Capacitors in a GM34063 controlled switching power supply

[carpenter]

I had a D'Link DL-1008 switch on the wall for over 10 years. It stopped working over the weekend. The fault was clear at first sight, electrolytic capacitors.
Two input 220uf/25V inflated and had around 10uF Replaced 2x 470uF/25V ESR 350mOhm
Switch source output capacitors 2x100uF 25V looked OK but were also around 10uF and were replaced with 2x220uV 16V ESP under 500mOhm. The output voltage is only 2.5V
In the switch there are two more electrolytic capacitors 100uF 25V. , also 15uF were defective , replaced with 220uF ESR under 500mOhm. The voltage on the capacitors is around 1.85V 2,5 minus the loss on the PNP transistor UTC 2SB772L.
Stepdond converter is drived by GM34063 1,5A input 3-40V 100kHz.
Where is the problem?
Switch works again but freezes
Two or three times a day it just freezes and needs to be reset
The adopter is a classic transformer with 4 diodes and a 1000uF 16V capacitor, measured 860uF ESR 380mOhm
The converter (probably the coil) can be heard a little, with the ear at max 5 cm you can hear whistling

A higher capacitor capacity shouldn't be a problem in this case, or am I wrong?

 

[hiphei]

My tips; put some flux on and resolder *all* powersupply smd components (around ic and ic included).. to my eye they look very bad.. you don't need much solder.. plenty of flux and just some solder on a tip of a soldering iron.. at least the resistors, controller ic, and q2 (maybe a switch?) I'd also confirm *all* powersupply electrolytics are replaced.. And all the other values you replaced maybe still alright, but i think replacing 10uF caps with that large may present an issue..i'd go back there but with new ones..!

--- Updated Jun 25, 2024 ---

My tips; put some flux on and resolder *all* powersupply smd components (around ic and ic included).. to my eye they look very bad.. you don't need much solder.. plenty of flux and just some solder on a tip of a soldering iron.. at least the resistors, controller ic, and q2 (maybe a switch?) I'd also confirm *all* powersupply electrolytics are replaced.. And all the other values you replaced maybe still alright, but i think replacing 10uF caps with that large may present an issue..i'd go back there but with new ones..!

Tip for myself..put the glasses on when reading.. (..there were no 10uF caps in the product..)

 

[KlausST]

Hi,

to be honest ... I don´t consider 500mOhms as LowESR. There easily may be switching currents with peaks above 1A ... so 500mV voltage ripple, which is huge.
Even if 2 are in parallel.

ESR is not the only thing you have to care for when selecting a capacitor. It also needs to be designed for the switching frequency (and the ESR should be specified at high frequency) and it needs to widthstand the ripple current.

I´d use a scope to verify the power supply voltages´ ripple and noise.

If you encounter a HF noise/ripple at the electrolytics ... you may solder additional ceramics in parallel .. check again with the scope if they improve the situation.

Klaus

 

[carpenter]

So I looked at it with an oscilloscope and made some adjustments.
The noise on the input capacitors was approx. 400mV
I provisionally added a 1000uF/16V KZH to the unused spot for the diodes
The noise dropped to about 250mV.
the noise on the output capacitors was about 70 mV
replaced one capacitor LowESR 1000uF 6.3V.
The noise dropped to about 25mV.
I don't know if it will help, I'll find out tomorrow during the stress test.
The electrolytic capacitors go away gradually, their capacity decreases and ESR increases, yet it worked without problems and the C parameters were worse than what I put there.I'll see what it does now.
By the way, the switching frequency is 23kHz Datasheet GM34063

 

[carpenter]

So unfortunately, the feed keeps freezing randomly.
I have already reduced the ESR of the capacitors at the input to below 80mOhm.

 

[hiphei]

Hi!

Repairing devices is different profession altogether compared to electronics circuit design and one has to set their brain into 'puzzle mode' and think in a multitude of levels at the same time (that's what the designer does too, but the levels are different)..the circuit, the material tear and wear, warranty for your labor and parts, etc.etc.. I've been working in repair business professionally for over 25 years and about half of the time as a company owner so i've had some time to hone the process of a repair..

