How about actually putting some figures into the equation. There have been various answers to the question like "Just add a MOV between phases" good answers if you only want the circuit to withstand one pulse then the chances are that the MOV would be destroyed and possibly the circuit the MOV put there to protect.
If the rise time of the pulse is very fast the MOV will stand no chance of withstanding such a pulse, it needs to have some common mode inductance to damp down the rise time of the pulse to for the MOV to have any chance of clipping the pulse down to a acceptable level. (Best performance use a iron dust ring core in common mode) Secondly by adding two "Y" capacitance "say 10nf from line to earth" to the input of the circuit this will again help the pulse be clipped down to a acceptable level. All this circuitry should be placed as close to the input of the mains to help the performance of the filter. Also a filter design program such as Spice, that is if you are in possession of it, does not really help very much as the results given can be sometimes way out to what is actually required, By using the above circuit this will give a better result, probably not a perfect result first time, but it will give you a good starting point. Regards Richard Garbett Approvals Department Research & Development Apricot Computers (MEPCD) Ltd. 2500, The Crescent Birmingham Business Park Birmingham B37 7YE > -----Original Message----- > From: HMellberg [SMTP:[email protected]] > Sent: 23 April 1998 17:26 > To: [email protected]; [email protected] > Subject: Re: Surge filter > > The best way to design a filter is to establish the performance > criteria > first. > > Determine the source impedance > Determine the load impedance > Determine the attenuation required and at what frequency > Determine the roll-off required > > Now you can determine what order filter you need and whether it is a > butterworth or eliptical based on component count and cost. > > At this point you best use a filter design program such as FILSAP and > then > characterize it with SPICE. > > Don't forget that common mode rejection filters are a bit more tricky > and > require all lines to have a transfer impedance. > > Last but not least in importance is the topical placement of filters > especially critical at high frequency. The closer they are located to > the > entry/exit point the better the performance. And, return path > inductance and > length, is not to be ignored. > > Hans Mellberg > Director of Engineering > Compliance Certification Services
