Author Topic: Inrush current burns switches  (Read 5190 times)

0 Members and 1 Guest are viewing this topic.

Offline just5554Topic starter

  • Contributor
  • Posts: 20
  • Country: lt
Inrush current burns switches
« on: June 11, 2019, 04:25:44 pm »
Good day, I have a problem with switches. I'm using push lockable illuminated switches. The problem is, when I turn on DPS5005 power unit, there is huge inrush current, that welds contacts. I've tried few switches, both failed after few attempts to turn on.

Video showing how it welds itself: https://drive.google.com/file/d/1-OyGRqnLu5fEOEtO4Wsgc-eF1KHVMxBO/view?usp=sharing

My question is: how to avoid this? (Besides installing not pushable switch) I'm not very keen to install mains voltage switch to switch power supply. 
« Last Edit: June 11, 2019, 04:49:26 pm by just5554 »
 

Offline Gyro

  • Super Contributor
  • ***
  • Posts: 10172
  • Country: gb
Re: Inrush current burns switches
« Reply #1 on: June 11, 2019, 05:18:35 pm »
I'm not really familiar with the DPS5005 but it looks as if you have two problems...

1. The magnitude of the inrush current due to capacitors and load.
2. DC current is notoriously difficult to break.

Contact welding is clearly a number 1 problem (although contact bounce brings number 2 in too).

Your easiest approach is going to be to use the switch to drive a 42V (48V more common) relay with appropriately rated high current contacts. Use a snubber (100R in series with 100nF) across the switch contacts to reduce residual arcing.

A more expensive approach would be to use a DC solid state relay instead of a mechanical one.
Best Regards, Chris
 

Offline wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Inrush current burns switches
« Reply #2 on: June 11, 2019, 05:28:21 pm »
It's not appropriate to use switch there. You have high capacitance on the output of SMPS, and high capacitance on the input of DPS5005. The result is that you basically short cap in SMPS thorough that poor switch. You need inrush current limiting, simply using a switch is not a good idea. And why would you need it there to begin with? You should switch SMPS on the mains side, not between PSUs.
 

Offline just5554Topic starter

  • Contributor
  • Posts: 20
  • Country: lt
Re: Inrush current burns switches
« Reply #3 on: June 11, 2019, 05:43:34 pm »
Thank you for your ideas. I think the best solution would be to buy same switch but rated for 220v and use it to switch whole unit. Although I do not like puting components ordered from china on mains voltages. But oh well, what could go wrong besides current leak to metal enclosure... :-X
 

Offline duak

  • Super Contributor
  • ***
  • Posts: 1048
  • Country: ca
Re: Inrush current burns switches
« Reply #4 on: June 11, 2019, 05:52:53 pm »
It will be difficult to find a switch or even a relay that will not weld closed - the current is so high.

I would not try to use the pushbutton to switch the 42 VDC current, instead I would switch the AC into the supply or control the 42 V supply directly.  Is the 42 V supply from a server?  These supplies usually have some sort of on/off control input.

If I really had to switch the 42 VDC current, I would use a power transistor large enough for the current and switch the base current of the transistor.  If you provide information on the power supply, switch, and the value of the capacitor on the input of the DPS5005, I can give more help.
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: Inrush current burns switches
« Reply #5 on: June 11, 2019, 06:01:12 pm »
Although I do not like puting components ordered from china on mains voltages. But oh well, what could go wrong besides current leak to metal enclosure... :-X

As you somehow limited to components from China? Just buy proper switch from reliable source like Arrows Mouser Farnell
 

Offline wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Inrush current burns switches
« Reply #6 on: June 11, 2019, 06:40:14 pm »
Although I do not like puting components ordered from china on mains voltages. But oh well, what could go wrong besides current leak to metal enclosure... :-X

As you somehow limited to components from China? Just buy proper switch from reliable source like Arrows Mouser Farnell
Even easier, just go to Lemona and buy some decent switch (although they have some crap too).
 

Offline just5554Topic starter

  • Contributor
  • Posts: 20
  • Country: lt
Re: Inrush current burns switches
« Reply #7 on: June 11, 2019, 07:09:55 pm »

Is the 42 V supply from a server?  These supplies usually have some sort of on/off control input.

