Psu ripple and hard disk problem

[NOCTURNE]

Hello to all
I have a 750 watt PSU for several years (about 5 years). my HDD making noise and  freezing just a millisecond I replaced the hard drive and realized the new one has the same problem again.
then I realized that the problem is with the power output ripple, and by replacing the input and output capacitors, the problem was solved for 9 months.

After that, it happened again and this time  bought some very high quality and expensive capacitors and changed all the capacitors. The problem was solved.

But now, after 2 months, the hard drive has the same problem again and I do not know why the capacitors break down so quickly again, even though they are much better than before, apparently, they can not clean the output ripple properly.

I wanted to know, do you know why it happened so quickly? Is there a problem with the electricity in the house? or because the output filter may have a fault that could damage the capacitor's ripple cleaning? I always used up to a maximum of 400 watts of power wattage, I am really frustrated.

[PKTKS]

A wild guess...

Since some couple years...  hardware has gone up to the 500W
easily with these "modern" GPUs now required.

The problem is with those absurd power hungry devices
is that such enormous power will be dissipated in components
as well..  capacitors were never meant to handle such HIGH FREQUENCY
at such HIGH POWER RATINGS..

Nevertheless they are being used daily and such stress will
certainly  make their life shorter.  (HEAT MOISTURE BRAND... POWER...)

My hope in the near future is that that FORM FACTOR  can be changed
enough to split such stupid power ratings among different PSUs
instead of overloading a single one with such load...

A single 1200W for a PC ?  this is a disaster waiting to happen.

Try  a different brand or  DOUBLE YOUR PSU to minimize HEAT...
and stress..

Paul

[tunk]

No expert on capacitors, but I'm left with the
impression that not all capacitors are created
equal. The new capacitors, were they low ESR?

[NOCTURNE]

A wild guess...

Since some couple years...  hardware has gone up to the 500W
easily with these "modern" GPUs now required.

The problem is with those absurd power hungry devices
is that such enormous power will be dissipated in components
as well..  capacitors were never meant to handle such HIGH FREQUENCY
at such HIGH POWER RATINGS..

Nevertheless they are being used daily and such stress will
certainly  make their life shorter.  (HEAT MOISTURE BRAND... POWER...)

My hope in the near future is that that FORM FACTOR  can be changed
enough to split such stupid power ratings among different PSUs
instead of overloading a single one with such load...

A single 1200W for a PC ?  this is a disaster waiting to happen.

Try  a different brand or  DOUBLE YOUR PSU to minimize HEAT...
and stress..

Paul

thanks for the replay 
it's a real 750 PSU, not peak 750 and I don't think its because it is not able to handle 400w since I checked the output voltages at the proper pressure and it was not dropping out of the standard voltage range
also, caps are new I getting more info in a replay to tunk

[wraper]

If capacitors you used for repair were general purpose type, even if super high quality, they will fail very fast.

[NOCTURNE]

No expert on capacitors, but I'm left with the
impression that not all capacitors are created
equal. The new capacitors, were they low ESR?

If capacitors you used for repair were general purpose type, even if super high quality, they will fail very fast.

thanks for the replays 
these are my output filter capacitors datasheets:
https://silmic.ir/wp-content/uploads/2020/05/ME-WX.pdf - 2200 and 25volt
https://silmic.ir/wp-content/uploads/2020/04/al-ky-e-2020.pdf 1000 35 volt
they all have high Rated ripple current number in their datasheets
----------------------------------------------
also, I wanna say the ripple iam talking about is not a ripple happening on output voltages transparently, since they all seem really stable as far as measured them and in standard range even in pressure.
but this ripple happens in very high frequency I think and only can see with an oscilloscope maybe?
---------------------------------------------------------
i wanna know if the caps failed after 2 mounts only or something causing the output filter to not working properly

[wraper]

Those capacitors are fine (unless dodgy stuff from ebay or aliexpress). Most likely there was some other problem in PSU.

[NOCTURNE]

Those capacitors are fine (unless dodgy stuff from ebay or aliexpress). Most likely there was some other problem in PSU.

doggy stuff  ? fake ones or the ones were on the shelf for a long time before selling? I don't think they are fake and also 2 mounts is a very very short time for such caps to fail

[wraper]

Those capacitors are fine (unless dodgy stuff from ebay or aliexpress). Most likely there was some other problem in PSU.

doggy stuff  ? fake ones or the ones were on the shelf for a long time before selling? I don't think they are fake and also 2 mounts is a very very short time for such caps to fail

Ebay and ali are full of counterfeits, much harder to buy genuine rather than fake crap. 2 months is about right for cheap general purpose capacitors to fail.

[fzabkar]

The +5V supply to the HDD is mostly down-regulated to Vcore and Vio levels for the digital logic. The +12V supply drives the spindle motor and voice coil. The +5V supply is also applied to the read/write amplifier ("preamp") via a filter inductor.

