# Fuse selection for Flyback application under Surge conditions

#### Bjtpower_magic

##### Junior Member level 3
Hi,
Hope you are doing well.

I am designing Flyback converter of 70W(Non-PFC).
I need help for selection for Fuse at the input under surge conditions.

I have surge parameters which is Live to Neutral 1KV Pulse for tr/tn=1.2/50uS
Nominal current requirement for application at lower voltage (170VAC) would be ~2A considering derating of 75percent and at 50degree ambient temp

Now everyone knows that, when we apply surge there is High short circuit current will through MOV (and obviously through Fuse).

Can anyone explain me how to select the Fuse and MOV for this application.
Input Voltage=170-265VAC
Output Power=70W
Input power=output power/Efficiency=70/0.8=88W
Input current=input power/V*PF=88/(170*0.5)=1.04A

Any simulations or practical approach would help

Regards,
Marx

#### KlausST

##### Super Moderator
Staff member
Hi,

Ideas:
* slow blow fuse (read fuse datasheet about trip time for the calculated current pulse)
* fuse after MOV
* series inductance to lower the current peak

Btw: Voltage (1kV) will not make the fuse to trip. You need to know the pulse current.

Klaus

#### Bjtpower_magic

##### Junior Member level 3
Hi KlaussT,
How to calculate Pulse current? how to derive it?

From which test we will pulse current? How?

Regards,
Marx

#### KlausST

##### Super Moderator
Staff member
Hi,

I can't which test you need to survive.
Pulse, voltage, waveform, energy should be specified by the test.
Current and it's waveform needs to be calculated including your circuit.

Klaus

#### Bjtpower_magic

##### Junior Member level 3
Hi KlausST,
Could you please give me example?
I have specified my test levels for Surge, could we get any idea from that?
What additional information do you need for calculations?

#### FvM

##### Super Moderator
Staff member
You can refer to manufacturer application notes for MOV selection. Epcos suggests S07K275 (7 mm disc, 275Vrms) for 1 kV surge capability.

The voltage + current waveform is defined e.g. in IEC 61000-5, for a discussion of network circuit parameters see e.g. https://www.pes-spdc.org/system/fil...e_generator_circuit_simulation_valuesrev3.pdf

In case of shorted surge pulse, you get I²t of about 1.5 A²s for 1 kV pulse, the fuse melting integral should be clearly higher. Depending on the characterictic, you end up at 1 to 2.5A fuse.
--- Updated ---

Here is a LTspice simulation of surge generator with MOV

#### Attachments

• 1.3 KB Views: 3
Last edited:

#### Bjtpower_magic

##### Junior Member level 3
Hi FvM,
How did you get i2t from the graph?

#### FvM

##### Super Moderator
Staff member
I got it from simulation. Irms^2*Twindow

#### Bjtpower_magic

##### Junior Member level 3
I got it from simulation. Irms^2*Twindow
It is hard to validate.. I dont know how to measure it.
Irms and Twindow from the graph.

Regards,
Marx

#### zenerbjt

##### Full Member level 2
What is the application?

Usually you can gauge what MOV you need from that.
For consumer goods, its usually just a 10mm MOV. 250-275VAC.
In truth, with a non-PFC’d flyback , you must have quite a big Electrolytic input capacitor?
That in itself will give you good transient protection. Most transients wont be able to charge that electrolytic up more than a few volts.

You say 1.2us/50us….so that’s a 1.2us initial surge.
Supposing you have an 82uF input electrolytic…..
Then a surge current of 3000 Amps would only increase the capacitor’s voltage by 43V in 1.2us….so do you really need a MOV at all? Your input cap is doing a very good job for you.

And seriously, how many 3000 Amp transients do you think any 10mm MOV can take before it pops? Not many I can tell you. A MOV suffers degradation every time it clamps a transient. So whatever MOV you specify…after a couple of transients it won’t be anything like the initial spec.

I would leave the MOV out….especially if your fuse is upstream of the MOV…..because you are going to increase the likelihood of the fuse blowing.

We used to send our SMPS’s off to Littelfuse (in USA or Canada?) and they used to do a surge test on them and report back the results. So you could just buy a littelfuse MOV and fuse, and call their apps guy and ask him your question.

#### FvM

##### Super Moderator
Staff member
The conclusions don't sound quite plausible, if the voltage is clamped by the rectifier and filter capacitor, the MOV won't be stressed at all. Also not affect the chance of blowing the fuse.

The surge specification is 1 kV, respectively you get 500 A peak current (effective network impedance is 2 ohm). How did you arrive at 3000 A?