If I really had to switch the 42 VDC current, I would use a power transistor large enough for the current and switch the base current of the transistor.  If you provide information on the power supply, switch, and the value of the capacitor on the input of the DPS5005, I can give more help.


Power supply is from Epson SX100 printer. Epson 1PM324578. Caps on output: 1000µF 50V, 470µF 50V

Switch is cheap'o from aliexpress, blue, 48V. https://www.aliexpress.com/item/32799185441.html?spm=2114.search0104.3.31.13d57c9dWl1we8&ws_ab_test=searchweb0_0%2Csearchweb201602_6_10065_10068_319_10059_10884_317_10887_10696_321_322_10084_453_10083_454_10103_10618_10307_10820_10821_10303_537_10302_536%2Csearchweb201603_52%2CppcSwitch_0&algo_expid=18ea8d2a-b87c-4b84-887f-0050970a54b3-4&algo_pvid=18ea8d2a-b87c-4b84-887f-0050970a54b3&transAbTest=ae803_4

Caps on DPS5005 input are: 2x 330µF 63V.

Although I do not like puting components ordered from china on mains voltages. But oh well, what could go wrong besides current leak to metal enclosure... :-X

As you somehow limited to components from China? Just buy proper switch from reliable source like Arrows Mouser Farnell
Even easier, just go to Lemona and buy some decent switch (although they have some crap too).

Have bought many components from there and can say, that their products price does not represent quality. Lemona can only provide non-branded switches, 6 times more expensive than from china.
 

Offline wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Inrush current burns switches
« Reply #8 on: June 11, 2019, 09:38:36 pm »
Lemona can only provide non-branded switches, 6 times more expensive than from china.
They have branded switches. Also you can order from Farnell, TME or other catalogs through them. Pay more for parts but you don't pay for shipping compared with ordering directly. Makes sense if you buy only a few cheap things.
 

Offline soldar

  • Super Contributor
  • ***
  • Posts: 3595
  • Country: es
Re: Inrush current burns switches
« Reply #9 on: June 11, 2019, 09:46:57 pm »
I recommend this one.
All my posts are made with 100% recycled electrons and bare traces of grey matter.
 

Offline Terry01

  • Frequent Contributor
  • **
  • Posts: 907
  • Country: scotland
Re: Inrush current burns switches
« Reply #10 on: June 11, 2019, 09:58:39 pm »
I used these for the DPS5030 and they were fine.

http://www.ebay.co.uk/itm/231439337713

Sparks and Smoke means i'm nearly there!
 

Offline lordvader88

  • Frequent Contributor
  • **
  • !
  • Posts: 935
  • Country: ca
Re: Inrush current burns switches
« Reply #11 on: June 11, 2019, 11:17:46 pm »
OP aside from upgrading the switch, do u have any NTC in-rush limiting thermistors ? They work great, as far as I know so far. I have been using them with linear transformers, I'm worried about my big cap's blowing up.

Maybe for now if u could find 1 of those in an old TV or SMPS, and try for now with your switch. Also a temporary fix would be to make a milli-ohm resistor to put in series with the small switch.


Soon I will finally play with the relays I've gotton over the years from TVs etc. Now that will make a nice switch in a multi-PSU I'll put all in the same box.
« Last Edit: June 11, 2019, 11:24:19 pm by lordvader88 »
 

Offline wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Inrush current burns switches
« Reply #12 on: June 11, 2019, 11:47:27 pm »
OP aside from upgrading the switch, do u have any NTC in-rush limiting thermistors ? They work great, as far as I know so far. I have been using them with linear transformers, I'm worried about my big cap's blowing up.
Until you remove power and then switch ON again in a short period of time before NTC cools down.
 
The following users thanked this post: Siwastaja

Offline Terry01

  • Frequent Contributor
  • **
  • Posts: 907
  • Country: scotland
Re: Inrush current burns switches
« Reply #13 on: June 12, 2019, 11:15:05 am »
Just looking at your schematic again... I put my switch on the AC + line between the mains AC 240v and the V+ on the PSU rather than between the PSU and the DPS5005.
I don't know if this is correct or not. Maybe one of the more experienced guys can give their opinion on this?
I am still learning too...
« Last Edit: June 12, 2019, 11:17:05 am by Terry01 »
Sparks and Smoke means i'm nearly there!
 

