Author Topic: Why does SMPS take 20 minutes to power up?  (Read 4198 times)

0 Members and 1 Guest are viewing this topic.

Offline max.wwwangTopic starter

  • Frequent Contributor
  • **
  • Posts: 493
  • Country: nz
Why does SMPS take 20 minutes to power up?
« on: September 10, 2023, 02:46:03 am »
I got a sewing machine to repair, which is said to have a faulty power supply. However, after watching a YouTube Video online and trying the suggested trick, it works all fine. The only thing is, I needed to turn on the power switch (no response at first whatsoever) and leave it there for some time (it's said to be up to one hour) before it powers up. In my case, it took 20 minutes before the machine turned on without any problems. Next time (according to the video, as long as you don't leave it idle for long), it will turn on immediately without any problems.

It's the rubbish SMPS type, not the traditional linear transformer thingie.

What do you think is likely the problem that makes an SMPS work but takes a long time before waking up. Have you seen this kind of behaviour of SMPS before? I've attached photos of both sides of the board to get an idea of its design.

Practically, since now I know the trick, I don't need to fix it. But it's always good to bring it back to the way it should be.

[Edit] With a second thought, it is possible that the PS has no problem but there is some problem in the main board of the machine. Unfortunately, now I have no way to ascertain if this is the case, because without leaving the machine idle for I-don't-know how long it takes, I'm unable to reproduce the problem!
« Last Edit: September 11, 2023, 07:32:48 am by max.wwwang »
Neutral | grounded
 

Online Kim Christensen

  • Super Contributor
  • ***
  • Posts: 1215
  • Country: ca
Re: Why SMPS takes 20 minutes to power up?
« Reply #1 on: September 10, 2023, 03:23:01 am »
[Edit] With a second thought, it is possible that the PS has no problem but there is some problem in the main board of the machine. Unfortunately, now I have no way to ascertain if this is the case, because without leaving the machine idle for I-don't-know how long it takes, I'm unable to reproduce the problem!

Well, you could check the DC voltages coming out of the power supply. If they are all there instantly at turn-on, then the problem is likely elsewhere in the machine.
 

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 2770
  • Country: au
Re: Why SMPS takes 20 minutes to power up?
« Reply #2 on: September 10, 2023, 03:30:16 am »
Also, carefully listen for any faint ticking from the PSU while it's in the fault state.
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 

Offline amyk

  • Super Contributor
  • ***
  • Posts: 8215
Re: Why SMPS takes 20 minutes to power up?
« Reply #3 on: September 10, 2023, 03:30:23 am »
Almost certainly bad caps. Their ESR is far too high when cold, but once they've warmed up they are just barely OK to work.
 
The following users thanked this post: mzzj, max.wwwang, 5U4GB

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 2770
  • Country: au
Re: Why SMPS takes 20 minutes to power up?
« Reply #4 on: September 10, 2023, 03:34:18 am »
Yes, likely the control ICs Vcc capacitor.
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 

Offline fmashockie

  • Regular Contributor
  • *
  • Posts: 205
  • Country: us
Re: Why SMPS takes 20 minutes to power up?
« Reply #5 on: September 10, 2023, 04:10:07 am »
You've got the SMPS board out of the unit.  If you can, power it up to mains (use an isolation transformer if you have one for safety purposes) and check the DC output voltages.  The AC/DC converter (likely IC1) could be going thru an auto-restart cycle (hiccuping) due to some kind of fault on the board.  It might eventually resolve itself as the board warms up like amyk said.

Also, if I am correct in describing the issue, there will come a time when it doesn't turn on at all. Ever.  And you will need to fix it.
« Last Edit: September 10, 2023, 04:13:37 am by fmashockie »
 

Offline asis

  • Regular Contributor
  • *
  • Posts: 214
  • Country: ru
Re: Why SMPS takes 20 minutes to power up?
« Reply #6 on: September 10, 2023, 07:16:50 am »
Hi,

It’s trite, this ECAP has degraded.
 

Offline RoGeorge

  • Super Contributor
  • ***
  • Posts: 6080
  • Country: ro
Re: Why SMPS takes 20 minutes to power up?
« Reply #7 on: September 10, 2023, 07:17:10 am »
Replace the electrolytic capacitors.  Start with the ones working at high frequency and low voltage.  The ones after the mains bridge are usually OK.  The smaller ones dry out, or leak, or whatever they do that they increase their ESR.  The electrolytics placed near hot components tend to fail first.  Higher temperatures damage them faster.

However, for damaged capacitors the ESR lowers with temperature increase, and that might explain why the SMPS eventually starts after 20 minutes.

