Chieftec HPC-360-202 5vsb jumped to 6V!

[goodpsusearch]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

Ok. I put back the original cap and the oscillation stopped.

I added an NTC thermistor 5A rated.

I did a partial recap on that power supply. After all, I have 3 Jou Jye 400Watt psu waiting for recap, those are very well built and heavier on 12V (29A) supplies, therefor I prefer to use my stock of Japanese caps on them.

And finally, I replaced the 2200uF 16V capacitor with 3300uF 16V.

Everything seems to be ok now. Thanks to everyone for your input.

[momaka]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

This power supply looks like the "Task" PSU I have. The PCB layout is completely different form mine, but I think it's the same manufacturer because:
- mine has the same mixture of Teapo/Jenpo caps
- SG6105D PWM controller
- half-birdge
- clean soldering and clean PCB traces
- same small 10 uF 50V cap before 12V PI inductor and larger cap after it
- also same oscillation behavior when I recapped it.

Btw, very good soldering:

I like it too.
Clean traces, leaded solder, and no SMD components - what more could you ask for?

I have no idea how two resistors and two caps can boost efficiency. I'd guess it's some kind of psuedo-PFC... but I haven't thought much about it.

I *think* they boost the voltage a bit across the primary caps, so when the PSU is in stand-by, it get's better efficiency. But when it's operating, like you I see no way how those caps can boost the voltage or improve efficiency. If anything, they are diminishing the efficiency (just look at how burned that PCB is!).

I replaced the 50V 10uF cap with 1000uF 16V. The power supply oscillates like crazy now!

Interesting.
I replaced mine (the 50V 10 uF cap) with a 16V 1200 uF Panasonic FM and I also got an oscillating output when I tried powering a 12V-based PC. Up to this point, though, I thought this was because I removed most of the minimum load resistors and because I also changed some of the other Jenpo caps on the output with Panasonic FM - but looks like that's not the case. However, when I put this PSU in a 5V-based PC, it doesn't oscillate anymore.

If you feel experimental, try adding that cap back in there and try the PSU with both 5V and 12V -based computers (if you have any that you don't mind sacrificing... I did my testing on an old Socket A PC with a 750 MHz Duron for the 5V test and a dead ECS P4IBMS with a 1.7 GHz P4 Willamate CPU for the 12V test). Later on, I also tried it in my dual 1 GHz Pentium 3 ASUS CUR-DLS (5V based) computer and it was still stable.

BTW, what cap brand and series did you use for the 16V 1000 uF? I *think* this PSU just doesn't like low ESR caps.

*** EDIT ***

Okay I did a bit of searching. UL number E144195 is/was for SIRTEC. My Task PSU has the same UL number too, so I was right.

You know sir, high tech power supplies like this often need re-capping

Ah, now I see what you meant by that comment

Man, if there was ever such profession as a power supply detective, I'd be one of the best.

[budm]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

Some Passive PFC designs:
https://cdn.badcaps-static.com/pdfs/...dfef6be8d2.pdf
http://electronicdesign.com/article/...ontrol-to-cfls
https://cdn.badcaps-static.com/pdfs/...361d1d92c6.pdf

[Behemot]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

OK, so it cuts a few harmonics. That's all? Oh man, come on

[goodpsusearch]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

Interesting.
I replaced mine (the 50V 10 uF cap) with a 16V 1200 uF Panasonic FM and I also got an oscillating output when I tried powering a 12V-based PC. Up to this point, though, I thought this was because I removed most of the minimum load resistors and because I also changed some of the other Jenpo caps on the output with Panasonic FM - but looks like that's not the case. However, when I put this PSU in a 5V-based PC, it doesn't oscillate anymore.


BTW, what cap brand and series did you use for the 16V 1000 uF? I *think* this PSU just doesn't like low ESR caps.

I did all my tests with 12V-based PC, because testing with 5V-based computers seems more and more pointless these days

I also believe that if I tested it with a 5V based PC the oscillation would stop.

The 1000uF 16V capacitor was a Jun Fu WG 0.055 ohm.

I also tried 100uF 25V Rubycon YXF 0.40 esr and I got little oscillation while running MEMTEST. 12V fluctuated between 11.96-12.05.

With the original Jenpo 50V 10uF cap I get stable main voltages when idle or light PC usage, but when in full load such as MEMTEST, the 3 main voltages fluctuate like this:
3.3V: 3.44-3.45
5V: 5.16-5.17
12V: 11.97-12.04 (approximately)

Panasonic FM are definitely too low esr for that psu. I hate when they use general purpose caps though. I think that Panasonic FC would be ideal for that psu. Too bad that they wouldn't fit Panasonic FK *might* be ok too.