[NOCTURNE]

The +5V supply to the HDD is mostly down-regulated to Vcore and Vio levels for the digital logic. The +12V supply drives the spindle motor and voice coil. The +5V supply is also applied to the read/write amplifier ("preamp") via a filter inductor.

thanks for the info but i couldn't quite understand what you want to say 

[fzabkar]

The +5V supply to the HDD is mostly down-regulated to Vcore and Vio levels for the digital logic. The +12V supply drives the spindle motor and voice coil. The +5V supply is also applied to the read/write amplifier ("preamp") via a filter inductor.

thanks for the info but i couldn't quite understand what you want to say 

I'm just saying that the HDD has a lot of electronics which is designed to filter out noise from the supply. I would be more inclined to blame the HDD rather than the PSU.

[wraper]

I'm just saying that the HDD has a lot of electronics which is designed to filter out noise from the supply. I would be more inclined to blame the HDD rather than the PSU.

FWIW I've read plenty of stories about HDD emitting weird noises and prematurely failing due to bad PSU. As of filtering on 12V, there is barely any. Also ferrite bead won't filter anything except very high frequency noise.

[NOCTURNE]

I'm just saying that the HDD has a lot of electronics which is designed to filter out noise from the supply. I would be more inclined to blame the HDD rather than the PSU.

FWIW I've read plenty of stories about HDD emitting weird noises and prematurely failing due to bad PSU. As of filtering on 12V, there is barely any. Also ferrite bead won't filter anything except very high frequency noise.

The +5V supply to the HDD is mostly down-regulated to Vcore and Vio levels for the digital logic. The +12V supply drives the spindle motor and voice coil. The +5V supply is also applied to the read/write amplifier ("preamp") via a filter inductor.

thanks for the info but i couldn't quite understand what you want to say 

I'm just saying that the HDD has a lot of electronics which is designed to filter out noise from the supply. I would be more inclined to blame the HDD rather than the PSU.

I found this on the internet seems somehow related to my problem this article is also indicating bad caps of the output filter

Power Supply works, but the game is flakey on it: Check the filter capacitors on the output section of the power supply. Look for ones that have split tops or ones that have tilted over or lifted up because the rubber plug has popped out of the bottom. If they all look OK then either shotgun them or check the outputs with an oscilloscope and look for trash high-frequency AC ripple on them. Replace the caps as necessary to clean up those outputs, fix any cracked solder joint, reassemble, and test.
http://electronics-diy.com/repairing-switching-power-supply.php
------------------------------------------------------------------------------------------------------
also, I am going to build a portable cap changer with electrolytic caps for output. I tested this before and this worked but somebody in another forum said: this time add some polyester caps for high-frequency ripples on output regarding your current flow this even increases your cap's lifespan.
----------------------------------------------
btw I know buying a new psu is best option but currently i cant do this since this kind of components are extremely expensive here for our currency.

[PKTKS]

I have also read some folks stating that these new trendy PSUs
in which they value most the badges and that crappy plus gold shit...

are indeed failing short to supply proper power on some GPUs in very
specific cases..  folks which measure the in rush current of some games
are getting current SPIKES of short duration
in orders of double or triple the actual capacity of the PSUs.

Igors LAB already checked these spikes with proper equipo in which
the time window used (resolution) exceeds even these new NVIDIA PCAT
recent MEAS tools


So in plain facts those PFC based boosters can not provide these
fast current surges ... and these power ratings fall shortly with
HIGH RIPPLE or even undercurrent capacity.

Boosters diminish the dynamic current capability which led
these cases to the last hope solution of OVERSIZING PSUs to
at least double ratings... in case these fast in rush current
can drop them down...

More and more cases I have read last months..

Sad conclusion that these PFC based PSUs
with badged efficient ratings are crappy garbage
for serious power hungry high current spikes demands...

Paul

[NOCTURNE]

I have also read some folks stating that these new trendy PSUs
in which they value most the badges and that crappy plus gold shit...

are indeed failing short to supply proper power on some GPUs in very
specific cases..  folks which measure the in rush current of some games
are getting current SPIKES of short duration
in orders of double or triple the actual capacity of the PSUs.

Igors LAB already checked these spikes with proper equipo in which
the time window used (resolution) exceeds even these new NVIDIA PCAT
recent MEAS tools


So in plain facts those PFC based boosters can not provide these
fast current surges ... and these power ratings fall shortly with
HIGH RIPPLE or even undercurrent capacity.

Boosters diminish the dynamic current capability which led
these cases to the last hope solution of OVERSIZING PSUs to
at least double ratings... in case these fast in rush current
can drop them down...

More and more cases I have read last months..

Sad conclusion that these PFC based PSUs
with badged efficient ratings are crappy garbage
for serious power hungry high current spikes demands...

