Should I cool an inrush limiter NTC or not? Advantages/Disadvantages

Go To Last Post
30 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

For inrush current limitation of a 500uF capacitor @540VDC (2 series connected 1000uF 400V rated caps.)

 

I use 5 Ohm NTC with a stedy state current of 30A. The capacity of the NTC is 270J.

 

At first I thought, I place the NTC away from the cooling fan. So that it can have the least resistance.

 

But then. I imagined. What happens if the user switch off the device and switch it on rapidly. ex in 1 sec.

 

My NTC is hot and had no time to cool. So it is possible that a very high inrush current flows.

 

So I came up with the idea that I cool my NTC. I place it downward on the pcb where the air flow. 

 

But this time I suspect that I should not cool too much. Because if I cool too much it can not have low resistance and the efficiency of the circuit is worse.

 

If someone switch the device rapidly off and on, The fan will not stop.(because of momentum). The caps will not have time to discharge and the NTC will cool down rapidly because of the momentum of the rotating fan just in that 1 second that the user switched the device off and on. 

 

I konw the idea sound not pretty cool, But I wanted to hear the opinions if I leave it upside of the pcb where there is no air flow, or place it on the downside of the pcb with air flow but with some enchancement that it will not be cooled completely. (a controlled cooling where the temp does not go down too much but goes down rapidly through the momentum of the fan when current is cut off)

 

 

Any ideas? 

 

 

 

 

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If the unit is on and the cap(s) are charged, then a quick off/on cycle will leave your cap(s) still charged, so not much in-rush current. 

Balance your in-rush NTC's against your bleeder resistors/caps RC discharge time and compare that to your NTC temp curve.

 

Hope that helps, YMMV!

 

 

Jim

 

 

 

(Possum Lodge oath) Quando omni flunkus, moritati.

"I thought growing old would take longer"

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ki0bk wrote:

If the unit is on and the cap(s) are charged, then a quick off/on cycle will leave your cap(s) still charged, so not much in-rush current. 

Balance your in-rush NTC's against your bleeder resistors/caps RC discharge time and compare that to your NTC temp curve.

 

Hope that helps, YMMV!

 

 

Jim

 

 

 

The discharge time of the capacitors through 100kOhm balancing resistors is approx 8 seconds to 1/3 of the Bus voltage.

 

But the cooling time (without forced cooling) of the NTC is 45 seconds. (after 45 seconds NTC builds up a consistant resistor that can block the inrush=

 

I can not close the bridge between 8 seconds and 45 seconds. There is an unsafe zone of 37 second.

 

Which way to go to close this gap?

 

1- way : increasing the value of the balancing resistors (risking the balance of the capacitors) 

 

2-way  : investigating the tricky controlled cooling optimisation of NTC through the forced cooling.

 

 

________________________________ We dream of a world where current does not need the voltage to flow.

Last Edited: Thu. Feb 14, 2019 - 06:22 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It is quite common to see heat shrink tubing over NTC's Presumably to keep them extra warm, so the losses are less, but maybe also what Jim says to balance it with the power supply cap's.

 

It also depends on the appliance. TV's for example are usually just plugged in a few times in their life and it seems that re-plugging a still hot TV is simply called "user error".

Usually this also does not have great consequences and can be looked through the fingers.

 

In other devices this will be completely unacceptable. For example a beefy audio power amplifier with a 500VA toroidal transformer and 40mF capacitors is almost guaranteed to blow the mains fuses or explode the rectifier diodes in such a case.

These have much more rigorous inrush current prevention circuits. It is common to have a (10 Ohm or so) power resistor in series with mains, which gets bridged after a few seconds with a relay after the capacitors are charged. (Most power resistors can easily tolerate a 50x overload during a short time, This is sometimes even specified in their datasheets for purposes like this).

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

Last Edited: Thu. Feb 14, 2019 - 07:04 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

incal wrote:
The discharge time of the capacitors through 100kOhm balancing resistors is approx 8 seconds to 1/3 of the Bus voltage.