I was tempted to play with the resistor values, but in the end I just reseated/bent them to stay as far as possible from the caps. Btw the 3.3V resistor is the lowest minimum load resistor I have seen in a power supply. It is only 7.5ohms! It is so low that the esr meter could not measure the 2x3300uF caps total capacitance. I had to remove the resistor to test them. I would replace the resistor, but 3.3V never gets bellow 3.43V, I was worried that if I put a higher ohm resistor, it could increase even more!

If you feel experimental, try adding that cap back in there and try the PSU with both 5V and 12V -based computers (if you have any that you don't mind sacrificing... I did my testing on an old Socket A PC with a 750 MHz Duron for the 5V test and a dead ECS P4IBMS with a 1.7 GHz P4 Willamate CPU for the 12V test).

Why bother looking for old/dead PCs? I just tried it on the 1st PC that was lying around: Jetway 848PDAP (recapped), Pentium 4 Celeron 2.2GHZ overclocked to 2,552GHZ DDR400 512MB

[goodpsusearch]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

Some Passive PFC designs:
https://cdn.badcaps-static.com/pdfs/...dfef6be8d2.pdf
http://electronicdesign.com/article/...ontrol-to-cfls
https://cdn.badcaps-static.com/pdfs/...361d1d92c6.pdf

Thank you very much, this is very very enlightening (pun intended )

So, the circuit works like this:

During each half-cycle of the rectified ac input, current through diode DVF2 and resistor RVF charges capacitors CVF1 and CVF2 in series to half of the ac peak voltage each. (RVF serves to reduce current waveform peaks as the capacitors charge.)

When the rectified ac input voltage drops below VPeak/2, the capacitors act essentially in parallel and supply load current to the dc bus until the rectified ac input again exceeds VPeak/2 on the next half-cycle.

This passive valley-fill PFC circuit provides good power factor (>0.9) and low total harmonic distortion, or THD (<30%), but has a major drawback: the dc bus ripple is 50%. If the ballast control circuit's operating frequency were fixed, the resonant output stage would produce a lamp current that followed the dc bus voltage.

The 50% ripple, then, would cause a very high lamp-current crest factor—ILAMP(peak)/ILAMP(average)—resulting in a reduced lamp life. An additional control circuit, consisting of RFMIN2-3, RBUS1-2, and QFMIN, works together with the ballast control IC to decrease this lamp-current crest factor by modulating the IC's operating frequency.

In this circuit the filter capacitors are charged in
series via the diode and resistor on each half cycle of the
rectified AC input. Each capacitor is charged to ½ of the
AC peak voltage, minus three diode drops - two in the
bridge rectifier and one in the diode between the two
capacitors. The purpose of the resistor is to reduce the
peaks in the current waveform as the capacitors charge.

Since each capacitor is charged to half the peak
AC voltage, they supply output current only after the bus
voltage follows the sinusoidal waveform down to Vpeak /2.
At this time the capacitors are essentially in parallel and
supply load current until the rectified AC input again
exceeds VP/2 on the next half cycle. The discharge duty
cycle for the capacitors is approximately 37% followed by
an idle period during which time the load is being
supplied directly from the rectified AC input. At the peak
of the input AC voltage there is an additional current to re-
charge the capacitors up to Vpeak. The magnitude and
duration of this current is a function of the depth of
discharge and the value of the resistor in the charging
circuit.


Although the circuit presents a reasonably good
Power Factor (>0.95) and the harmonics can be tamed by
the L-C input filter, the major shortcoming of this circuit is
the 50% bus ripple voltage

But that circuit is different from the one found in computer power supplies, this one has 3 diodes instead of 2.

[budm]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

More learning:


http://www.powerint.com/en/forum/low...lter-capacitor
"One final word of advice, make sure you size your post filter inductor and capacitor correctly to avoid causing problems with your control loop. In general, most PI products are designed with a control loop crossover frequency of approximately 1kHz. When choosing your post filter components, choose them so that they filter frequencies approximately one decade higher than your crossover frequency. This will help reduce problems with control loop instability."


http://www.ve7zsa.net/technical/advs...ply_b1_r00.htm
It talks about thr resonance of the Pi filter, so if you change the cap to bigger value, it may be what you are seeing.
"One problem inherent with the choke-input filter is that resonance may occur where the first inductor and first capacitor are in series across the output of the rectifier circuit. If they are in resonance at the ripple frequency , an acceptor circuit is formed which passes a very high current. This current causes a large voltage at the ripple frequency to appear across the capacitor. Thus the ripple is increased and the effect is opposite to that which is desired in filtering, the elimination of the ripple. This large increase in voltage or current can cause damage to components in the circuit, especially to the rectifier if they exceed the peak current and PIV ratings for the diodes. Since the resonant frequency depends on the inductor and capacitor, they have to be chosen so that the value LC is at least twice the value for resonance at the ripple frequency to prevent resonance effects."

[Pentium4]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

What would cause oscillation just by changing the cap even though the replacement has better ratings?