Online Siwastaja

  • Super Contributor
  • ***
  • Posts: 9328
  • Country: fi
Re: Inrush current burns switches
« Reply #14 on: June 12, 2019, 11:22:13 am »
Until you remove power and then switch ON again in a short period of time before NTC cools down.

This. NTC is touted as an automatic inrush current limiter silver bullet on the forums, but it's a low-cost mass market solution which applies to a fairly small subset of inrush problems, and requires extensive analysis over edge cases to work reliably.

90% of the time, you want an actual inrush current limiting circuit, not an NTC. Whenever NTC happens to work, it's still a lot more than "just use an NTC".
 

Offline duak

  • Super Contributor
  • ***
  • Posts: 1048
  • Country: ca
Re: Inrush current burns switches
« Reply #15 on: June 12, 2019, 07:09:27 pm »
Just5554, please excuse my delay in responding.  I want to give a solution to your problem that will work reliably.

Here is the problem as I see it: the 42 V power supply has 1470 uF of output capacitance.  The DPS5005 has 660 uF of input capacitance.  To charge these capacitors to 42 V, about 0.6 Joules is required (equation: Energy = 1/2*C*V^2).  Unless a lossless device such as a series inductor is used to drop the voltage difference between the two units, the resistance of the wires, switch and capacitors will have to dissipate the same amount of energy that is used to charge the capacitors.

The initial current  is remarkably high; assuming 42 V with a series resistance of 0.5 ohms, one would see 84 Amps.  If a 1 ohm series resistor is added to limit current, the initial current will be 28 A.  This will cause a problem though, at maximum output current of the DPS5005 of 4 A, the 1 ohm series resistor will have to dissipate 16 W.

The switch will take a finite time to close.  As the switch is closing and the contacts  are just beginning to touch, the current is concentrated in a few very small areas.  If the contacts are not capable of handling the current, the material will first melt then solidify thus welding the contacts together. 

here is a link to a problem with relay contacts sticking: https://www.eevblog.com/forum/projects/relay-latching-and-getting-stuck/

An NTC  thermistor could be used but it would have to have a high enough cold resistance to limit current to something the switch could handle, yet it would need a low enough resistance to not drop too much voltage at full current.  Also, there could be a problem when going from zero to full output load current.  The thermal time constant may be too long, and it will limit current until it comes up to temperature.

I could not find more information on the Epson power supply.  What is the maximum output current?  Does it have any control wires in addition to the output wires?

I went to the website for the switch but I cannot see a current rating of any kind.  Does the switch have any ratings printed on it?

What is your experience with electronics?  Would you be able to construct a small circuit with one or two transistors, diodes and resistors?

I've worked out a circuit using a 2N6284 darlington transistor rated for 20 A,  but haven't completed the analysis.
 

Offline wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Inrush current burns switches
« Reply #16 on: June 12, 2019, 07:26:05 pm »
The initial current  is remarkably high; assuming 42 V with a series resistance of 0.5 ohms, one would see 84 Amps.
Resistance should be way lower than 0.5 Ohm
 

Offline jmelson

  • Super Contributor
  • ***
  • Posts: 2851
  • Country: us
Re: Inrush current burns switches
« Reply #17 on: June 12, 2019, 07:51:23 pm »

Power supply is from Epson SX100 printer. Epson 1PM324578. Caps on output: 1000µF 50V, 470µF 50V

Caps on DPS5005 input are: 2x 330µF 63V.
So, when your switch closes, you are connecting a 1000 uF cap charged to 42 V to a 660 uF cap at zero volts.  Assuming 1 Ohm of wiring resistance, the current will be around 42 A, possibly much more if the resistance is less.  Very obvious why the contacts are damaged.  A rule is that half the energy transferred from the charged to the uncharged cap will be lost as heat.  You can figure out how much energy this is, and it will be a LOT.

Jon
 

Offline wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Inrush current burns switches
« Reply #18 on: June 12, 2019, 08:10:19 pm »
Unless wires are very thin or long, expect <200 mOhm including capacitor ESR.
 