Don't bother much measuring each electrolytic capacitor, just replace all the smaller ones.
 
The following users thanked this post: max.wwwang

Offline max.wwwangTopic starter

  • Frequent Contributor
  • **
  • Posts: 493
  • Country: nz
Re: Why SMPS takes 20 minutes to power up?
« Reply #8 on: September 10, 2023, 07:37:59 am »
Also, carefully listen for any faint ticking from the PSU while it's in the fault state.

Where would this ticking be coming from? Are there relays in the circuit? I don't think so. Keen to know. Thank you.
Neutral | grounded
 

Offline max.wwwangTopic starter

  • Frequent Contributor
  • **
  • Posts: 493
  • Country: nz
Re: Why SMPS takes 20 minutes to power up?
« Reply #9 on: September 10, 2023, 07:39:08 am »
Well, you could check the DC voltages coming out of the power supply. If they are all there instantly at turn-on, then the problem is likely elsewhere in the machine.

Yes, I can check this to find out if the problem happens again next time.
Neutral | grounded
 

Offline wraper

  • Supporter
  • ****
  • Posts: 16690
  • Country: lv
Re: Why SMPS takes 20 minutes to power up?
« Reply #10 on: September 10, 2023, 07:42:36 am »
Practically, since now I know the trick, I don't need to fix it. But it's always good to bring it back to the way it should be.
Until a year later this trick will no longer work. I'd say most likely it's a small electrolytic capacitor nearby transformer, with less probability capacitors on the output. But it's better to first measure actual output voltage to confirm if it's actually a PSU issue to begin with.
 
The following users thanked this post: magic

Offline max.wwwangTopic starter

  • Frequent Contributor
  • **
  • Posts: 493
  • Country: nz
Re: Why SMPS takes 20 minutes to power up?
« Reply #11 on: September 10, 2023, 07:44:35 am »
Hi,

It’s trite, this ECAP has degraded.

Good to hear; that's promising! And thanks for pinpointing the ECAP. Its values are 25V 100uF.
Neutral | grounded
 

Offline max.wwwangTopic starter

  • Frequent Contributor
  • **
  • Posts: 493
  • Country: nz
Re: Why SMPS takes 20 minutes to power up?
« Reply #12 on: September 10, 2023, 07:53:56 am »
Yes, likely the control ICs Vcc capacitor.

There are two 8-pin ICs on the board, one is UC3842B (bottom right in the photo, near the AC input), which is a "HIGH PERFORMANCE CURRENT MODE PWM CONTROLLER". Another I'm not sure, its markings are:

ST(logo) CHN
063AC
K2X651

Which IC do you think have the faulty Vcc capacitor, and why? Thank you.
« Last Edit: September 10, 2023, 08:11:31 am by max.wwwang »
Neutral | grounded
 

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 2770
  • Country: au
Re: Why SMPS takes 20 minutes to power up?
« Reply #13 on: September 10, 2023, 08:01:06 am »
Also, carefully listen for any faint ticking from the PSU while it's in the fault state.

Where would this ticking be coming from? Are there relays in the circuit? I don't think so. Keen to know. Thank you.
The control IC, likely UC3842, is supposed to become active only when the Vcc capacitor, as pointed out, has charged to 16V from the HVDC via the resistor next to it. If there is a fault that prevents sustained operation of the control IC, it will stop and then try again when the Vcc capacitor charges to 16V again. The repeated restarts usually cause ticking from the transformer.
It's possible for the Vcc capacitor itself to be the cause.
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 
The following users thanked this post: max.wwwang

Offline magic

  • Super Contributor
  • ***
  • Posts: 6721
  • Country: pl
Re: Why SMPS takes 20 minutes to power up?
« Reply #14 on: September 10, 2023, 08:15:07 am »
The capacitor near UC3842 would be my first suspect too.

My first thought was that perhaps it fails to reach 16V at all and hence the controller never starts, until it the cap has been powered for a very long time.
But I suppose if capacitance is too low and/or ESR too high the capacitor could also fail to hold charge delivered from the AUX winding and cause PSU restarts.

Another common reason for continuous PSU restarts, not related to the capacitor, is overload or short circuit on the output.
 

