450V rated Linear regulator suffers high voltage spike to its input

[ocset]

Hello,

We are using a 450V rated LR8 linear regulator in our offline  product. If the product  happens to get  switched on at mains peak then the LR8 gets a voltage of 600V peak at its input (as in the scope shot attached).     
However, if we put a 100nF X2 capacitor across live/neutral, then the peak voltage applied to the LR8 is 500V. …However, the peak duration is longer even though its less voltage (as shown in attached scope shot).
Which is  less harmful to the LR8? 

LR8 Linear regulator datasheet:
http://www.microchip.com/wwwproducts/en/LR8

Do you think these  overvoltages could cause long term damage to the LR8? 

 

[Delta]

Remember you are dealing with a mains voltage of 292v RMS for periods of several minutes.

[capt bullshot]

Do you think these  overvoltages could cause long term damage to the LR8? 

Yes.

This is clearly stated in any datasheet I know of:
-------------------
 Stresses above those listed under “Maximum Ratings” may cause permanent damage to the device. This is
a stress rating only and functional operation of the device at those or any other conditions above those indicated in the
operational listings of this specification is not implied. Exposure to maximum rating conditions for extended periods
may affect device reliability.
--------------------

From my experience, modern devices have really very little margin from their nominal ratings to destruction. Older ones often had large margins, and some designers made use of that. Absolutely no more recommended with today stuff.

[NiHaoMike]

What output voltage are you going for? If it's for a low voltage bias supply, it would make sense to use a resistor to drop a lot of the voltage. And it would also make it a lot easier to add additional protection.

[Ice-Tea]

Do you think these  overvoltages could cause long term damage to the LR8? 

How is that even a question???

[Zero999]

Add a series resistor, say 10k, and a transient suppression diode, such as the P6KE400, before the regulator. That should limit the input voltage to a safe level.
https://www.diodes.com/assets/Datasheets/ds21502.pdf

[dmills]

Concur, add some series resistance, the ringing is probably cable inductance interacting with the input cap, something of a classic killer of switchers (But you usually see it with DC supplies) the tell being the frequency change when you add the 100nF. What inrush current are you measuring?

One does not get reliability by running anywhere close to (never mind exceeding) abs max ratings which seems to be something you guys keep on doing, and then being surprised when you see the same old problems.

73 Dan.

[ocset]

What output voltage are you going for?

3V

[Zero999]

What output voltage are you going for?

3V

How much current?

You definitely need a series resistor, to reduce the power dissipation in the regulator, perhaps even a low pass filter, might help. Try 10k and 100nF.

[T3sl4co1l]

Do you think these  overvoltages could cause long term damage to the LR8? 

How is that even a question???

I lowkey suspect treez is an ex-pat Russian mechanic forced into electronics for some other reason.

Yuo see Ivan, it not bad if regulator overvoltage a little. Puor vodka on top, it fix.

Tim

[BravoV]

....450V rated
.....gets a voltage of 600V

Do you think these  overvoltages could cause long term damage .....

Will you ask the same question again if it was struck by lightning ? 

[Red Squirrel]

....450V rated
.....gets a voltage of 600V

Do you think these  overvoltages could cause long term damage .....

Will you ask the same question again if it was struck by lightning ? 

Depends on the duty cycle of the lightning. 

[PartialDischarge]

Treez, take a look at Figure 1 and table 2 of this document, 450V is low for a semiconductor to be connected to mains

https://library.e.abb.com/public/6f03cdd0f2264ff48f2992e62497dd5a/Voltage%20ratings%20of%20high%20power%20_%205SYA%202051NLay.pdf

[Zero999]

Treez, take a look at Figure 1 and table 2 of this document, 450V is low for a semiconductor to be connected to mains

https://library.e.abb.com/public/6f03cdd0f2264ff48f2992e62497dd5a/Voltage%20ratings%20of%20high%20power%20_%205SYA%202051NLay.pdf

Come on, that document is targeted at industrial controls, not the domestic environment. This is evident from the fact that the lowest voltage in the table is 400VAC.

If you take a look at any piece of equipment, designed for use in the home or office on 230V, you'll find the semiconductors are usually rated to 400V. 450V is also common and I've seen the odd 500V here and there, but 400V is sufficient for most applications. MOVs and spark gaps are used to protect against higher voltage surges and generally do their job quite well.