If we talk about that specific product, i'm not familiar with it at all..however, i know the following process will get You a long term results..

Firstly..this product is not new and under a warranty (assumption by me).. Being an older product that's been in use over the warranty period typically brings different faults than the products have while they are new; older products have typically problems caused by time..eg. thermal expansion related faults, electrolytic capacitor faults, some materials can get leaky (eg. the red glue that was used in the 90's and early 00's to hold smd components while being flown with solder..), and semiconductors going unstable (possibly due to internal cracking of the substrate, junction, or something else), etc.

Since Your product supposedly goes into this category one must conclude that it has worked when it was new.. then one have to make a conclusion that once it's not working anymore there must have gotten some kinda fault..that conclusion sounds stupid, but one must think it from another perspective..meaning *it has been designed properly* and there is nothing wrong with the values and the types of components..so trying to fix the apparatus by replacing component values (be it capacitance, esr, inductance, whatever) or types won't get you anywhere..(sometimes you can get away for a short while by changing values to compensate the problem..but that just goes around..after a while it comes back since the fault wasn't repaired and compensation is not right anymore..)

As Klaus already wrote, esr is not the only thing that can go wrong inside the capacitor and by measuring it you still can't know whether the cap works or not..of course you can rule out the most obvious caps..

If the thing to repair this is a hobby then the product downtime or labor cost is no issue..and one can spend time to measure stuff and try stuff, etc..change parts..test it..do that again.. BUT..if the product downtime or labortime is an issue one must get to the core of the poodle sooner more likely than sooner..

I'd say quite confidently that all power supply or power line electrolytic cap wasn't younger than the next to it, so they most likely will fail next.. If You need to use the product, why not replace all of the power supply and power supply line capacitors? If you fiddle with the meter/oscilloscope/etc. for an hour, you can buy a pretty beefy bag of electrolytics with the same cost..so 5-10 caps is no issue..also then you have one solid *reference point* to make a right diagnose; It's not a capacitor fault anymore.. (you have to be methodical)
Alright..now we know it's not a capacitor issue..what is it you have left.. pcb, other passive components and semiconductors..
Let's think about pcb..what can go wrong there..? Either you got a crack or a leak/short.. Let's think a standard soldered component; You have the pcb material itself (fiberglass or pertinax, etc), then you have the foil/foil pad that's copper, then you have the solder itself and finally a component leg..all different material and all have different thermal expansion rate..so eventually it will crack.. and the solder is most typically the one's that's cracking..(may of course be any other of the 4 materials, but my experience tells me it's 9 out of 10 the solder that cracks..)
I remember old Philips K9 crt tv chassis was typically having intermittent faults and there was no other way to solve it than to draw a grid on the board with about 2 by 2 inch boxes and take the iron and solder each box from upper left corner to lower right corner and voila..no intermittent faults anymore..
So..resoldering all bad looking joints should be the first and foremost operation to perform with *every repair* but as a second thing to try out is to resolder everything possible (if you can not solder the smd chips, maybe you should take it to someone who has proper smd soldering station..?).

If you still didn't make it work, it can be eg. semiconductor issue..you might get it diagnosed with hair dryer and cold spray..but if it is a microcontroller issue, chances are poor to find a controller with proper firmware..

..so..back to my first post..have you reflown the bad looking joints and replaced *all* of the supply caps? I mean do we have a reference point here..? (i don't think you can find the designer reference values and ripples available so measuring that stuff very thoroughly is a waste of time here..)

Best Regards,
hiphei

--- Updated Jul 3, 2024 ---

I'd like to add something..

If we had schematics or even perhaps service manual, we might be able to solve like uC power on reset circuit problems or clock oscillator circuit problems, i2c bus block issues, eeprom issues and so on which all may present themselves as intermittent hang up.. but unfortunately most often there are none such documents available.. :/

BR,
hiphei