From: Chuck Hill ([email protected])
Date: Wed Jan 24 2001 - 19:04:37 PST
Its difficult to see the entire schematic just from your words below but
perhaps I can suggest some things for you to consider.
The common mode choke and capacitors are a network that forms
resonances. The Q of the resonances is limited by the losses in the
capacitors and choke. The resonant circuits are excited by the switching
transients from the DC to DC converter--these transients are mainly
current. The current in the resonant circuits (and the capacitors) is
magnified by the Q, that is, it is largely circulating current. This large
circulating current causes localized heating within the capacitor resulting
in premature failure.
I've seen this many times before in switching supplies. I've also seen
designs that pay no attention to the AC current rating of capacitors, yet
seem to work just fine. Good designs should ensure all components are
operated within specifications.
At 07:25 PM 1/24/01 -0500, Michael Nudelman wrote:
>I have a situation.
>I have a rather large board (3'x1.5')
>It'a telecom board. I have 48V DCDC on it (big one, 200W). The EMI filter is
>a common mode choke with a ceramic 0.01uF cap 100V in parallel with 10uF
>100V cap before and after the choke. In front of it is a slow-start module,
>which is a mosfet controlled by an RC in its gate. It ramps in-rush current
>ove couple of ms.
>It is a module, once suggested by Lucent for its DCDCs in their appnote and
>it is a standard solution on all our boards.
>Now, here what happens.
>Before i/o this filter we used to have a Filter module by the Mfr who does
>the DCDC, and few times it blew up on us.
>We designed it out, and temprarily installed wires shorting out the Filter.
>Boards worked normally, until we had to pass some tests and started
>installing a daughterboard, containing the filter, described in the
>Soon after large batch of boards was assebmled the front end ceramic cap
>(facing the 48V righ after the slow start module , the 0.01uF one)started
>blowing up in a large percentage of the boards (10-20%).
>Sometimes the cap that is the rear end (after the CM choke) blows up. But
>not as often as the front end one.
>Notice, that after the slow start, in parallel to this EMI filter and DCDC,
>there is another EMI filter, exactly same, with CM choke of about twice the
>inductance of the first, and same set of capacitors. Neither capacitor in
>this filter blows.
>The only difference here I see is that the first filter is on the
>daughterboard, and the second is on the main board. The second power supply
>however is small (10W)
>Now about the board: for almost 3 feet the feeding 48V line runs from
>connector to the slow start, after which it connects to the EMI filters
>within an inch and then the filters goes to the DCDCs.
>Now my theories:
>1. Wrong capacitor is being installed. This was checked - the spool has
>right part # on it, but who knows what's in there.
>2. The daughterboard is being stressed with the caps already mounted, which
>cracks the caps, and they flame.
>These two theories made me do an experiment: I took 30 boards and exposed
>each to 100V for 1 minute. Nothing happened.
>The slow start module creates an amlifier (common gate one), and a noise
>created during plugging connector in, amplifies times R(load)/Rsource,
>where Rload may be significantly larger than R source. R source is
>essentially output impedance of the humongous power supply, and R load is
>DCDC (it does not draw much current until it is turned on ) plus CM choke.
>Of course, at first moment it is shorted by a capacitor.
>The noise due to charging pre-slowstart board capacitance, though short, is
>very significant. I saw the 80Mhz oscillaton with current up to 80A for few
>The theory is far-fetched, but....
>Anyone has a similar experience? Is it my paranoia and it is just
>mechanically mishandled/wrong capacitors?
>Sorry for long text
>**** To unsubscribe from si-list or si-list-digest: send e-mail to
>[email protected] In the BODY of message put: UNSUBSCRIBE
>si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
>si-list archives are accessible at http://www.qsl.net/wb6tpu
**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected] In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
This archive was generated by hypermail 2b29 : Tue May 08 2001 - 14:30:44 PDT