[PartialDischarge]

Of course it is, but any *well designed* system with semiconductors connected directly to mains, even for 220V should follow those guidelines. Ligthings transmit across power transformers, and inductive loads are present in every home

Do a study for example on offline switched buck converters, these are eminently low power and aimed at home appliances (viper family, Ucc28880, topswitch, ncp112x....). Their mosfets are rated 600V and upwards and I guess there is a reason.

400V does not even cover a short term increase of 10% in line voltage over the maximum of 264Vrms as defined in EN 50160.

[Ice-Tea]

Of course it is, but any *well designed* system with semiconductors connected directly to mains, even for 220V should follow those guidelines. Ligthings transmit across power transformers, and inductive loads are present in every home

Which is why you have MOVs and the like.

Do a study for example on offline switched buck converters, these are eminently low power and aimed at home appliances (viper family, Ucc28880, topswitch, ncp112x....). Their mosfets are rated 600V and upwards and I guess there is a reason.

Yes, the reason is called 'reflected voltage'.

400V does not even cover a short term increase of 10% in line voltage over the maximum of 264Vrms as defined in EN 50160.

No? Because that would give a peak voltage of 372V

[PartialDischarge]

Which is why you have MOVs and the like.

Yes, they are so effective that I wonder why not install a couple of these at the point of entrance of every home and voila, surges and lightings are gone for ever and for every equipment inside the house /sarc. MOVs and the like can only take so much, so the higher rated, in voltage terms they are the better, which means that the semiconductors will have to cope with more. Unfortunately few have read the life expectancy data on these devices. I did write *well designed* system for a reason.

Yes, the reason is called 'reflected voltage'.

Buck converters don't suffer reflected voltage, but again refer to the previous answer, the higher rated a mains connected semiconductor is, the better.

No? Because that would give a peak voltage of 372V

Nope, and I know reading correctly the standards is a bit difficult, the 10% applies over the 10% of the nominal 240V

[Ice-Tea]

Which is why you have MOVs and the like.

Yes, they are so effective that I wonder why not install a couple of these at the point of entrance of every home and voila, surges and lightings are gone for ever and for every equipment inside the house /sarc. MOVs and the like can only take so much, so the higher rated, in voltage terms they are the better, which means that the semiconductors will have to cope with more. Unfortunately few have read the life expectancy data on these devices. I did write *well designed* system for a reason.

If you design regular home appliances like you would industrial control gear (like ABB) you won't make a penny, ever. In this case, "well designed" is "over designed". You could also design according to MIL specs or space specs. Granted, the result might be better or more reliable but nobody will ever buy it. The art of electronics is not so much to design something that will never ever break and has the best possible specs imaginable. It's also about designing something that makes economical sense. It's perfectly possible to design a system that will handle surges and has a  450V semicon

Yes, the reason is called 'reflected voltage'.

Buck converters don't suffer reflected voltage, but again refer to the previous answer, the higher rated a mains connected semiconductor is, the better.

Sure. Bring on the 2kV semiconductor. And what does a buck have to do with this?

No? Because that would give a peak voltage of 372V

Nope, and I know reading correctly the standards is a bit difficult, the 10% applies over the 10% of the nominal 240V

Gee, thanks. Guess I should take those reading comprehension classes again. This is what the spec says:

Under normal operating conditions the voltage variation should not exceed  ± 10%: Overall  experience  has  shown,  that  sustained 
voltage  deviations  of  more  than  ±10%  over  a  longer  period  of  time  are  extremely  unlikely,  although  they  could  theoretically  be  within  the  given  statistical  limits  of  2.3.2.  Therefore,  in  accordance  with  relevant  product  standards  and  application  of  IEC  60038  end  users’  appliances  are  usually  designed  to  tolerate  supply  terminal voltages of ±10% around the nominal system voltage, which is sufficient to cover an overwhelming  majority  of  supply  conditions.  It  is  hence  neither  technically  nor  economically  viable  to  generally  give  appliances  the  ability  to  handle  supply  terminal  voltage  tolerances  broader  than  ±10%.  If,  in  single  cases,  evidence  is  given,  that  the  magnitude  of  the  supply  voltage could depart beyond this limit for a longer period of time, additional measures are to be taken  in  cooperation  with  the  local  network  operator.  The  same  applies  in  cases,  where  specific appliances have an increased sensitivity with respect to voltage variations. Situations like those arising from faults or vo
ltage interruptions, the circumstances of which are eyond the reasonable control of the parties, are excluded.

Must be me, but I can't find the 10% on top of the 10%. Again: maybe just me. Even then, the 450V device Microchip designed specifically to operate ofline would pass that requirement.

