VOGONS


Reply 20 of 31, by mockingbird

User metadata
Rank Oldbie
Rank
Oldbie
keropi wrote:

^ no page to order , I need to pick specific caps in dimensions (else they won't fit) and I can't just give a telephone order and receive Lelon caps 🤣

Send me a PM. I can tell you where to get authentic quality caps shipped inexpensively.

mslrlv.png
(Decommissioned:)
7ivtic.png

Reply 22 of 31, by luckybob

User metadata
Rank l33t
Rank
l33t
keropi wrote:

...damned minimum orders, a 5-7eur job ends up at 30+ :\
looking for alternatives like ebay atm

And this is why I was going to suggest NO. Not when you can get a BRAND NEW high efficiency supply for just a few dollars more.

I dont know if newegg ships to greece, but look at these units: http://www.newegg.com/Product/Product.aspx?It … N82E16817139026

The only time I'd even CARE about the 5v rail is if I was using an old AMD with 5v connector. Anything else isn't going to use that much 5v.

Cheap old power supplies are just a disaster waiting to happen. Even "fixing" them is hit and miss for even the experienced. ITS JUST NOT WORTH THE RISK TO YOUR HARDWARE should something go wrong.

It is a mistake to think you can solve any major problems just with potatoes.

Reply 23 of 31, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
keropi wrote:

300W Heroichi PSU: worth refurbishing?

That depends entirely on the specific model you have.

Many Heroichi (aka HEC, aka Compucase) PSUs were built with incomplete (only partial) input filters. (The input filter is a 2-way filter that, in addition to other things blocks, line noise both ways.)
Incomplete input filters can result in that background 'buzz' noise in other nearby AC powered devices (like radios, TV's, etc.. most anything with speakers) or it can result in the same noise getting into your PC (and PC speakers) from crappy PSUs in other devices.

With Heroichi the crappy input filter setup is found in the Orion series, some other low-end units and it is common in OEM models.
There are OEM Heroichi models with full filters though, so you just have to look. (Know because I have some.)

HEC PSUs in general are good reliable PSUs. (Aside from caps issues common to most older PSUs of any brand.)
A HEC with a compete input filter setup is absolutely worth rebuilding (recapping) if you need the -5v and don't care about the efficiency rating.
.

Last edited by PCBONEZ on 2015-11-14, 15:44. Edited 1 time in total.

Reply 24 of 31, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
Logistics wrote:

Power supplies, unless specifically designed to use polys, are not the appropriate place for a poly upgrade.

True but you have the 'why's' kind of wrong.

Logistics wrote:

Polymer caps are very fast, and would require all the circuitry behind them to be equally quick.

This concern only applies to signaling applications and is NA to filtering ripple on a DC circuit.
"Fast" is just the wrong idea/word and it's out of place in a discussion about ripple filters.

Logistics wrote:

In the VRM circuit of a motherboard, behind a CPU you would normally half the value in capacitance because the goal here is only to smooth the incoming power.

I'm the originator of this particular rule - years ago whilst on BadCaps.net after a discussion with kc8adu.
It was based on the info in a tech-doc. An application guide for a VRM controller chip.
That was very early in the poly-modding days and while it's usually perfectly fine to use it, more recent knowledge requires adding some notes.
#1: It applies only to motherboard VRMs. (NOT PSUs or other areas on a motherboard.)
#2: There is a minimum total capacitance needed even in a motherboard VRM (both high and low sides). That minimum capacitance is largely dependent on the CPU or (more specifically) the 'size' of load transients when the CPU loading changes. Boards designed with wet-lytics tend (in the VRM) to have a total uF far in excess of what is actually needed. That was the solution to lowering ESR at the time. If the original caps on a board had a total capacitance on the low side then halving it could result in a system that crashes on large CPU load transients. - It is best to find a motherboard with a similar (powered) CPU that has polys as original and use that total capacitance as your minimum.

