 separate DELNIS, a DECServer 100,
and a MicroVAX 3100) suffer no ill effects.

I think we need to know more about the hardware so we can try to work out what
electrical faults we need to look for.  On the H4005 the electrical tolerances
and design info (a circuit schematic ?) might help us figure out what did the
damage without affecting anything else, and for the ethernet some information
about earth voltage differences, ripple voltage tolerance and so on may help.
If you know of anyone who could help obtain this information or already knows
what we ought to be looking for, perhaps you could pass our request on to
them.

By the way, we are working on the basis that the failures are not caused by
simple ageing (because the failures have started suddenly and are bunched
together, and the first failure was so dramatic!), but we might be 
wrong(?)."


Thanks,
Dave.

T.R	Title	User	Personal Name	Date	Lines
901.1	Check out recently done/added work for problems...	NETCAD::BATTERSBY		`Tue Feb 11 1997 10:09`	15
	if you do a dir/title=h4005 you will see that there have been other entries regarding H4005's. I'd suggest browsing through this short group of entries, to see if your scenario fits something discussed in these entries. I might also add though that if your UK DECUS member is suddenly seeing failures, to investigate recent changes/work done on their LAN to see if this work might be the source of cause for these recent failures. Things like a newly added segment of thickwire added, which has the bare metal surface of a barrel connector, or an N connector which is in contact with some electrical potential other than earth ground. Perhaps there is a newly added cable segment that isn't/wasn't ever properly grounded. Look perhaps for causes like these. Bob
901.2	Ungrounded coax, static discharges???	STKHLM::WEBJORN	Gullik Webj�rn Network Advisory	`Wed Feb 12 1997 08:41`	17
	Most probable cause if multiple H4005's are going dead is that the coax has lost it's ground connection. The typical isolation between AUI and coax should stand about 3 KV. Any fan or rotating belt in an airconditioner or something in the building ventilation can charge dust and particles with minute charges. As the dust or particles accumulate on the cable( ungrounded that is ), eventually the charge build up until you get arcing somewhere. If that somewhere is within a H4005 I can imagine the DC-DC converter dying. Since the AUI cables are fairly well shielded and grounded it's less likely that 'the other end', the adapter, DELNI etc suffers. This is the reason thinwire has 1 Meg ohm static discharge resistors built in. Gullik
901.3		CHEFS::KERRELLD	To infinity and beyond...	`Wed Feb 12 1997 10:39`	3
	Thanks! I'll pass on the advice. Dave.
901.4	Dust????	NOTED::defctb.lkg.dec.com::Bill	Bill Melaragni, HPN, DTN 226-6670	`Thu Feb 13 1997 08:04`	22
	RE .2: I'm not sure i buy THAT! There are many reasons i don't like that explanation, but here are just a couple: 1) The H4005 has the same 1 Mohm bleed-off resistor as the DESTA and other thinwire MAUs. 2) I find it next to impossible to believe that static discharge will fry any of the components in the MAU. The tranfer impedance of the static charge is on the order of megohms. I could imagine, maybe, damaging the DP8392 transceiver chip, but not the DC/DC. It's too tough. No, i believe the problem is same old one: capacitors that have been exposed to temperatures exceeding the tolerances. There are several electrolytic caps in the DC/DC that are rated for a "low" temperature of 80'C. These caps, when exposed to these temps for any length of time, begin to loose their electrolytic properties. At this point the caps look like short circuits and roast themselves and the DC/DC. My opinion. bill
901.5	Dc-to-DC Converter ECO'd to fix this	NPSS::KIRK		`Thu Feb 13 1997 08:43`	36
	This has been addressed with an ECO to replace DC-to-DC Converters. Details are in my mail'ed response to the base noter. From: NPSS::KIRK "Dick Kirk DTN 226-7048" 12-FEB-1997 14:24:24.14 To: CHEFS::KERRELLD CC: ANTON,WONKKA,DELNI::NICKERSON,KIRK Subj: H4005 problem Dave, We have ECO'd the H4005 (new rev D02) about a year ago to replace the DC-to-DC converters. The new converters are rated for a higher temperature range that the previous ones. We had been seeing a number of failures of the DC-to-DC converters after the warranty period, but before the expected failure times. The failure mechanism was that a capacitor failed, shorted, and the circuit drew enough current to fry a resistor, and in some cases cause smoke. Time to failure seemed to be related to time at elevated temperature. The ECO was implemented about February 1996. H4005s of revision D02 will have the new DC-to-DC converters. They will be more resistant to high temperature conditions. Please let me know if you need any further information. Regards, Dick Kirk Network Product Support