Offline max.wwwangTopic starter

  • Frequent Contributor
  • **
  • Posts: 493
  • Country: nz
Re: Why SMPS takes 20 minutes to power up?
« Reply #15 on: September 10, 2023, 08:23:11 am »
The capacitor ESR change when warming up theory makes sense. The strange thing here is that even when the culprit (supposedly) capacitor has completely cooled down, when I plug it in and turn it on it works immediately without hesitation. According to the video I mentioned, as long as you don't leave it idle for very long (I take it as, say, more than a year), and keep regular use of the machine, then it works just fine. Since I got this machine only yesterday, so I don't have experience yet as to how long idle time will trigger this fault. All that I have experienced so far is, since the 20 minutes of 'warming up' for the first time, it behaves as if it had no problems at all!
Neutral | grounded
 

Offline max.wwwangTopic starter

  • Frequent Contributor
  • **
  • Posts: 493
  • Country: nz
Re: Why SMPS takes 20 minutes to power up?
« Reply #16 on: September 10, 2023, 08:27:07 am »
The capacitor near UC3842 would be my first suspect too.

My first thought was that perhaps it fails to reach 16V at all and hence the controller never starts, until it the cap has been powered for a very long time.
But I suppose if capacitance is too low and/or ESR too high the capacitor could also fail to hold charge delivered from the AUX winding and cause PSU restarts.

Another common reason for continuous PSU restarts, not related to the capacitor, is overload or short circuit on the output.

Basically what you are saying here is that what looks like the process of the PSU 'warming up' is simply its repeated restart due to a failure somewhere. Is that correct?

Since the machine is now working (supposedly with the warmed-up capacitor) the way it should, can we reasonably rule out the possibility of overload of short circuit of the user side of the PSU -- that is the main board of the machine?

Thanks.
Neutral | grounded
 

Offline magic

  • Super Contributor
  • ***
  • Posts: 6721
  • Country: pl
Re: Why SMPS takes 20 minutes to power up?
« Reply #17 on: September 10, 2023, 09:16:10 am »
Either continuous restarting or not starting at all. At any rate, it's clear that there is no major damage if everything works normally after those 20 minutes.

It is said that old electrolytic capacitors can sometimes be "fixed", or at least improved, simply by charging them. It's called "capacitor re-forming" and it appears to apply to capacitors with damaged oxide layer and hence reduced breakdown voltage and increased leakage current. The flow of leakage current triggers an electrochemical reaction which consumes electrolyte (if there is anything left to consume) and rebuilds the oxide layer, until the cap stops leaking and holds its rated voltage again.

This could explain why waiting gets the machines back to operation. But if the problem repeats, it is obviously better to locate and replace the failing capacitor. I would also consider replacing other similar capacitors on this PCB.
 
The following users thanked this post: abraxa, max.wwwang

Online tautech

  • Super Contributor
  • ***
  • Posts: 27908
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
Re: Why SMPS takes 20 minutes to power up?
« Reply #18 on: September 11, 2023, 01:16:35 am »
Yes, likely the control ICs Vcc capacitor.

There are two 8-pin ICs on the board, one is UC3842B (bottom right in the photo, near the AC input), which is a "HIGH PERFORMANCE CURRENT MODE PWM CONTROLLER". Another I'm not sure, its markings are:

ST(logo) CHN
063AC
K2X651

Which IC do you think have the faulty Vcc capacitor, and why? Thank you.
As others have said, the SMPS PWM controller IC.
These often fail in older gear and usually they are charged via a resistive dropper chain from HV DC then when the UVlockout point is exceeded the IC kicks off for its VCC to be sustained by the flyback winding on the transformer via halfwave rectification.
If the VCC cap is not in good order (leaky and/or diminished capacitance) the IC supply will fail to meet UVlockout and the IC will not start.
It's a good idea to replace these caps with a Low ESR equivalent to better withstand the charge pulses from the flyback winding.

Whenever you have successive starts kick any DUT into life start checking caps, properly checking them too not just in circuit.
To better understand how SMPS work download and study the controller IC datasheet and the often included typical application schematic as so often it's closely copied by the SMPS manufacturer.

SMPS ticking symptoms are another matter where normally the overcurrent sense is kicking in and out and the noise originates from transformer magnetizing currents building to overcurrent limits then failing, rinse and repeat.
This may stem from dud output caps or a short/excessive load further from the supply.
Avid Rabid Hobbyist
Siglent Youtube channel: https://www.youtube.com/@SiglentVideo/videos
 
The following users thanked this post: abraxa, max.wwwang

Offline max.wwwangTopic starter

  • Frequent Contributor
  • **
  • Posts: 493
  • Country: nz
Re: Why SMPS takes 20 minutes to power up?
« Reply #19 on: September 11, 2023, 07:32:29 am »
As others have said, the SMPS PWM controller IC.
These often fail in older gear and usually they are charged via a resistive dropper chain from HV DC then when the UVlockout point is exceeded the IC kicks off for its VCC to be sustained by the flyback winding on the transformer via halfwave rectification.
If the VCC cap is not in good order (leaky and/or diminished capacitance) the IC supply will fail to meet UVlockout and the IC will not start.
It's a good idea to replace these caps with a Low ESR equivalent to better withstand the charge pulses from the flyback winding.