 

Really?  My math says RC of 500uF across 100k gets you a time constant of more like 50 seconds.  Maybe there's a lot of drain somewhere else, or leaky capacitors?  S.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If the psu is over 100W and you’re in the EU, then you need power factor correction. This takes care of the inrush current.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Scroungre wrote:

incal wrote:
The discharge time of the capacitors through 100kOhm balancing resistors is approx 8 seconds to 1/3 of the Bus voltage.

 

Really?  My math says RC of 500uF across 100k gets you a time constant of more like 50 seconds.  Maybe there's a lot of drain somewhere else, or leaky capacitors?  S.

 

Sorry the balancing resistor should be 20k.  I thought about disturbing the balance of the series connected capacitors by using 100k balancing resistors instead of 20k to increase the discharging time to match 45 seconds NTC cooling time. The matched value is then 100k. But 100k is a little bit risky. 

 

The discharge time for 0.37 of the Voltage is around 9-10 seconds for 20k. 

 

Allowing a risk for capacitor balancing by using 100k instead of 20k? Or shifting the risk to NTC with 45 seconds cooling time? 

 

 

________________________________ We dream of a world where current does not need the voltage to flow.

Last Edited: Thu. Feb 14, 2019 - 08:13 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Kartman wrote:
If the psu is over 100W and you’re in the EU, then you need power factor correction. This takes care of the inrush current.

 

The PSU is 5000W. But as far as I know the PFC was not mandatory for 3 phase supply also in EU. Is the new regulation also so tight for 3 phase supplies.

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have no idea about the EU requirements for 3 phase.
In the olden days, triacs would be used to implement a soft start on large psus.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Actually, I'm happy to hear the balancing R's are not 100k.  That bothered me too, especially at the voltage and capacitance you were using.

 

Can you stick a block of metal on the NTC?  Something that'll absorb short-term heat transients but not a "heatsink", really.  No fins?

 

Otherwise you might be looking at more complex circuits involving switching FETs and power resistors.  S.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Kartman wrote:
I have no idea about the EU requirements for 3 phase. In the olden days, triacs would be used to implement a soft start on large psus.

 

EN-61000-2-3 dictates a minimum PF of 0.8 for all networks. But the three phase network with suitable passive DC capacitors in worst case shoud not go  under a PF of 0.85.

IMHO if the passive filters are well arranged a 3 phase supplied PSU may not need an active power factor correction to satisfy EN-61000-2-3

 

But I am not %100 sure.

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Scroungre wrote:

Actually, I'm happy to hear the balancing R's are not 100k.  That bothered me too, especially at the voltage and capacitance you were using.

 

Can you stick a block of metal on the NTC?  Something that'll absorb short-term heat transients but not a "heatsink", really.  No fins?

 

Otherwise you might be looking at more complex circuits involving switching FETs and power resistors.  S.

 

Yes, I also tend to implement such a solution. A heat spreader for the NTC that will not function as a perfect heatsink, but decrease the cooling time when current is cut off. This can be material with more contact area (ex.silicon) with relative low thermal conductivity compared to metal 

 

A construction that holds the NTC at 200 degrees instead of 210 degrees when working but spreads the heat quicker when the current is cut off. Maybe a mixture of silicon and metal to find the optimum cooling curve.

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Any way to add a restart timeout?  i.e. once the power switch or loss of power occurs, a timeout must take place before input can be enabled, or a soft start is started? 

 

Jim

 

 

(Possum Lodge oath) Quando omni flunkus, moritati.

"I thought growing old would take longer"

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ki0bk wrote:

Any way to add a restart timeout?  i.e. once the power switch or loss of power occurs, a timeout must take place before input can be enabled, or a soft start is started? 

 

Jim

 

 

This will require an additional relay /contactor or a thyristor. At these power levels they are costly and in the long run because of the worn outs on the contact platings failure sensitive. 

Rather then investing in relay/contactor or a thyristor, I may consider investing in active power correction ic + one IGBT + a pfc diode. They would cost apprx. the same.

________________________________ We dream of a world where current does not need the voltage to flow.

Last Edited: Thu. Feb 14, 2019 - 10:17 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Instead of NTC...how about a beefy N chan fet...use a "slow" rc to charge up the gate (but not toooo slow, so as to not grossly overheat FET)*, then A pulldown diode on the gate quickly resets it  upon loss of power for your next powerup cycle.

