I would like to make a good surge protector for a device that will have a hardwired mains connection. The protected electronics will consist of a ~10W SMPS and a low-power AC load (<100W) controlled with a triac.

I plan to use a mains filter (nothing fancy) behind some combination of GDTs (Gas Discharge Tubes), MOVs (Metal-Oxide Varistors) and/or TVSs (Transient Voltage Suppressors). This is meant to work at 110V (US) but the SMPS is universal so it would be nice if it worked with 230V as well (mostly because this is the standard voltage in the place where I live).

I understand that I want the faster device (MOV, TVS) to have higher voltage rating than the GDT so both have a chance to work when really needed. Unfortunately both GDTs and MOVs have rather lousily specified I-V characteristics and their "breakdown voltage" is more of a "breakdown region".

A GDT with typical 230V breakdown won't work with 230V AC since it may start conducting at as low as 180V. If OTOH I go with a 350V typical (280V minimal) my protection is guaranteed only at 420V since this is the maximum ignition voltage, right? This leaves little room for the choice of the MOV breakdown voltage (which is a voltage range as well) and still places heavy burden on the downstream elements. Is this a lost case and the only solution is to have all the downstream components rated >900 V?

I welcome any articles, suggestions or rules of thumb for designing such (simple) surge protectors.


A MOV is an essential part of your mains-side safety circuitry (despite what user886922 says). 300V MOVs are commonly used to protect universal input (85-264VAC) power supplies. The MOV is sized such that the energy it can crowbar safely is more than enough to blow the mains fuse, which will provide the necessary disconnect. It should be located 'inside' any EMI filter components as surges tend to be amplified by EMI inductors (think boost converters).

EMI filter capacitors (X- and Y- rated parts) have dielectric strength far beyond their rated voltage, so there's little concern there. Sometimes I see GDTs in parallel with the Y-caps (line-to-earth, neutral-to-earth) but often they're omitted because of the toughness of the safety capacitors.

I wouldn't be surprised if your existing 10W SMPS doesn't already have some combination of GDTs and MOVs.

  • \$\begingroup\$ True but by the time the fuse pops the surge would have had time to damage the protected equipment, depending on its rise time - eg lightning strike - so I mantain that the presence of GDT/MOV on mains equipment is mostly useless and a feel-good measure. All expensive equipment that requires complete protection that I've seen uses a UPS type design. \$\endgroup\$ – Kristoffon Aug 23 '11 at 19:40
  • \$\begingroup\$ @user886922 Having been involved in many AC/DC power supply designs, I respecfully disagree. Surge testing as per IEC61000-4-5 will show you that having a MOV is very helpful in passing this requirement, which is as close to a real-world lighting strike as you're going to get in a lab. Your argument appears to me to be "a real lightning strike will blow up everything, why bother?" which doesn't sit well with customers - some measure of protection is still better than no protection at all! \$\endgroup\$ – Adam Lawrence Aug 23 '11 at 20:36
  • \$\begingroup\$ The SMPS is a cheap one so even the mains filter is not populated on the PCB. This is the reason why I want to add some external protection and filtering. Do I understand correctly that you recommend the MOV before the common mode choke? \$\endgroup\$ – jpc Aug 24 '11 at 11:29
  • \$\begingroup\$ The MOV should be as close to the bridge rectifier as possible - any EMI circuitry should be between the fuse and the MOV. That's the practice I'm accustomed to. \$\endgroup\$ – Adam Lawrence Aug 24 '11 at 12:38

MOVs/TVSs/etc on mains lines are mostly useless. Ask yourself if you can picture a combnination of MOV/TVSs etc being effective agaisnt all the power available at the mains.

Any effective suppression will work by disconnecting the mains from the load, which complicates the design significantly as you need both a means of detecting a surge as well as switching the load off before damage is done.

A simple yet bulletproof protector is the no-break - by lowering then raising the voltage in a controlled manner you pretty much guarantee no surge can pass through.

  • \$\begingroup\$ Thanks for the answer. 1200V protection one can get from series a MOSFET or IGBT also does not look like it will work with all surges (which IIUC frequently go into several kVs). A relay OTOH will not react quickly enough for this to be useful. Could you comment on that? UPS would work but only if the protected equipment is more expensive than the UPS :) (it is unfortunately not the case in my project). \$\endgroup\$ – jpc Aug 22 '11 at 21:20
  • \$\begingroup\$ PS. Do you have any good articles about or maybe even books mentioning this subject? \$\endgroup\$ – jpc Aug 22 '11 at 21:20
  • \$\begingroup\$ those are the trade-offs, pretty much. sorry, don't have any references to cite. \$\endgroup\$ – Kristoffon Aug 22 '11 at 22:34
  • \$\begingroup\$ A MOV/GDT combined with a fast acting fuse upstream makes sense. The safety device breaks through, makes the fuse blow, disconnecting the circuitry downstream. \$\endgroup\$ – try-catch-finally Dec 24 '17 at 18:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.