Whenever you have successive starts kick any DUT into life start checking caps, properly checking them too not just in circuit.
To better understand how SMPS work download and study the controller IC datasheet and the often included typical application schematic as so often it's closely copied by the SMPS manufacturer.

SMPS ticking symptoms are another matter where normally the overcurrent sense is kicking in and out and the noise originates from transformer magnetizing currents building to overcurrent limits then failing, rinse and repeat.
This may stem from dud output caps or a short/excessive load further from the supply.

Thanks. Will have a close look at the datasheet and application notes. I watched a couple of videos of Dave not very long ago and I think understood the basics of SMPS, but unfortunately most of it is forgotten!
Neutral | grounded
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 27908
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
Re: Why does SMPS take 20 minutes to power up?
« Reply #20 on: September 11, 2023, 07:45:26 am »
Just remember UVlockout must be surpassed for SMPS to start.
They must have a startup VCC usually via a dropper network from HV DC and these dropper resistors go open too.
The VCC cap must be sound for the IC to start.

Normal failure points are in no order:
HV DC dropper resistor/s
VCC cap
Flyback diode

Not lots to check but if it's a tick fault then you can normally assume the primary side is okay but while in there always check the IC VCC cap, you'll be surprised how often you find sick ones about to give problems.
Avid Rabid Hobbyist
Siglent Youtube channel: https://www.youtube.com/@SiglentVideo/videos
 

Offline max.wwwangTopic starter

  • Frequent Contributor
  • **
  • Posts: 493
  • Country: nz
Re: Why does SMPS take 20 minutes to power up?
« Reply #21 on: September 11, 2023, 07:19:30 pm »
Just remember UVlockout must be surpassed for SMPS to start.
They must have a startup VCC usually via a dropper network from HV DC and these dropper resistors go open too.
The VCC cap must be sound for the IC to start.

Normal failure points are in no order:
HV DC dropper resistor/s
VCC cap
Flyback diode

Not lots to check but if it's a tick fault then you can normally assume the primary side is okay but while in there always check the IC VCC cap, you'll be surprised how often you find sick ones about to give problems.

That really covers everything. Thank you very much. Will have a poking around with my homemade isolation transformer, DMM, and scope, and take this opportunity to refresh my knowledge on this subject. Have a great spring day sir.
Neutral | grounded
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 27908
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
Re: Why does SMPS take 20 minutes to power up?
« Reply #22 on: September 11, 2023, 07:40:00 pm »
Consider powering just the SMPS IC with an external LV supply to be able to check it's functionality without having HV DC to contend with otherwise a differential probe is required as these things are generally at elevated voltages.

Safest approach is checking the passives and diodes when unpowered and then power up with a dim bulb tester to better contain any magic smoke.
Avid Rabid Hobbyist
Siglent Youtube channel: https://www.youtube.com/@SiglentVideo/videos
 

Offline max.wwwangTopic starter

  • Frequent Contributor
  • **
  • Posts: 493
  • Country: nz
Re: Why does SMPS take 20 minutes to power up?
« Reply #23 on: September 12, 2023, 04:38:13 am »
Consider powering just the SMPS IC with an external LV supply to be able to check it's functionality without having HV DC to contend with otherwise a differential probe is required as these things are generally at elevated voltages.

Safest approach is checking the passives and diodes when unpowered and then power up with a dim bulb tester to better contain any magic smoke.

Great point (safety first, in terms of both personal safety and safety of DUT). With regard to LV, a few more questions:

What range of voltage would be good for the test/diagnosis purposes?
Would a stepped down low AC input from the AC plug achieve the same thing? I believe so. That would be easier.
Neutral | grounded
 

Offline magic

  • Super Contributor
  • ***
  • Posts: 6721
  • Country: pl
Re: Why does SMPS take 20 minutes to power up?
« Reply #24 on: September 12, 2023, 05:00:03 am »
Will have a poking around with my homemade isolation transformer, DMM, and scope, and take this opportunity to refresh my knowledge on this subject.
This can be as dangerous to the PSU as it is to yourself. If you really must play with a switcher and see how it works, use some other junk that you don't care about.

If this PSU ever acts up again, simply measure VCC voltage to UC3842 with an isolated DMM and replace that stupid capacitor if it's less than 16V or whatever is the UVLO turn-on threshold specified in the datasheet.

Or simply replace the stupid capacitor right away and forget about it...
Again, other junk can be used for experiments if you want to understand how things work.
« Last Edit: September 12, 2023, 05:02:38 am by magic »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf