Inrush current limitation into 4kW offline SMPS?

[cupoftea]

Hi,
We are doing a 4kW SMPS, 240VAC in. Its 4 interleaved Boost PFC’s at the input. Their total output capacitance is 4.3mF.
For inrush limitation we are using five 100R FMP300 , 3W resistors in parallel…..then we short them with a relay.

(there is also a 5W SMPS which runs throughout the inrush event)

The inrush episode lasts for some 700ms and is as attached……

**..Blue shows current in one of the FMP300 resistors
**..Red shows power dissipation in one of the FMP300 resistors
**..Green shows the PFC output capacitor voltage.

FMP300 datasheet
https://www.farnell.com/datasheets/3046080.pdf

Nothing in the datasheet pertains to one-off events such as inrush…do you think they will survive?

LTspice also attached

 

[d123]

Hi,

I am not being facetious: Instead of asking here, because you should know how to do this, why don't you just a) calculate the power dissipated in each resistor, and follow up with a second opinion from your trusty calculator, and b) calculate (or use an online or offline) tool) temperature rise in the resistors, c) calculate change in resistance with power dissipated due to ppm/ºC?

As you know, I don't know, and I'm painfully tired, it's been a long day but 3W resistors and kilowatts sound like snap, crackle and pop due to woefully miserly skimping to me right now.

I assume the choice of paralleled resistors is for cost-effectiveness and simplicity (and PCB real estate), and to use the well-known PMOS controlled by a BJT current source charging a capacitor at the rate required to avoid blowing things up so as to (vaguely) control the inrush current is just not practical for kW power circuits and would be worse as it would require paralleled MOSFETs, and it introduces a start-updelay that I find irrationally irritating...

 

[scopeprobe]

Hi,
We are doing a 4kW SMPS, 240VAC in. Its 4 interleaved Boost PFC’s at the input. Their total output capacitance is 4.3mF.
For inrush limitation we are using five 100R FMP300 , 3W resistors in parallel…..then we short them with a relay.

(there is also a 5W SMPS which runs throughout the inrush event)

The inrush episode lasts for some 700ms and is as attached……

**..Blue shows current in one of the FMP300 resistors
**..Red shows power dissipation in one of the FMP300 resistors
**..Green shows the PFC output capacitor voltage.

FMP300 datasheet
https://www.farnell.com/datasheets/3046080.pdf

Nothing in the datasheet pertains to one-off events such as inrush…do you think they will survive?

LTspice also attached

Just an observation but 4 Interleaved boost PFC's sounds a lot if complexity and expense for a 4 KW PFC. Just a quick google search suggest you could have simpler single stage options using analog or digitial controllers, just examples https://www.ti.com/lit/ug/tiduez3/tiduez3.pdf?ts=1652037766610 https://assets.nexperia.com/documents/user-manual/UM90003.pdf

 

[Easy peasy]

I expect the 5 R's will survive - this is an easy thing to test on the bench for repeated turns ons - and measure R temp - then also check if any are open ckt. ( don't forget to flatten the caps for each test .. - using a similar 5 R load - same energy you see ).

 

[std_match]

I think you should use the specified "Periodic-pulse Overload" as a starting point, and it is 4 times the RCWV, which will give 16 times the rated power. The 3W resistor could then dissipate 48W for one second.
If you want a peak power of 1 kW you should use a power resistor with a more detailed pulse power specification.

Will you somehow limit the minimum power cycle time? What happens if someone tries to switch on and off the SMPS repeatedly?

I would look at an NTC or a PTC:

PTC Thermistors For Inrush Current Limiting | Ametherm

Limit Inrush Current in industrial and commercial power products with NTC or PTC Thermistors. Applications and Samples.

www.ametherm.com

 

[barry]

Hi,

I am not being facetious: Instead of asking here, because you should know how to do this

It has become obvious to me that "cupoftea", or "treez", or whatever their name is, knows pretty much nothing. They post 4 or 5 questions A DAY about things they should know, or should be able to find out by doing a minimum amount of research or self-education.

I'm annoyed with their constant spamming of this board. It is becoming the Cup-of-tea Can't Figure Another Thing Out forum.

But that's just my opinion.

 

[cupoftea]

Just an observation but 4 Interleaved boost PFC's sounds a lot if complexity and expense for a 4 KW PFC. Just a quick google search suggest you could have simpler single stage options using analog or digitial controllers, just examples https://www.ti.com/lit/ug/tiduez3/tiduez3.pdf?ts=1652037766610 https://assets.nexperia.com/documents/user-manual/UM90003.pdf

Thanks very much , very much appreciated your kindly supplied App Notes..... the Totem pole PFC is very interesting. It would have been interesting if they had said why their examples only do 4kW at 200+VAC.....at 120VAC, one of them is only good to 2kW, and the other example, did not even try getting any output at 120VAC.
I'll have to read closer to see how they write the code for it....does the design engineer have to write the code?, or is the code pre-written, and you just kind of fill in a code spec somewhere for it?

