On Sunday, Mar 14, 2004, at 15:13 US/Eastern, Stefan Wimmer wrote:
> --- In AVR-Chat@yahoogroups.com, John Johnson <johnatl@m...> wrote:
>>> Do you have a snubber network across the relay contacts ?
>> no, what do you recommend? 10ohm and 10nf in series? I couldn't
> find a good web resource, so I'm kind of guessing.
>
>
> This subject is asked astonishingly often (not only in this mailing
> list). Are people involved in embedded systems nowadays pure software
> people? Anybody remember physics in school/university? No (E-)
> engineers here??
>
>
> Ok, back to the basics:
> As one (I hope at least _one_ of the readers) might remember, at
> least 90% of all problems in physics (where EE does belong to) can be
> solved by energy balancing.
>
> So what do we have in steady state? There is a circuit which is
> closed by your relay contacts. Either the load itself and/or some
> stray inductors are "charged" with some current flowing thru it. The
> energy stored in the magnetic field is: 1/2 L * I * I.
>
> Now when you switch off the relay, this energy has to go somewhere.
> The current "wants to continue flowing" and according to the equation
> U = -L dI/dt we get some really high voltages with even very small
> (stray) inductors, as long as the switch off time (dt) is short
> enough. This high voltage will cause some nice arcing at the contacts
> and due to the negative differential resistance of an arc, RF is
> generated (Tune a radio to the LW or SW range while your relay
> switches if you don't believe this - you will hear the RF, no matter
> what exact frequency you dialed in).
>
> Now what can we do?
> Best would be to tune down dt (make it longer), but that would kill
> your relay contacts by high current arcing (which is a different kind
> of animal since the current path is here provided by metal vapors in
> the (too slowly) growing gap between the contacts) in no time. So we
> have to take care of the current and give it an alternate path (while
> attenuating it - we want to switch off - don't we? :-)). A good
> measure would be a capacitor parallel to the relay contacts. It is
> held discharged as long as the contacs are closed. When they open,
> the current can (at first) continue to flow and charge the capacitor.
> This will result in an increasing voltage across the capacitor which
> will finally stop the current flowing.
>
> If we want to know how big the capacitor has to be - we're back to
> energy balancing:
> 1/2 L * I * I = 1/2 C * U * U.
>
> Now you can reorganize and resolve for C.
>
> And for the voltage rating keep in mind, that your capacitor will be
> charged to the peak voltage of your (AC) operating voltage plus what
> you just calculated above.
>
> That's it.
>
>
> Really?
> No!
>
> What happens if the relay contact close again? Oops! Yes, they will
> short circuit a charged capacitor. Well, since we don't really want
> to weld the contacts shut, we should include a resistor to limit the
> current to a value that can be handled by the relay contacts
> additionally to the load current (iow: the difference between the
> current rating of the contacts and the (inrush) current of the load).
> Usual Ohm's law applies (calculating with Umax as discussed above).
>
> If you "guesstimated" your (stray) L in the equation above, it is
> good practise to double or triple your C and calculate the needed R
> accordingly, just to be on the safe side.
>
> Now you have your snubber circuit (RC-network) without any magic,
> guessing (ok, stray inductance) and head scratching (I hope :) ).
>
>
>
> Sorry for any wrong wordings or phrases, but english is not my native
> language. An if I sounded a little brisk - maybe I just needed a
> little venting :)
>
>
>
>
> ------------------------ Yahoo! Groups Sponsor
> ---------------------~-->
> Upgrade to 128-bit SSL Security!
> http://us.click.yahoo.com/LPJzrA/yjVHAA/TtwFAA/dN_tlB/TM
> ---------------------------------------------------------------------
> ~->
>
>
> Yahoo! Groups Links
>
>
>
>
>