Digital Power supply protection help needed!

[KAKAH]

hello i am using a digital psu from dell, the model is da-2 its a 12v supply with all kinds of protection and a stand by mode. the problem is that its quite noisy and when used for a longer period of time, it kills the small psu that powers up my computer from the 12v. my idea was to put huge capacitor on the 12v line to make it stable and i got 2x47000uf16v in parallel with the load. everything works perfect till the point i put the caps to the line, instantly kicks off the protection with a big spark. what could be wrong with using the caps to stabilize the circuit?? could a diode work between the psu and the caps? what could be the thing that makes the protection go off?

 

[cowboybob]

hello i am using a digital psu from dell, the model is da-2 its a 12v supply with all kinds of protection and a stand by mode. the problem is that its quite noisy and when used for a longer period of time, it kills the small psu that powers up my computer from the 12v. my idea was to put huge capacitor on the 12v line to make it stable and i got 2x47000uf16v in parallel with the load. everything works perfect till the point i put the caps to the line, instantly kicks off the protection with a big spark. what could be wrong with using the caps to stabilize the circuit?? could a diode work between the psu and the caps? what could be the thing that makes the protection go off?

(BOLD emphasis is mine)

Are the cap(s) connected to the 12VDC line before, or after, you power up the PSU?

If after you're, essentially, putting a dead short across the 12VDC line.

Discharged caps that size would represent, in that first instance of being attached, essentially a very big, practically zero ohm load. Little wonder there's a "big spark". Which is probably what's causing the protection circuit(s) to kick in.

Try connecting the cap(s) with the PSU turned off. Then try powering up and see what happens.

Be CAREFUL. Those caps can bite...

 

[KAKAH]

thank you for your advice, ill be more cautious. yes the spark was only there when i tried to apply the caps after turning the supply on, the original plan was for the caps to connect trough a double switch at the same time turning the psu from standby to ON, witch i understand wasnt a good idea, but still the protection instantly turns the supply off, even if the caps are on the load before turning it on. So the bigger question is How do i make this turning on part with the caps on more smooth,easy for the psu to not set the protection off?

 

[cowboybob]

Try putting a low (say, maybe 100 ohms, or less) 5watt (or higher) resistance in series with the cap(s), on the + side of the PSU.

That reduce the inflow current.

 

[KAKAH]

or what exactly is that could be setting the protection off? the huge load?

 

[KAKAH]

yea the thing is power supply is 18a
intheory the max w the computer is 200w . <couldnt a 5w resistor burn or really heat up?

 

[Mosaic]

Here is your problem & solution.
https://www.newark.com/pdfs/techarticles/ametherm/Ametherm_PPT_FEB07_REV6DistyTutorial.pdf

 

[cowboybob]

yea the thing is power supply is 18a
intheory the max w the computer is 200w . <couldnt a 5w resistor burn or really heat up?

Only if it were continuous.

The RC time constant of 100 ohms and 52,000uF is about 6 mS. Pretty short and no where near enough to heat up the resistor. Once the caps are charged, current drops to a very small level.

Matter of fact, you might even increase the resistance some. Play with it.

More I think about it, 1 or 2,000uF would would probably do it. Just don't know the level of noise you're dealing with.

 

[KAKAH]

ok so i need an ntc thermistor how would i calculate the inrush current for the capacitors of 2x47000uF 16v when i have a load in parallel of maximum 200w and the optimal resistance . max the supply is supposed to handle is 18a . and as i understan i should put the thermistor in parallel with the cap?

 

[KAKAH]

well 10000uf wasnt really enough but did make a quite difference but i didnt have problems like this

 

[KAKAH]

so cowboybob how would i calculate the perfect value? id really love to buy the right components with the first try as its so expensive here.
iknow basic ohms law but how would i calculate the inrush current of my cap bank and the reduction of it that the added resistor makes?

 

[KAKAH]

and does a regular resistor take power from the supply all the time? as i understood thermistor gradually decreases(or increases?) the resistance so after reducing the first peak and then works as it would not be there?

 

[Mosaic]

This will help you size the inrush thermistor:
https://www.soselectronic.hu/a_info/resource/a/pdf/ntc.pdf

Do NOT place it in parallel with the cap! You can place them in parallel with each other to increase current handling capacity.
The inrush thermistor , will drop in resistance as current demand heats it up, thereby providing lower voltage drop & heating than a simple resistor. They generally run a bit hot to touch.

Edit: it might be a good idea to tie the both the +ve & -ve low volt rails to Earth via two 0.1uF WIMA film caps. to handle HF switching noise harmonics that might escape the electrolytics due to their ESR and higher Dielectric Absorption. Also, placing a 10K resistor across the electolytic terminals to ensure removal of charge once power is off is useful.

 

[cowboybob]

Look at page 12 of the site Mosaic provided. use the upper example first. And NO, the NTC thermistor is not placed parallel to the cap, rather in series.

Until I can get more information about the problem(s) you're experiencing with the PSU (beyond the added capacitor surge issue) I'm not sure what to recommend. On the surface of it, 52,000uF caps seems like major overkill for a 200W PSU (essentially a PC power supply).