The App notes you kindly post are especially interesting since, there has been in depth discussion on this forum about PFC stages >1.5kW.....its been said that 1.5kW is about the maximum for a Boost PFC (I appreciate totem pole PFC may be different?)...and if more power is needed, then more than one Booster should be used.

Also, it would be interesting to compare the input AC filter size for a 4stage, 4kW interleaved Boost PFC, and a 4kW single stage Boost PFC....ditto for Totem pole PFC.

The Totem pole PFC is very interesting though....previous discussion on here has mentioned problems with very high dv/dt and common mode noise issues, when compared to standard Boost PFC.

What would be really interesting, would be to see the Boost PFC inductor for a single stage Boost PFC that could do 4kW from 100VAC.....that would be a chunker!.....it would need to be a torroid, and sendust or similar, and likely a couple stacked on top of each other......but what windings would they use for that much current?........probably couldnt be single strand as it would be too hard to bend it round the torroid in quick enough production time.

Though thanks for the App Notes on Totem pole PFC.... this is a gem for the future.....said to be going to totally replace boost PFC in future.......i have heard that Apps Engineers working for semico's, have made Demo Boards of Totem Pole PFC's, and then rather than write up there work for presentation to the public, instead, they have just "done a runner" and taken up consultancy designer roles making totem pole PFC's for customers!...and getting well minted doing it!!.........some semico's are now trying to get round this by outsourcing some of the Totem pole PFC demo board work to distributors......i know this is happening because when i interviewed at Anglia components, they were making a demo board Totem Pole PFC!

 

[scopeprobe]

Thanks very much , very much appreciated your kindly supplied App Notes..... the Totem pole PFC is very interesting. It would have been interesting if they had said why their examples only do 4kW at 200+VAC.....at 120VAC, one of them is only good to 2kW, and the other example, did not even try getting any output at 120VAC.
I'll have to read closer to see how they write the code for it....does the design engineer have to write the code?, or is the code pre-written, and you just kind of fill in a code spec somewhere for it?

The App notes you kindly post are especially interesting since, there has been in depth discussion on this forum about PFC stages >1.5kW.....its been said that 1.5kW is about the maximum for a Boost PFC (I appreciate totem pole PFC may be different?)...and if more power is needed, then more than one Booster should be used.

Also, it would be interesting to compare the input AC filter size for a 4stage, 4kW interleaved Boost PFC, and a 4kW single stage Boost PFC....ditto for Totem pole PFC.

The Totem pole PFC is very interesting though....previous discussion on here has mentioned problems with very high dv/dt and common mode noise issues, when compared to standard Boost PFC.

What would be really interesting, would be to see the Boost PFC inductor for a single stage Boost PFC that could do 4kW from 100VAC.....that would be a chunker!.....it would need to be a torroid, and sendust or similar, and likely a couple stacked on top of each other......but what windings would they use for that much current?........probably couldnt be single strand as it would be too hard to bend it round the torroid in quick enough production time.

Though thanks for the App Notes on Totem pole PFC.... this is a gem for the future.....said to be going to totally replace boost PFC in future.......i have heard that Apps Engineers working for semico's, have made Demo Boards of Totem Pole PFC's, and then rather than write up there work for presentation to the public, instead, they have just "done a runner" and taken up consultancy designer roles making totem pole PFC's for customers!...and getting well minted doing it!!.........some semico's are now trying to get round this by outsourcing some of the Totem pole PFC demo board work to distributors......i know this is happening because when i interviewed at Anglia components, they were making a demo board Totem Pole PFC!

Most PFC's over a 1.5KW probably start to consider the use of Multiple AC Line Phases to spread the load which is probably more practical due to the large currents involved, this then requires multiple PFC controllers which i can understand the added complexity. I'm not an expert on power conversion but 4KW off a single 100V phase will be close to a 40A circuit breaker which is beyond most household wiring which is possibly why its not seen often. Industrial applications would almost certainly use 3 phases for this to keep their power lines balanced. Semiconductor manufacturer arn't stupid, they make parts to satisfy where there's a large market demand . The maths would probably tell a clearer story which is why power conversion is often a iterative design process.

 

[cupoftea]

4KW off a single 100V phase will be close to a 40A circuit breaker which is beyond most household wiring which is possibly why its not seen often. Industrial applications would almost certainly use 3 phases for this to keep their power lines balanced

Thanks, yes, i also questioned the 40A off single phase 120VAC for this reason.....but the kit we are making is not "domestic" kit.
But Before this, i thought 4kW from single phase , 120VAC, would pretty much never get done...and it would always be three phase.