Last edited:

#### zenerbjt

##### Full Member level 2
Yes you're right the fuse would be just as stressed because the MOV wouldnt come into conduction. And as such, just forget the MOV...the Big Al el cap will save the day.
That is the one beauty of Al El caps...they have wet electrolyte which spreads the heat from the surge very well...whereas many other caps would crack and break under the surge......apart from X2 caps but they dont come in 82uF, 450V.

In serious truth, how many mains transients will be able to life the voltagte of a 82uF capacitor more than a few volts.?.....i'd say virtually none. Specially since most equipment exists on the phase with multiple other equipments, which also have movs and caps etc etc.

Most main transients are due to failure of a high power device on the mains, where a huge short circuit current flows, and then its suddenly broken by a fuse blowing.....and since the inductance of the mains wiring isnt very much, there isnt much energy in the stray inductive current, so most mains transients are pitifully small............a lot of mains transients dont travel very far becuase they get absorbed by the mains wiring resistance and the stray capacitance of the wiring.

Many of the fuses of high power devices have large RC dampers across them, which absorbs the inductive energy when the fuse blows....so you dont get mains transients at all.......its only very old equipment that doesnt have this.
--- Updated ---

Attached is an LTspice sim of your 1kv transient. Just convert to .asc and run it.

You will see that the voltage of the El cap hardly rises very much at all...no MOV needed. Keep MOVs out of it if you can...they wear out and fail and blow the fuse when they fail....rendering your Power supply useless.
--- Updated ---

To proove my point, shove an LR8 450v regulator on the mains post bridge bus, in parallel with a 22uF cap.......i bet you you can leave it for years and the LR8 will not blow...because no transient will lift the 22uf cap voltage high enough to blow it....by the way, an LR8 volt reg is a good "transient spotter" as it will blow if ever exposed to 500V or more.

Mains transients are over-hyped.....most input filters will gobble them up.

#### Attachments

• 2.2 KB Views: 1
Last edited:

#### Bjtpower_magic

##### Junior Member level 3
I got it from simulation. Irms^2*Twindow
Hi FvM,

#### FvM

##### Super Moderator
Staff member
It's Irms and and measurement interval in LTspive waveform viewer. In the shown short circuit example 175²*100µs = 3 A²s

#### Bjtpower_magic

##### Junior Member level 3
Yes.. got it for Melting point of Fuse when short circuit conditions.

For MoV, We need to remove the resistor R5? and check for the current passing through MoV?

Last edited:

#### FvM

##### Super Moderator
Staff member
Yes remove short for MOV test. However, as others pointed out, the input capacitor will sink most of the current.

#### zenerbjt

##### Full Member level 2
Has the OP run the simulation of post #12 yet?
I think it shows pretty conclusively, that the MOV will not come into conduction with a 1kV 1.2/50us transient. As FvM said, this is due to the input capacitor taking all the current. The inductive input filters will initially block the current , but will then saturate, and the capacitor will then port the current, and keep the voltage too low for the MOV to come into serious conduction. I think if the MOV model is put into the sim of post #12, then we will see that the MOV doesnt conduct after a 1kV 1.2/50us transient.

Also, if the transient is 1kV, and the mains is 240VAC, then as we know, most transients occur at the mains peak, and so we can effectively subtract 339V from the 1kV transient, and say its actually a 661V transient.

Last edited:

#### Bjtpower_magic

##### Junior Member level 3
Has the OP run the simulation of post #12 yet?
I have checked the simulation not run because it doesnt had surge Components.

I think it shows pretty conclusively, that the MOV will not come into conduction with a 1kV 1.2/50us transient. As FvM said, this is due to the input capacitor taking all the current
Which capacitor are you refering? Bulk capacitor? Y Capacitor?
Also, if the transient is 1kV, and the mains is 240VAC, then as we know, most transients occur at the mains peak, and so we can effectively subtract 339V from the 1kV transient, and say its actually a 661V transient
Yes.. 661 will pass through MoV for short time.

#### zenerbjt

##### Full Member level 2
Yes.. 661 will pass through MoV for short time.
The voltage on the MOV will not reach 661V, because the capacitor, your bulk input electrolytic, will stop the voltage from getting that high...you must have a bulk electrolytic input cap, because you say you're at 70w and non-PFC'd.

I have checked the simulation not run because it doesnt had surge Components.
It doesnt need surge components, because the bulk cap clamps the voltage down low.........it shows a 1kv 1.2/50us transient, and the voltage stays below 450V at your flyback....no MOV needed there.

FvM, in post #16, recognises that its the bulk cap that is clamping the transient in your case....which is of a 1kv transient of 1.2/50us, and into a mains impedance of 2 ohms.

Last edited:

#### FvM

##### Super Moderator
Staff member
We didn't yet see a complete circuit with rectifier and filter capacitor. In so far you are talking only of an expected circuit and it's surge absorption capabilities.