This is "my" minimum total uF (to keep out of trouble) after looking at many MANY boards and the specs for the CPUs they run.
It (the values) applies to sockets 603, 604, 771, 423, 478, 775 (and probably everything up) as the CPU loadings are in the same ranges.
(Does not apply to odd-balls like Pentium-M or Atom boards which can use less.)
(I can't help you AMD guys. I don't see enough AMD boards.)
- Low (volt) side of VRM:
Low end consumer boards usually have (a minimum) of 5040-5600uF (9x or 10x 560uF)
Higher end consumer boards usually have at least 6500uF - I use this as minimum for all consumer class boards I repair.
Server boards tend to have 8000-10000uF per CPU. (Clearly these are designed with heavy load transients in mind.)
- High side of VRM (the 16v caps) - Most boards have either 3 or 4 caps here.
4pc of 330uF(min) or 3pc of 470uF(min) - I prefer 4pc of 470uF or 3pc of 680-820uF

The uF in PSU filters is ultra important for a few reasons.
#1: It affects the timings at BOOT time.
When a PC BOOTs the rails on the board have x-amount of time to get to ~80% (or so) of rated voltage or the power-good signal won't issue.
Smaller uF caps won't hurt you on that but caps with too high a uF may.
#2: It affects how well the PSUs respond to load changes and how much the voltage dips when the load suddenly goes up. In this case more uF means more reserve power for transients.
#3: The cap is not the whole OP filter. There are also inductors. The values of the inductors and caps are chosen such that the filter is optimized for the expected ripple frequency.
Changing the uF too much without also changing the inductor may 'de-tune' the filter such that less ripple gets removed.
- If you use too low of uF in your PSU (as with some poly-mod idea) then #2 or #3 may bite you.

Too low (excessively low) ESR in you PSU OP caps can bite you too. In some PSUs it can actually make ripple worse.
But that is incredibly hard to explain in detail and I'm not in the mood. Real basic follows.
Essentially (as far as ripple goes) you are lowering the resistance to ground. Like a near-short. The ripple current may go way high between the switchers and the OP filters.
High current = heat.

Poly-modding PSU's is possible but it's not straightforward and you really need an O'Scope to check your results, least you enjoy frying things.
.

Last edited by PCBONEZ on 2015-11-14, 16:17. Edited 5 times in total.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 25 of 31, by keropi

User metadata
Rank l33t++
Rank
l33t++

IIRC I already recaped this psu and currently use it in my p1 mmx build
Thanks for the extra knowledge PCBONEZ

🎵 🎧 PCMIDI MPU , OrpheusII , Action Rewind , Megacard and 🎶GoldLib soundcard website

Reply 27 of 31, by TELVM

User metadata
Rank Oldbie
Rank
Oldbie

Greetings PCBONEZ, glad to see you here around respect-048.gif . I suggest everybody pay close attention to what this man has to say, he really knows the stuff.

PCBONEZ wrote:

... #2: There is a minimum total capacitance needed even in a motherboard VRM (both high and low sides). That minimum capacitance is largely dependent on the CPU or (more specifically) the 'size' of load transients when the CPU loading changes ...

There is this tiny and cheap ASUS AM1I-A mini-ATX mobo that comes with one single 270uF poly on VRM-high:

h0x83ous.png

Now many people including myself have been experiencing the occasional random reboots on these systems.

I suspected that even for frugal CPUs like AM1 ones (~25W TDP), just 270uF might be a too little amount of VRM-high capacitance. So I soldered a 3300uF Panny FM on the 'P4' wire, right close to the EPS12V connector, to help with transient handling:

6OZYGyhb.png

No more reboots after that.

Let the air flow!

Reply 28 of 31, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
TELVM wrote:
Greetings PCBONEZ, glad to see you here around http://smileys.on-my-web.com/repository/Respect/respect-048.gif . I suggest every […]
Show full quote

Greetings PCBONEZ, glad to see you here around respect-048.gif . I suggest everybody pay close attention to what this man has to say, he really knows the stuff.

PCBONEZ wrote:

... #2: There is a minimum total capacitance needed even in a motherboard VRM (both high and low sides). That minimum capacitance is largely dependent on the CPU or (more specifically) the 'size' of load transients when the CPU loading changes ...

There is this tiny and cheap ASUS AM1I-A mini-ATX mobo that comes with one single 270uF poly on VRM-high:

~~see image in previous post~~

Now many people including myself have been experiencing the occasional random reboots on these systems.

I suspected that even for frugal CPUs like AM1 ones (~25W TDP), just 270uF might be a too little amount of VRM-high capacitance. So I soldered a 3300uF Panny FM on the 'P4' wire, right close to the EPS12V connector, to help with transient handling:

~~see image in previous post~~

No more reboots after that.

That was a very good way to fix the problem in that situation.

There is another 270uF-16v over by the RAM slots that may be in parallel (through the board) with the one you pictured, but still, 540uF is just not enough on the +12v rail.
I'm not surprised that board suffers for random reboot issues.

