Chinese DVR Power Supply

[hgg]

Hi guys!

I have a chinese DVR that just stopped functioning.  It turns out that it was its external power supply.
Its a 19V, 3A switching power supply.  One of the capacitors leaked.  I measured its ESR and its way
too high.  All other capacitors are fine.


Faulty Capacitor

Below are some photos of the PCB:


PCB Top


PCB Underside

The faulty capacitor has a value of 470?F @ 35V but I only have a 680?F @ 35V one.
My question is, do you think that its ok to replace the faulty one with a larger capacity value?

I've read that in some situations that this might not be a good idea, such as:

-- 1) If total capacitance is too high, the turn-on surge current could be high
        enough to trigger the OCP (over-current protection) in the ATX supply and           
        on the on-board regulators themselves.

-- 2) In this type of circuit, incorrect values can cause parasitic oscillation, which
        will cause over-current, overheating and premature cap failure.

If this capacitor is a filtering one maybe it might be ok to replace it with a larger one.
Its near the output stage of the power supply.

What do you think?

Thank you!
George.

[SeanB]

If it fits and is a low ESR type it will work. If this one has done a pop then the other ones will be close to the same stage as well though, they will need it as well.

[hgg]

If it fits and is a low ESR type it will work. If this one has done a pop then the other ones will be close to the same stage as well though, they will need it as well.

Hello SeanB!

The 680?F I have, has a 0.07? ESR and the other two have 0.015? ESR.
The other two look that they are in pretty good condition.

What do you think?

Thanks!

[SeanB]

The others are probably after a LC filter, so they do noty have the ripple current, so just place the one you have in, and look to see if it works. If so then order the 3 as low ESR devices and do the lot.

[hgg]

I've just replaced it with the one I had and it works, but when I connect it to the scope
I get the following 200mV peak to peak signal spikes:  (sorry about the video quality)



This is the first time that I connect this power supply to the scope so I don't know if
that was happening before as well.  Do you think that this might be the result of the
bigger sized capacitor or is it just a crappy power supply...?

Thanks.
George.

[SeanB]

Probably a combo of both a junky power supply and the other caps being no longer any good any more ( if they were in the first place which is doubtful) and barely doing any filtering. Most likely replacing them with any other cap that fits ( 100-470uF 35V) will improve things a lot. If you have any high value ceramic SMD caps of an appropriate voltage rating try soldering one across each capacitor as well.

[hgg]

Ok SeanB, thank you!

I think I will replace them all with some new Panasonic ones and see what happens.
I might try your suggestion for the SMD ceramics as well.

Thanks again!
George.

[hgg]

I started testing all the switching power supplies I had and interestingly enough all of them had
similar problems with the transients at different frequencies.  The best was the one from s SONY
VAIO with very little ripple and the smallest transient of about 164mV Peak to Peak.


snagit

I also tested an IBM and a noname Chinese one that had the worst peak to peak of around 800mV...
So it seems that the power supply of the DVR had a value (180mV) in the low range.

Is this an inherent problem of all switching power supplies?

George.

[SeanB]

Yes, you can reduce it with a few LC filter stages, but it will always be there.

[wraper]

Is this an inherent problem of all switching power supplies?

George.

I doubt that they have any problems. It's all about how you measure them. Watch this video:
And SMPS is a source of noise itself. Measuring SMPS transients without load is a useless waste of time BTW.

[hgg]

Is this an inherent problem of all switching power supplies?

George.

I doubt that they have any problems. It's all about how you measure them. Watch this video:
And SMPS is a source of noise itself. Measuring SMPS transients without load is a useless waste of time BTW.

Hi,

But David was measuring a linear power supply and the actual noise came from a switching power
supply!  So switching power supplies have this kind of problem especially if they are not well designed.

[amyk]

I also tested an IBM and a noname Chinese one that had the worst peak to peak of around 800mV...

Are these for laptops? 800mV out of around 20V is 4%, nothing to worry about for a rail that gets further stepped down inside the laptop to 12, 5, 3.3, 2.5, 1.x, etc. In fact most laptops will run fine even with an input of 12-15V.

[wraper]

Hi,

But David was measuring a linear power supply and the actual noise came from a switching power
supply!  So switching power supplies have this kind of problem especially if they are not well designed.

Situation is similar and don't forget that this is SMPS which generate common mode noise by itself. No doubt they have big ripple on output, something like <200mV but not 0,8V. So measuring SMPS output with long wires and without load will show something horrible but not what actually is there. For example, measuring buck converter output on PC motherboard can often show something like 10V transients on 1,5V rail. But they are not there for sure, otherwise ICs on that rail would be dead. Also without load output voltage may be unstable. Anyway only measuring properly with load will show how SMPS works on it's intended usage scenario.

[hgg]

Hi,

"amyk", yes all the switching power supplies were from laptops from 16V to 19.5V.

"wraper", maybe your are right.   Will I have a more accurate result if I measure them with a
power resistor as a dummy load ?

[wraper]

Hi,

"amyk", yes all the switching power supplies were from laptops from 16V to 19.5V.

"wraper", maybe your are right.   Will I have a more accurate result if I measure them with a
power resistor as a dummy load ?

There may be some "fake" transients because of the wire loops but result should be near to what there is actually. You may also want to look on a max load because it will show how much manufacturer saved on capacitors because ripple will increase with load.

[hgg]

I see.  Ok, thanks I will try it.