 

*want fet hot as an inrush absorber, but yet not burning up

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The main problem with Fet is that it is always there as a series connected device by the time the device working. This means additional power loss and especially for a 540VDC bus voltage,the mosfet rating should be 1000V which is not a main stream stream device. I would use the switch (extra igbt) for the power correction circuit rather than the mosfet. This could also limit the inrush current through the standard PFC inductor.

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The main problem with Fet is that it is always there as a series connected device by the time the device working. This means additional power loss

Perhaps, but I'd expect a full-on fet to drop less than a thermistor-inrush limiter...for example an inrush limiter might drop from 30 ohms to 0.5 ohm as it heats up.  It may be different in the high-power KW world. 

 

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

avrcandies wrote:

The main problem with Fet is that it is always there as a series connected device by the time the device working. This means additional power loss

Perhaps, but I'd expect a full-on fet to drop less than a thermistor-inrush limiter...for example an inrush limiter might drop from 30 ohms to 0.5 ohm as it heats up.  It may be different in the high-power KW world. 

 

 

 

The problem with FET is it should be a high voltage type. And such kind of mosfet should have a high Rdson. (except very pricy and new SiC Mosfets.).

I am planning to add an inductor and enchance this inductor through an IGBT to get a power factor corrector. IMHO , this setup will eliminate the need for NTC. Or the NTC will cooperate with the inductor.

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

    NTC is an easy fix of the inrush current issue for small power cases. Once you go for higher power, like in your case, other methods should be applied.

 

Paulvdh wrote:
It is common to have a (10 Ohm or so) power resistor in series with mains, which gets bridged after a few seconds with a relay after the capacitors are charged.

+1. 

 

    The most efficient PFCs are those that resemble to a boost converter. However, like with any boost converter, there IS an inrush current there. Another disadvantage is that you will end up with a higher DC voltage. Comparing with a resistor and relay, it is significantly less efficient and more costly. You need a big inductor in there, powerful semiconductor, IC, Diode.

 

    I doubt that any solution involving semiconductors can beat the old fashion relay one. The good thing is that the relay never disconnects under load so you don't need a very big one.

 

    You may want to check about crest factor regulations first. Even you have a natural decent power factor, the crest factor is significant. Maybe look how VFD are made ?

 

Last Edited: Sat. Feb 16, 2019 - 04:30 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

angelu wrote:

    NTC is an easy fix of the inrush current issue for small power cases. Once you go for higher power, like in your case, other methods should be applied.

 

Paulvdh wrote:
It is common to have a (10 Ohm or so) power resistor in series with mains, which gets bridged after a few seconds with a relay after the capacitors are charged.

+1. 

 

    The most efficient PFCs are those that resemble to a boost converter. However, like with any boost converter, there IS an inrush current there. Another disadvantage is that you will end up with a higher DC voltage. Comparing with a resistor and relay, it is significantly less efficient and more costly. You need a big inductor in there, powerful semiconductor, IC, Diode.

 

    I doubt that any solution involving semiconductors can beat the old fashion relay one. The good thing is that the relay never disconnects under load so you don't need a very big one.

 

    You may want to check about crest factor regulations first. Even you have a natural decent power factor, the crest factor is significant. Maybe look how VFD are made ?

 

 

If I switch the relay after 0.5-1 sec (after the 10-50Ohm resistor has charged the capacitor), it means that the relay switches at zero current. The control system can then start the PSU after the relay completely closed under 0A. So the contacts of the relay will not worn out. 

Also when shutting down , the control system stops the PSU main Igbts and wait till the DC BUS current returns to zero. Again it can be possible to turn the relay at 0A off.

In this way the relay will last for a long time. 

 

If I think again, this seems not a bad solution.

But I hesitate if I should implement also a PFC stage or not. Because at 5000W the power factor and the crest factor of the current to average current can be so high that I should use an overdimensioned circuit breaker.

 

I investigate some Ics like FAN9672 or UCC28070 (2 interleaved switch with 180 shift). With continuous conduction mode.

 

PFC with 2 or 3 interleaved switch will also reduce the capacitor ripple current and I may use lower value capacitors. The high voltage capacitors are also expensive. 