Do you have access to an oscilloscope? It would be useful to see the condition of the DC output(s) of the PSU.

Just saw Mosaic's post. I type entirely TOO slowly...

 

[KAKAH]

thank you guys for your feedback a lot, ive learned quite from this.
well i think the supply works good and isnt broken but it is still noisy as most of the digital ones, and i need it to really be as inaudible as possible, when i tried with 10k uf it made a big difference, but still not even close to what i need. I have a sound interface with max input of 3v and more aimed at the audible spectrum that i can use as a rough oscilloscope in my digital audio workstation. max 96khz sampling rate.

i try as much as i can, but i still cannot fully grasp the calculations for the thermistor. as i understand i need the biggest one( s464 8-20amps,1-10 ohms, 6.7w) is only 10 ohms gonna be enough to limit this current? 6.7w? i think id use max 160-170w of power when doing something cpu heavy. and it wouldnt be to nice if the thermistor limiter would run hot all the time even when finished doing its job(as this is a computer).
i dont see the big image of variables this limiter would have to handle and for how long times.
Could i maybe use a resistor for the few miliseconds and then have some simple schematic with relay for instance that shorts the resistor out of the circuit when the cap current drops? as i understand thermistor does this by its thermal properties by itself but the temprature should be as low as possible in my case.
**broken link removed**

another thing, could i limit this current(and have the volt drop) only for the capacitors in the circuit and have the load before the limiter and caps, but still get the stabilization??

it might be a good idea to tie the both the +ve & -ve low volt rails to Earth via two 0.1uF WIMA film caps. to handle HF switching noise harmonics that might escape the electrolytics due to their ESR and higher Dielectric Absorption. Also, placing a 10K resistor across the electolytic terminals to ensure removal of charge once power is off is useful.

thanks for the ideas. where do i find the low voltage rails?(pardon my lack of knowledge ) and wouldnt the resistor eat my current and overheat with the power im dealing in this circuit while in use?

 

[Mosaic]

KAKAH: I begin to think that you are guessing at what you're doing, so let's back up a bit.
Correct me if necessary, but I assume you don't know about electronic fundamentals such as V=IR, P=I*I*R or about things such as low pass filters for attenuating high frequency noise?
Perhaps you should seek some hands on assistance from a more experienced person, since if you don't have the vocabulary to understand the terminology, you are going to make mistakes that could be costly when dealing with hundreds of watts of power.


That said:
Switching power supplies or SMPS emit higher frequency noise in the multi Khz range as opposed to 120Hz AC hum found in linear supplies. Therefore a low pass LC filter should make a big difference attenuating the noise.
Since you don't have access to an oscilloscope you can't actually see or quantify the noise.
Please read here:
https://www.eevblog.com/forum/beginners/lc-filter-basics-for-smps/

If this is too difficult for you, then you shouldn't be doing this. Get someone to help who understands.
Also download LTspice as a free simulator to model what you're doing without melting anything or making smoke!
https://www.linear.com/designtools/software/

 

[Mosaic]

BTW , when sizing your inductors note that you can parallel them for higher current capacity, doing so reduces their value as shown:
https://en.wikipedia.org/wiki/Series_and_parallel_circuits#Inductors_2

Jameco carries some decent toroid inductors, plus its possible to wind your own, but that's another tutorial to learn.
https://www.jameco.com/webapp/wcs/s...d=10001&freeText=toroid&search_type=jamecoall

SMPS noise
https://www.mastec.co.nz/JST/PK8.html

Here is some info on handling common mode noise:
https://www.murata.com/products/emc/knowhow/pdf/26to30.pdf

 

[KAKAH]

well thanks alot mosaic! do you think its worth building a filter without seeing the noise on a scope? like would a 3khz lp could be effective even if the problem lies much much higher? wouldnt the coils have to be massive for this current?

 

[Mosaic]

At a peak 18 A load you can use 4 of these in parallel to get u to 20A and about 97 uH inductance. Use that in your filter calcs.
https://www.jameco.com/webapp/wcs/s...-1&catalogId=10001&pa=387330&productId=387330

The lower the LPass the better. Also tying the +ve line and -ve line to earth with 0.1 uF caps will help with the noise control as well.

BTW inductors also help smooth surge currents.

 

[Mosaic]

Hi Again:
I did a sim in LTspice with a .02 ohm impedance DC supply and an overlaid 1 ohm impedance, 0.4V, noise signal @ 10Khz into a 1 ohm load for about a 12A draw. Using a simple LC filter with the 4 parallel Jameco toroids (net .018 ohm series resistance) and a .04 ohm ESR, 220uF average electrolytic capacitor you realize a 3db attenuation starting at 1Khz. https://circuitcalculator.com/lcfilter.htm

The light blue trace shows the power up from 0V @ 1 millisec and the superimposed 10Khz 'noise' = thick line.
The red trace is the filtered V after the LC filter. Note it settles at 3ms and is quite clean.
The green shows the current through the inductor and the teal trace is the current flowing in the capacitor. You can see the startup current surge into the cap is almost 18Amps. The Flux eq for the inductor simulates saturation @ 18Amps load.