[mikeselectricstuff]

Depending on your input range you'll be dropping at least 150v across the regulator, so dropping a significant part of this across a resistor isn't going to hurt, will spread out the heat and reduce stress on the regulator.

If you don't need universal input, so your input range is smaller, you may be able to lose the regulator and use a simple resistor+zener or shunt regulator.
 

[Zero999]

Which is why you have MOVs and the like.

Yes, they are so effective that I wonder why not install a couple of these at the point of entrance of every home and voila, surges and lightings are gone for ever and for every equipment inside the house /sarc. MOVs and the like can only take so much, so the higher rated, in voltage terms they are the better, which means that the semiconductors will have to cope with more. Unfortunately few have read the life expectancy data on these devices. I did write *well designed* system for a reason.

If you design regular home appliances like you would industrial control gear (like ABB) you won't make a penny, ever. In this case, "well designed" is "over designed". You could also design according to MIL specs or space specs. Granted, the result might be better or more reliable but nobody will ever buy it. The art of electronics is not so much to design something that will never ever break and has the best possible specs imaginable. It's also about designing something that makes economical sense. It's perfectly possible to design a system that will handle surges and has a  450V semicon

I agree. It also wouldn't surprise me if industrial installations are more prone to higher surges, because the cables will be thicker and therefore lower impedance.

Nearly every PC PSU I've seen has 400V rectifier diodes, a 400V electrolytic and 400V MOSFETs and surprisingly very few fail. MOVS and gas discharge tubes help, but semiconductors will also go into avalanche breakdown, absorbing the spike and resume normal operation, if the spike doesn't contain enough energy to damage them. I wouldn't be surprised if the rectifier diodes inside the PC I'm using at the moment have absorbed a few high voltage transients, yet still work perfectly.

[capt bullshot]

Nearly every PC PSU I've seen has 400V rectifier diodes, a 400V electrolytic and 400V MOSFETs and surprisingly very few fail. MOVS and gas discharge tubes help, but semiconductors will also go into avalanche breakdown, absorbing the spike and resume normal operation, if the spike doesn't contain enough energy to damage them. I wouldn't be surprised if the rectifier diodes inside the PC I'm using at the moment have absorbed a few high voltage transients, yet still work perfectly.

Usually, it's the rather large electrolytic capacitor that absorbs the transient energy. First, the transients energy must put enough charge into the cap to let the voltage rise, this takes care of a lot of transients, and second, they act like an zener diode if you exceed the rated voltage by some 10%.
So the voltage across the cap is limited to maybe 440V by the capacitor itself, that helps to protect the semiconductors from too high a voltage. If you put enough energy into the transient, the rectifier diodes fail due to high single pulse current, and the electrolytic may vent. They do vent in a spectacular way if the energy is high enough.

[Niklas]

Google "overvoltage category" and check the needed impulse withstand voltage for 230 V rated appliancies. Electrolytic capacitor are quite slow and are almost transparent to fast transients. The EMI filter at the PSU's input reduces emissions and increases immunity.

[David Hess]

Do you think these  overvoltages could cause long term damage to the LR8?

Yes, an input protection circuit is going to be required.  I would use some combination of a passive series and shunt protection network.

If continuous faults are possible, then I would use a high voltage transistor cascode which can also disconnect the input.  Foldback current limiting would be a good idea also.  Of course if you implement all of this, then the LR8 may be superfluous.

[fcb]

Cool - I was not aware of the LR8.

*looks for a project to use it in*

[ocset]

Thanks, We cannot add anything on the PCB itself , as there's no room.
We can only add stuff to the live and neutral wires that connect to the product. –Hence the X2 capacitor we added, as described in the top post.

The "500v peak" scope shot in the top post was with a 100nF X2 capacitor connected between live and neutral. There was also 25uH (10 metres) of coiled up mains cable ahead of that so that we could simulate line inductance.
When we changed the coiled up cable to 100metres of (coiled up) length, then the voltage peak goes up to 550V (as attached)…..youch! (100m of mains cable had an inductance of 73uH )
Obviously, the more “line inductance” we add, the higher the peak voltage to the LR8 is going to go….youch again!

So, the attached 550v peak scope shot is with 73uH of mains cable and then a 100nf x2 capacitor, then the product

We are wondering what is the typical maximum line inductance in the installation that our lamp will see?
It’s a lamp that gets put in pub gardens, outdoors.