 

If I can reduce the capacitor value through PFC, the extra cost could balance out. So I will prefer to implement the PFC.

 

But I am not sure. The fan9672 or Ucc28070 has some comparable implementations reaching 5000W , but they are all for 230Vac. (for a boost voltage of 400Vdc). But my bus voltage is 540VDC. So I am not sure if these Ics are suitable for my voltage range.

 

 

________________________________ We dream of a world where current does not need the voltage to flow.

Last Edited: Sat. Feb 16, 2019 - 09:54 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

    With a three phase bridge, the max current should not be that bad. Even if the max current is high, the average current is still under 10A which is what heats the relay contacts. Given the fact that the relay version is so easy to implement, I would give it a try before building a costly and bulky PFC. At 5000W, a such PFC is not small, and it takes time to design, PCB which is critical, attach semiconductors to heatsink, test. The relay option is easy to test. A 16 - 25A may be enough. For test, no need to have them on the PCB. The power dissipation on the relay contacts and coil is way lower than on a 5000W PFC.

 

    

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

angelu wrote:

    With a three phase bridge, the max current should not be that bad. Even if the max current is high, the average current is still under 10A which is what heats the relay contacts. Given the fact that the relay version is so easy to implement, I would give it a try before building a costly and bulky PFC. At 5000W, a such PFC is not small, and it takes time to design, PCB which is critical, attach semiconductors to heatsink, test. The relay option is easy to test. A 16 - 25A may be enough. For test, no need to have them on the PCB. The power dissipation on the relay contacts and coil is way lower than on a 5000W PFC.

 

    

 

I also have concern over the energy loss of the PFC. It is not neglectable.

 

Energy loss on :

The fast recovery diodes (2.5-2.6V x 10A) 

IGBTSs + the loss (3V x 10A)

Also Bypss Diode (3Vx2-3A)

 

There are all extra power loss. I am considering seriously if the passive power correction can be a solution. 

 

I have some room for an input inductor. What abaout adding an inductor at the input. (10mH to 50mH). I get some power factor correction and I limit the inrush current. Something like a PI filter for example? But the values for the inductor at 300Hz is really huge.....

 

What may be the disadvantages of the input inductor?

 

Would you prefer the relay after the rectifier bridge (only 1 contact), Or before the rectifier bridge for all 3 contacts (3 contact relay) so that each contact has lower current. I tend to use 1 relay and 1 resistor after the rectifier bridge.  

________________________________ We dream of a world where current does not need the voltage to flow.

Last Edited: Sun. Feb 17, 2019 - 07:31 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'm a bit confused here.

When I look at the subjects of your last 10+ questions, I start wondering why you are asking these on a forum for AVR uC's

Most people here are into software (and have some hardware / electonics knowledge), but there may be better places for these questions.

Should I cool an inrush limiter NTC or not? Advantages/Disadvantages
Transformator Flux balancing capacitor selection
lm331 cheap alternative as smd
roller bearing guided small aggragtes on elliptical path
SMD ferrite beads to supress noise when driving mosfets
Paralelling 2 BJTs NPN-PNP totem pole
SOT23-6 package, Any experience for DIY reflow soldering
MLCC versus Mica capacitor for high power RF transmitter
74HC86 has schmitt triger inputs or not?
Level shifter or not ?
Schmitt inverter CMOS or LSTTL
5V to 3.3V market trend which to choose for power electronic

Source: https://www.avrfreaks.net/user/11188/content

 

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Paulvdh wrote:

I'm a bit confused here.

When I look at the subjects of your last 10+ questions, I start wondering why you are asking these on a forum for AVR uC's

Most people here are into software (and have some hardware / electonics knowledge), but there may be better places for these questions.