Paul

damn dude nice info  i didn't know there's Npcat equipment out there for real test, although for my case this is not the problem...
do you know which brand have better psu for these specific scenarios or i should go for double the wattage every time buy a psu ?
thanks

[wraper]

So in plain facts those PFC based boosters can not provide these
fast current surges ... and these power ratings fall shortly with
HIGH RIPPLE or even undercurrent capacity.

Oh my, you again with baseless PFC hate as if it's what's wrong with this world . And cheapest and crappiest PSUs made from shit and sticks, you are advocating for, supposedly should work fine? Would you mind to provide a quote from Igor saying there is something wrong with PFC?

[wraper]

https://www.igorslab.de/en/the-battle-of-graphics-card-against-power-supply-power-consumption-and-peak-loads-demystified/5/

Summary of measurement results
Modern switching power supplies should have no problems with the flowing currents produced by graphics cards. I couldn’t measure really sensationally high currents with any single card and in any of the different load situations. Spikes up to about 10 milliseconds are present, no question about it, because that is in the nature of things, because demanding games also cause very different load changes.

[PKTKS]

geez how can i hate a circuit....

plain simple facts i am tired to see...

BOOSTERS ARE NOT GOOD DYNAMIC CURRENT SOURCES...

it took some time to someone with the right tools and skills to spot that..

check the current probe setup used and how to see that spotted.. IgorsLab videos provides quality tests

[PKTKS]

... my next question is ..

how can someone without that awesome setup..  decides which PSU can or can not cope that current surges
..

i have seen a lot of these modern stuff failing.. or just shuting down..

when not deadly crapped by some intense surge during a game session

[PKTKS]

Having 4 R&S scopes with those current sensing probes
attached is probably not affordable for folks just wanting
to evaluate if some budget PSU will actually "works" with
some gaming card...

Actually it seems cheaper to oversize the PSU itself..

Just an amazing quality MEAS setup...

https://www.igorslab.de/en/power-recording-graphics-card-power-supply-interaction-measurement/3/


Wish one of those R&S with the probe ...
Paul

[PKTKS]

He actually used a specific HAMEG setup (as I understand..)
with proper carefully intervals to spot some tricky "design"
limitations of today "modern" stuff.

Reading the whole thing with care one thing actually
spots my eyes even much more than just the BAD PFC design..

Regarding GPUs (which are not like CPUs)  the typical 431 FB network
(which mostly also uses a Type3  compensator) is just as limited
to answer these fast in rush currents and compensate the regulation
fast enough..

very interesting thoughts on how regulation can put a bad delay
which will certainly impact other devices which are - as of today -
attached to the same POWER RAIL...  CPUS disks...

IMHO GPUs require a separate PSU.. wo PFCs and even the TL431 seems
totally obsolete to answer that fast in rush..

Very interesting readings..

https://www.igorslab.de/en/power-recording-graphics-card-power-supply-interaction-measurement/

Paul

PS.. please note carefully his comments on these..

"Over the last few months, I have been able to measure different manufacturer solutions in the supply of graphics cards. Initially, it was smaller cards such as different models of the GeForce GTX 750 Ti, whose power consumption did not really exceed the 60 watts on average, but depending on the model and the clock can generate very high power peaks. Then the 75-watt limit for fractions of a second falls quite clearly, as the graphic clearly shows us:"

[PKTKS]

... now 2 more very interesting pages ...

specific to  CAPACITOR issue
https://www.igorslab.de/en/power-recording-graphics-card-power-supply-interaction-measurement/8/

and the previous pages comparing SINGLE RAIL x MULTI RAIL designs..
the ripple spawned across the MOBO... and the HEAT expansion..
interesting pages.. how these sum up to shut down the system
or even dead short something...

And how bad these "modern" digital control designs can actually be
more a problem than a solution...

Again.. this reading is very needed to think problems like this
one on the original post OP.

Paul

[exe]

I realized that the problem is with the power output ripple

Did you use an oscilloscope to confirm this?

[NOCTURNE]

... now 2 more very interesting pages ...

specific to  CAPACITOR issue
https://www.igorslab.de/en/power-recording-graphics-card-power-supply-interaction-measurement/8/

and the previous pages comparing SINGLE RAIL x MULTI RAIL designs..
the ripple spawned across the MOBO... and the HEAT expansion..
interesting pages.. how these sum up to shut down the system
or even dead short something...

And how bad these "modern" digital control designs can actually be
more a problem than a solution...

Again.. this reading is very needed to think problems like this
one on the original post OP.

Paul

It an interesting article so still I don't know what cap characteristics in the datasheet going to be more important when it comes to my problem
I found some of the previous caps datasheet, not all of them but 2200 uf for output there is 1000 uf as well i don't know what was their name
http://www.elna.co.jp/en/capacitor/alumi/catalog/pdf/rj4_e.pdf

I realized that the problem is with the power output ripple

Did you use an oscilloscope to confirm this?

sadly I don't have one but I know at least changing output filter caps will solve the issue for a while.