Offline rachdatu

  • Regular Contributor
  • *
  • Posts: 54
  • Country: ch
Re: Inrush current burns switches
« Reply #19 on: June 14, 2019, 09:56:50 am »
I built an inrush current limiter for my ebike and it works. I then tried to use the same circuit for these BPS units and it did not work. It just blows the FETs and I don't know why. There is a long thread in the endless-sphere.com forum about these limiters. There are a lot of versions... You can start here (https://endless-sphere.com/forums/viewtopic.php?f=3&t=40142&start=100

For your info, I built the "Inrush Limiter 3" from Richard Fechter with 2 irfb4110.

I built as well a power supply with a dps module and I simply switch from the mains (230VAC) side.
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: Inrush current burns switches
« Reply #20 on: June 14, 2019, 12:45:54 pm »
I built an inrush current limiter for my ebike and it works. I then tried to use the same circuit for these BPS units and it did not work. It just blows the FETs and I don't know why.

It's thermal runaway. VGS for paralleled FETs most likely will differ slightly, they do not start conduction all at the same VGS voltage. When you parallel FET's, especially w/o current-balancing resistors, one will open first, alone dissipating all power and explode, then next FET starts conduction and explode and so on. Some FETs are better for "linear applications", other are notoriously bad. Basically you shall check which FETs are used in high power programmable DC loads and use those, preferably with some current balancing resistors in form of prolonged PCB traces or just wire. There are few discussions about FETs in linear mode here.

For your info, I built the "Inrush Limiter 3" from Richard Fechter with 2 irfb4110.

What are values of components, what is resulting dV/dt? Pay close attention to "Fig 8. Maximum Safe Operating Area" of the irfb4110 datasheet, make sure you operate FET *deep* into "safe operation area".
« Last Edit: June 14, 2019, 12:52:21 pm by ogden »
 

Offline TheWelly888

  • Frequent Contributor
  • **
  • Posts: 344
  • Country: gb
Re: Inrush current burns switches
« Reply #21 on: June 14, 2019, 12:53:02 pm »
That reminds me of my old Amstrad PC (1991 vintage) which had such a high inrush current on powering up that it welded the contacts of the mains socket switch!
You can do anything with the right attitude and a hammer.
 

Offline wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Inrush current burns switches
« Reply #22 on: June 14, 2019, 01:00:13 pm »
It's thermal runaway. VGS for paralleled FETs most likely will differ slightly, they do not start conduction all at the same VGS voltage. When you parallel FET's, especially w/o current-balancing resistors, one will open first, alone dissipating all power and explode, then next FET starts conduction and explode and so on. Some FETs are better for "linear applications", other are notoriously bad. Basically you shall check which FETs are used in high power programmable DC loads and use those, preferably with some current balancing resistors in form of prolonged PCB traces or just wire. There are few discussions about FETs in linear mode here.
With MOSFETS in linear operation simply using current balancing resistors is not nearly enough. Each MOSFET should have individual gate drive with current sense. In this regard they are worse than BJT. On the other hand when using them as switches, you can simply connect them in parallel.
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: Inrush current burns switches
« Reply #23 on: June 14, 2019, 01:36:34 pm »
With MOSFETS in linear operation simply using current balancing resistors is not nearly enough.

For DC load indeed you want to balance dissipated heat as evenly as possible. Yet this is not programmable DC load. Refer to chapter "5. Partially enhanced (linear mode) power sharing":

https://assets.nexperia.cn/documents/application-note/AN11599.pdf

Quote
MOSFETs developed specifically for linear mode operation are available from NXP. They can simplify the design of paralleled groups of MOSFETs intended for linear mode operation.
« Last Edit: June 14, 2019, 01:51:42 pm by ogden »
 

Offline wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Inrush current burns switches
« Reply #24 on: June 14, 2019, 02:03:31 pm »
With MOSFETS in linear operation simply using current balancing resistors is not nearly enough.

For DC load indeed you want to balance dissipated heat as evenly as possible. Yet this is not programmable DC load. Refer to chapter "5. Partially enhanced (linear mode) power sharing":

https://assets.nexperia.cn/documents/application-note/AN11599.pdf

Quote
MOSFETs developed specifically for linear mode operation are available from NXP. They can simplify the design of paralleled groups of MOSFETs intended for linear mode operation.
You either need closely matched MOSFETS for that to work reasonably well or prohibitively high resistance source resistors for low voltage applications if you just use random MOSFETS.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf