View Single Post
  #9   Report Post  
Posted to sci.electronics.repair
[email protected] stratus46@yahoo.com is offline
external usenet poster
 
Posts: 475
Default DTV decoder repair, capacitor gets too hot

On May 30, 4:45*am, Jeroni Paul wrote:
I'm repairing a DTV decoder Saivod DT2012 (chip Cheertek and tuner

LG)
that had a swollen capacitor in the power supply. 6 months ago I
already replaced this capacitor with a low ESR high quality 105 deg
replacement but it just lasted 6 months and is swollen again.

The capacitor is EC205 1000uF 16V and is the filter for 5V output

in a
switching power supply. Typical design, small transformer,

rectifier
diode, first capacitor, inductor and second capacitor with TNY265

as
primary controller chip. The second capacitor remains cold and

checks
fine on the ESR meter. The problem is in the first capacitor, I
replaced it again with a new 105 deg type and it works but I notice
the capacitor gets very hot in just 10 minutes. All the capacitors

in
the supply check fine on the ESR meter. Tried different brands of
capacitors and the ones having the smallest ESR work better but

also
get very hot.

It seems to me this receiver is too exigent on the quality and ESR

of
this capacitor. Do you think this is normal and I should be looking
for a really good capacitor or should I look for another problem
elsewhere? Could I do something to reduce the stress it is put on?

The waveform on this capacitor looks like a sawtooth, rises really
fast and falls slowly until half cycle, then remains constant for

the
remaining half cycle. Amplitude around 0,4Vpp to 1Vpp depending on

the
capacitor used.


Where I used to work they had Samsung DTV tuners that had a similar
issue. In that case we mounted the capacitor on its side to get it
farther away from the diode next to it that was the source of the
heat. You might look into the Nichicon HN series caps or possibly an
organic polymer cap. AFAIK the HN caps are the lowest (?) ESR of the
'lytics but the polymers are supposed to be even lower. Would a larger
value cap help here? What is the actual load current and what is the
frequency of the ripple?