[Behemot]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

Those beter ratings? Oscilator must be based on some set parameters, if you change one, whole system changes and may even collaps. Here if you change capacitance, you have to change inductance as well.

[budm]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

"I replaced mine (the 50V 10 uF cap) with a 16V 1200 uF Panasonic FM "
It was 10uF, new one is 1200uF! So the resonance and filter frequency went way down.
http://www.powerint.com/en/forum/low...lter-capacitor
"One final word of advice, make sure you size your post filter inductor and capacitor correctly to avoid causing problems with your control loop. In general, most PI products are designed with a control loop crossover frequency of approximately 1kHz. When choosing your post filter components, choose them so that they filter frequencies approximately one decade higher than your crossover frequency. This will help reduce problems with control loop instability."

[momaka]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

I did all my tests with 12V-based PC, because testing with 5V-based computers seems more and more pointless these days

True. But some slightly older PSUs just aren't meant for 12V systems.

Btw the 3.3V resistor is the lowest minimum load resistor I have seen in a power supply. It is only 7.5ohms!

I have seen 6.8 Ohm twice now in Deer/Allied/L&C PSUs. Gotta make sure those crappy caps don't get cold feet

Actually, a hot resistor has 2 effects on the caps:
1) It keeps the caps hot so their ESR decreases a little - that way manufacturers can get away with using slightly inferior caps.
2) If the resistor is hot enough, it will make the caps fail faster (especially if they are crappy - it's actually guaranteed). That way, PSU lasts just a little bit (but hopefully past warranty) and then the user has to buy a new PSU again. Keeps the world economy going!

[momaka]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

What would cause oscillation just by changing the cap even though the replacement has better ratings?

It's not about the ratings of the cap, it's about the whole system as Behemot mentioned.

In a PC SMPS, the output voltages are usually filtered by an inductor in series with 1 or more capacitors. This forms an LCR (inductor-capacitor-resistor) circuit (where the resistance R comes primarily from the ESR of the capacitor). And the thing about LCR circuits is that they filter output ripple "differently" at different load levels. Now, there's nothing wrong with that, but the PWM controller may not "like" some of the ripple at certain output loads. This can cause the PWM controller to get "confused" which may in turn lead to more unstable voltages. They key to keeping the PWM controller "happy" is called compensation - which essentially consists of a network of capacitors and resistors placed on the feedback loop before it's being fed to the PWM controller. The values of these resistors and capacitors are carefully picked so that the PWM controller only sees the "overall" regulation of the outputs and doesn't get "distracted" or "confused" from the output ripple.
So in short, if the compensation is not designed properly, you can get lots of oscillations or out-of-spec voltages. It is also important to note that the compensation in each PSU is designed specifically for the output LCR filter circuit in that PSU. Some designs can be more tolerant of specification variations in the LCR filter circuit while others may not be - as is the case with this particular power supply.

[budm]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

"Btw the 3.3V resistor is the lowest minimum load resistor I have seen in a power supply. It is only 7.5ohms!" This only 0.44A, only about 1.5Watt of power.

[goodpsusearch]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

Have you ever seen a psu with an output voltage filter like this?

It's the 5vsb LCR filter of a power supply I am working on now. Do I have to remove coil that is in the output?

[momaka]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

"Btw the 3.3V resistor is the lowest minimum load resistor I have seen in a power supply. It is only 7.5ohms!" This only 0.44A, only about 1.5Watt of power.

Yes, but try touching it after a minute or so. 1.5W for the case size of that resistor will burn you and may even get hot enough to burn wires and PCB over time.
If you're making a SMPS and it doesn't work well without very large minimum load resistors, you're doing it wrong...
okay, maybe not wrong, but definitely not optimally.

Just FYI, I have several HiPro PSUs that have no minimum load resistors. And they run very stable with both 5V-based and 12V-based computers. IIRC, JhonnyGuru also tested one a while back, and it had very very clean outputs with minimum amounts of noise, even when driven past the labeled wattage.

Do I have to remove coil that is in the output?

Depends how big it is. If it's small and has very few turns, you can probably leave it. And if was designed like that, then do leave it alone unless you have problems.

[budm]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

They probably figure that 1.5W of power dissipation will be kept cooler with internal fan running. More likely marginal design just to last 2~3yeras before people will upgrade the computer.

[goodpsusearch]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

I opened a new thread about the power supply with the strange 5vsb ripple filtering circuit:

https://www.badcaps.net/forum/showthread.php?p=283259

[picturevalve]

Re: Chieftec HPC-360-202 5vsb jumped to 6V!

I have one of these which has died, I've replaced all the caps reading high esr including c34, c36 and c37 in the 5vsb section. All o/k now except the 5vsb is low at 4.4v. Any pointers welcome as to likely problem? Mant thanks.