 

[prairiedog]

5x3W metal film resistors for inrush on a 4,300uF cap??? That's hilarious. As a big spender, you'll need to spend more than pennies in the 4kW league. Your power supply project will not succeed with that mindset, prepare for exploding resistors.
The LT spice sim gave peak power over 1.1kW each and average over 100W each for the charging event. The 10A diode was 16A peaks for several cycles.
Yes but do you think it will work?

 

[cupoftea]

Thanks, we're thinking of moving over to NTC's instead of resistors......or, in combination with resistors.....te NTC will just immediately get hot and reduce its resistance rather than blow......i've seen offline 200-265VAC power supplies with a small 1cm NTC of 80R doing inrush into 480uF (output of Boost PFC)...so in other words......we woudl need 10 of them in parallel.........but then we only have 8R of inrush limit......so maybe we will seek those white 'bathtub' resistors...thats what was doing inrush on the 10kW three phase electric drives i worked on.

I actually think that we are going to end up putting in a wee flyback converter which will pump the 4.3mF up to 390V , and then enable the PSU......by switching on two back to back IGBTs....inrush resistors for this duty just dont seem to exist.

This app note shows a 3r3 NTC being used to inrush into 1.1mF....wont get much limitation from 3R3...and thats the cold resistance....

https://www.infineon.com/dgdl/Infineon-ApplicationNote_EVAL_2.5KW_CCM_4PIN-ApplicationNotes-v01_00-EN.pdf?fileId=5546d4624fb7fef2014fd65081616257

 

[cupoftea]

To be sure of being able to inrush into the 4.3mF from 240VAC.......two pllel 47R HSC100 resistors will be needed it appears...(no need for heatsink for them though)

https://www.farnell.com/datasheets/463742.pdf

As the attached sim shows, (LTspice), the joules of energy during the inrush goes up to 100Joules in 600ms....the datasheet says its good to 180 Joules when pulsed.

Also, the power peak goes to 2.4kW....and you are allowed a power peak of 2.5kW for 1 second....and this is for much less than one second.

From research on the web, it appears that for mains inrush into >2mF, you are supposed to either use NTC's, or a precharging regulator.......inrush resistors are not supposed to be used, however, we cant use NTC's , so will do the two 47R's , and then work on the pre-charge reg in a feww days time.

The piece of info thats not available in any datasheet is the power_per_time_interval parameter......in this case, what is the power that it can dissipate in 600ms. No datasheet has this info available.

The missing parameter from all datasheets is "power per piffelingly short interval of time".

Where inrush is concerned....

"2500W for 10ms" is_not_equal_to "250W for 1 second" is_not_equal_to "2.5W for 100 seconds"

 

[FvM]

Considering the fact that there thousands of SMPS on the market that manage inrush current limiting somehow, the thread has a decent smell of reinventing the wheel.

 

[cupoftea]

I must admit, i find this very interesting and have a few inrush bias supply PCBs that i was thinking of trying to selling.....(customer provides the inrush resistor)...PCB does the settable delay and short-out with b2b IGBT, and hi side supply for IGBTs.

....But anyway.....i am going to put it forward now, its impossible, with calculation, to calculate, for say a 100-265VAC PSU with say 4.3mF inrush cap'ance.....its literally impossible to calculate exactly the cheapest , smallest resistor that can be reliably used.......it has to be done emperically......the data (eg SHC etc etc, just isnt there.

As you know, Most people will benefit from the fact that there PSU...
1....isnt ON/OFF/ON'd very much
2.....Will not always be ON'd at mains peak
3...The mains wil not often be 265VAC when they switch ON.

During testing, this PSU will be ON/OFF/ON'd frequently, and the NTC will likely be hot.

 

[D.A.(Tony)Stewart]

You would have to know the thermal time constants ** Rth of the metal oxide and the ceramic to estimate the Safe operating Area (SOA) but at least you can test the case with a thermocouple time response.

They are rated for surge but only at 4x rated power for 1 second cycled on and 25s off for 10k cycles

Regarding:
"Periodic-pulse Overload IEC 60115-1 4.39 4 times RCWV 10,000 cycles (1 Sec. on, 25 Sec. off)"

Although these are all constant energy, it is not evenly distributed, due to the series elements in material heat conduction, so it is the thin oxide layer that heats up first and Arrhenius Law dictates there will be rapid aging during this short I^2t event just like in fuses.



Tungsten lamps aren't cycled many times per day but most often blow during inrush, and by lowering voltage to derate the power 25% (miser-lamps) The lifetime is extended greatly by temperature reduction.

--- Updated May 10, 2022 ---

Resistance values > 10 Ohms for the <= 75W versions suffer from lower pulse dissipation capacity I presume due to the surface area reduction for heat spreading to laser cut the spiral resistance and thus a higher Rth for the oxide layer.

I think a suitably derated ICL with a low PTC operating temperature point or with bypass relay and restart delay of the thermal time constant would be suggested. In PC's PSU's they use about a 2 second inhibit restart time for 0.5kW with PTC's.