Should I cool an inrush limiter NTC or not? Advantages/Disadvantages
Transformator Flux balancing capacitor selection
lm331 cheap alternative as smd
roller bearing guided small aggragtes on elliptical path
SMD ferrite beads to supress noise when driving mosfets
Paralelling 2 BJTs NPN-PNP totem pole
SOT23-6 package, Any experience for DIY reflow soldering
MLCC versus Mica capacitor for high power RF transmitter
74HC86 has schmitt triger inputs or not?
Level shifter or not ?
Schmitt inverter CMOS or LSTTL
5V to 3.3V market trend which to choose for power electronic

Source: https://www.avrfreaks.net/user/11188/content

 

 

I thought that this is the general electronics topic of the Avr forum. I am a member of this forum since 2001. And when I look at the above questions I see some questions abaout the generel electronics. I have also some entries for AVR software and Hardware under the related topics. 

 

Any link for any better place for this kind of questions is also appreciated. 

 

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Sometimes I forget all those new sections on this forum.

I think they're only here for 10 years or so.

Maybe I'm just becoming an olt fart, blowing in the wind.

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

incal wrote:

 

I thought that this is the general electronics topic of the Avr forum. I am a member of this forum since 2001. And when I look at the above questions I see some questions abaout the generel electronics. I have also some entries for AVR software and Hardware under the related topics. 

 

Any link for any better place for this kind of questions is also appreciated. 

 

 

I noticed that some people here are very knowledgeable on electronics, however, on average, I'd say we are stronger on the software side of things.

That's my case too, so I don't really know a forum where people are very good with general electronics, but maybe try this one: http://www.eevblog.com/forum/ind...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Paulvdh wrote:

Sometimes I forget all those new sections on this forum.

I think they're only here for 10 years or so.

Maybe I'm just becoming an olt fart, blowing in the wind.

 

We are all blowing in the wind. I can not believe that my first entry is from 2001. 

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

El Tangas wrote:

incal wrote:

 

I thought that this is the general electronics topic of the Avr forum. I am a member of this forum since 2001. And when I look at the above questions I see some questions abaout the generel electronics. I have also some entries for AVR software and Hardware under the related topics. 

 

Any link for any better place for this kind of questions is also appreciated. 

 

 

I noticed that some people here are very knowledgeable on electronics, however, on average, I'd say we are stronger on the software side of things.

That's my case too, so I don't really know a forum where people are very good with general electronics, but maybe try this one: http://www.eevblog.com/forum/ind...

 

I also notice the eevblog videos (Dave) and forum from my google searchs. Thanks for the info.

________________________________ We dream of a world where current does not need the voltage to flow.

Last Edited: Sun. Feb 17, 2019 - 06:22 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

incal wrote:
I tend to use 1 relay and 1 resistor after the rectifier bridge. 

    This is what I would do as well.

 

    Do you have some pics of the design so far that you could post here ?

 

    

Paulvdh wrote:
I'm a bit confused here. When I look at the subjects of your last 10+ questions, I start wondering why you are asking these on a forum for AVR uC's Most people here are into software (and have some hardware / electonics knowledge), but there may be better places for these questions.

 

    Come on Paul, you are a heavy contributor on these type of topics. Actually now the discussion is focusing on the idea you presented first: the relay.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

angelu wrote:

incal wrote:
I tend to use 1 relay and 1 resistor after the rectifier bridge. 

    This is what I would do as well.

 

    Do you have some pics of the design so far that you could post here ?

 

    

Paulvdh wrote:
I'm a bit confused here. When I look at the subjects of your last 10+ questions, I start wondering why you are asking these on a forum for AVR uC's Most people here are into software (and have some hardware / electonics knowledge), but there may be better places for these questions.

 

    Come on Paul, you are a heavy contributor on these type of topics. Actually now the discussion is focusing on the idea you presented first: the relay.

 

After some investigation I also saw recommendation for implementation of the relay on some TI datasheets. TI Datasheet recommends a relay paralel to the NTC. This is also a good enchancement. Because it offers some redundancy.

If the relay worns out after long time, the device will continiue to function properly. Only slightly decreasing energy efficiency.

 

For 540VDC Bus voltage 1Ohm or 2.5Ohm NTC + a parallel relay seems the optimum solution. The relay adds an additional enegy efficiency and does not have a critical function on main functions of the PSU.

 

As it is a mechanical device, it is always better to know that if it fails, the device continiues to work without any problem.

________________________________ We dream of a world where current does not need the voltage to flow.

Last Edited: Mon. Feb 18, 2019 - 09:50 AM