If you are savvy with replacing caps on motherboards you might consider replacing those 270uF with 820 or 1000 uF caps.
Those aren't cheap in 16v but you only need two.
....

I have not been a fan of Asus since the Pentium-III era.
IMHO Asus went south after that both in engineering and business practices.

Asus denied (through customer support) that capacitor issues existed on boards that had 100's of complaints on-line about capacitor problems.
That included boards with bad Nichicon HM and HN caps even after Nichicon made a public announcement that certain date-codes were defective.
For that matter they pretended there there was no such thing as a capacitor plague at all.
That's just not honest and I avoid companies that have a an apparent standing policy to lie to customers to get out of warranty work.

In the socket 478/775 era Asus did too many 'bad engineering practice' type cheap-down stunts for me.
Those were usually things that a component level tech would see tracing out circuits but most others wouldn't catch it.
They don't get caught because few people with the skills and/or equipment are willing to spend the time on a potentially unrepairable board.
They did things like skimping on EMI/RF filters and bypass caps near chips or ports. (Usually small SMD components that no one really looks at.)
Also they knowingly used caps with high failure rates all through that era.
OST (brand caps) were WELL KNOWN to be unreliable since 2003-ish and Asus continued using them in droves anyway.
-- Your board with only 540uF on +12v exists as evidence Asus hasn't changed their cheap-down practices.
-- As they buy caps in bulk using 2x 820uF there instead of 270uF -might- cost them 50 cents more a board. Probably less.

I have seen a vast number of complaints about Asus boards built since 2000 just quitting ~ for no apparent reason ~ .
From my observations Asus has far more 'mystery failures' than other brands. With other brands you can usually figure out why it failed. Not so much with Asus.
These mystery failures inspired me (while over at badcaps.net) to coin the expression "ASDS" aka "Asus Sudden Death Syndrome".
....

I did not mean to derail the thread. I seem to have a talent for that.
If the mods could move the stuff not pertaining to the PSU to it's own thread, that would be just ducky with me.
- Thanks.
.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 29 of 31, by TELVM

User metadata
Rank Oldbie
Rank
Oldbie

The MSI AM1I mobo has 2x 270uF on VRM-high and doesn't seem to suffer from random reboots:

vCd5rVsm.png

Might be ASUS went a bit too far on penny-pinching to keep AM1I-A costs down.

PCBONEZ wrote:

... you might consider replacing those 270uF with 820 or 1000 uF caps. Those aren't cheap in 16v but you only need two ...

That would be the elegant solution. As it happens I already had a bunch of 3300uF FMs at hand and modding the PSU cable was less disruptive.

BTW, here at Vogons there is this thread on troubleshooting and recapping ancient PSUs: A tale of two PSUs

Let the air flow!

Reply 30 of 31, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
TELVM wrote:

The MSI AM1I mobo has 2x 270uF on VRM-high and doesn't seem to suffer from random reboots:

I had trouble finding good photos of that board but I did see one that showed an additional 270uF-16v and 4x 100uF-16v.
Those are most likely all in parallel through the board on +12v rail so the total uF on +12v would be 1210uF.
That's more than double what the Asus board has and pretty close to the 3x470uF or 4x330uF I usually see on full sized boards.

How much uF is in the PSU's +12v OP filters may also affect the rebooting problem.
Did you run them with different PSU's?

I haven't done a whole lot with mini-ITX boards but I do have some Supermicro X7SPA variants in use and an EBM-945GME (supports laptop C2D) I've never bothered to test out.
I use the X7SPA in my NAS and my Media-Server.
The EBM-945GME is a cool little board but the power connector is not ATX. It's a 1x4-pin header that needs +12v and +5v. Never got around to finding an appropriate PSU for it.
Looks like this: http://www.backplane.com.au/index.php?Page=Ai … 25&product=3359
.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 31 of 31, by TELVM

User metadata
Rank Oldbie
Rank
Oldbie
PCBONEZ wrote:

... How much uF is in the PSU's +12v OP filters may also affect the rebooting problem ...

Yep from what I've read around the web it seems like some PSUs can cut it with these boards without reboots, while others can not. Perhaps depending on their +12V OutPut caps' combined uF and ESR.

PCBONEZ wrote:

Did you run them with different PSU's?

Just with this cheap and simple Be Quiet L8 300W (APFC, double-forward, group regulation, passive rectification). It has three Teapo SC 16V 3300uF 10mm on +12V OP, one before and two after the PI coil, each of the latter on one of the two +12V rails.

3lHEDlyk.png

93snAp1T.png

Let the air flow!