"Unapproved" input filtering

[momaka]

Re: "Unapproved" input filtering

Okay here are some pictures of the new input filtering and an "overall" look of the PSU. Looks decent. Replaced the 20A ultra fast on the 12V with a 30A ultra fast, and replaced all the caps.

Nice work, as usual.
30A on the 12V should be plenty for just about anything. Maybe even a light duty welder?

I love power supplies

We share the same passion, I think . They are very interesting beasts, aren't they?

KME obviously wanted to be cheap by getting non safety approved input filtering but they must have had a reason to put a safety approved X cap before the fuse?

Probably what we were just discussing - some X caps are designed not to go short circuit failure mode, so perhaps that's why they felt it is safe to install one before the fuse.

How can they get away with this? Sounds very dangerous...

If power supplies were to be thoroughly inspected before being imported, such PSUs surely shouldn't make it to the market here. However, that is not the case. Also, there are a lot of "IFs" involved for a PSU with non-safety Y caps to become dangerous. First, your outlet/house grounding must be faulty. And secondly, the non-safety Y cap between L and G must fail short circuit in order for the computer case to become energized.

http://www.kemet.com/Lists/Filestore...aRFIandSMD.pdf
http://www.capacitorguide.com/filter-capacitor/
http://www.tecategroup.com/capacitor...capacitors.php

Very useful links - at leat the last two. The PDF in the first had its text in very light color and I couldn't read it at all. Perhaps Adobe Reader works with it? I have SumatraPDF, so maybe that's why?

...and capacitorguide implies that X caps can fail in both ways. So this supports the argument to check the datasheet to really know how a particular X cap will act in failure mode.

Yes, capacitor guide does word it in a very interesting way:
- "...as well as the fact that the failure mode of metallized capacitors tends toward open circuit, while the failure mode of ceramic capacitors tends towards short circuit, which is potentially more dangerous to users."

I guess the keyword here is "tends". So yeah, you are right, indeed - always check the datasheet. Or if in doubt and can't find a datasheet, just place any X and Y caps after the fuse, and pray that the cheapo glass fuse in the cheapo PSU doesn't fail to fail when the EMI caps fail
(lol, I think my high school English teacher would get a kick out of that sentence ).

Tecate (there's a beer brand in MX by the same name )

Is that a shockingly good beer too?

Beware of the possibility of instabilities if you shift the dominant pole around. Depending on the margins in that supply's compensation, you might not be able to get away with putting a cap before that PI-choke.

Yeah, I've seen that before in a Sirtec-built PSU of mine. It had an empty 10 mm spot before the PI coil on the 12V rail (actually, there was a 100 uF cap in that spots), and a 1200 uF cap after the PI coil. I removed the 100 uF cap without thinking and added another 1200 uF (a Panasonic FM, at that). Not sure if the PSU was meant to power only 5V-heavy loads or what, but it definitely wasn't happy with a normal 12V load and almost no 5V load. Oscillations and whinning was the result with pretty unstable voltages (though the PSU didn't manage to explode). I removed that 1200 uF Panny FM and subbed a Teapo SM or SC in its place. That removed some of the while and oscillations, but not completely. I didn't bother testing it further, but my guess is that Sirtec probably cut a few corners by installing that small 100 uF cap instead of a proper bigger cap, and then they "fixed" the compensation up a bit to account for that. But that's only my guess

[Wester547]

Re: "Unapproved" input filtering

Yeah, I've seen that before in a Sirtec-built PSU of mine. It had an empty 10 mm spot before the PI coil on the 12V rail (actually, there was a 100 uF cap in that spots), and a 1200 uF cap after the PI coil. I removed the 100 uF cap without thinking and added another 1200 uF (a Panasonic FM, at that). Not sure if the PSU was meant to power only 5V-heavy loads or what, but it definitely wasn't happy with a normal 12V load and almost no 5V load. Oscillations and whinning was the result with pretty unstable voltages (though the PSU didn't manage to explode). I removed that 1200 uF Panny FM and subbed a Teapo SM or SC in its place. That removed some of the while and oscillations, but not completely. I didn't bother testing it further, but my guess is that Sirtec probably cut a few corners by installing that small 100 uF cap instead of a proper bigger cap, and then they "fixed" the compensation up a bit to account for that. But that's only my guess

Some older Sirtecs were even worse in this respect, using only a 10uF 50V capacitor before the ferrite choke with a 1000uF 16V capacitor after for +12V filtering. Isn't it true though that the capacitor before the coil (the input capacitor) does the most filtering and that the one after does the least (the output capacitor), unless the one on the input has failed? If that's true, then I can't imagine how a single 10uF or 100uF capacitor with some "help" could possibly be enough to filter the ripple from even a single hard drive, optical drive, or even a single +12V DC fan. Unless Sirtec designed the PI filter so the cap after does the most filtering.

[Pentium4]

Re: "Unapproved" input filtering

Looks like you had fun- I like this!

kaboom is back! and yes I did have fun, as usual. I like to sit in the corner at work and solder. My co-workers call it my "therapy" for the day

I did the same in a supply I upgraded without any trouble. It, too, was a half-bridge, 5V-based design, so less emphasis on the +12 regulation may work in your favor.

Are you talking about the first or second PSU I posted here?

Beware of the possibility of instabilities if you shift the dominant pole around. Depending on the margins in that supply's compensation, you might not be able to get away with putting a cap before that PI-choke.

Thanks for the heads up. What would be the "consequences" of said instabilities?

Hope you can find something here that helps:
https://www.badcaps.net/forum/showthread.php?t=14405

Wow! Talk about a gutless wonder. The title of the thread is very appropriate. That thing got quite an upgrade, thanks for sharing! Did you do ripple tests before and after? Do you by chance have a solder shot too?

You are welcome! Yeah I really suggest you to try this mod because even if this units have been designed to use only one cap for 12V rail IMHO one cap is not enough specially if the power supply is pushed hard (ie- a somewhat power-hungry system), I don't have a load tester nor a scope but this mod resolved the issues a friend was having in his custom-built entry-level gaming computer (FX-8120, HD 7770, N68-VS3 FX, 8GB DDR3, Hard Drive and DVD burner) that system was presenting sudden shutdowns and random BSODs, the system was used only for six months before the issues appeared, he was reluctant to change the PSU (Pixxo) so I offered him a recap and he agreed. Long story short the recap only fixed the BSODs but shutdowns persisted to a lesser extent until that second cap in 12V rail was added and it's been working perfect for a year now.

Wow, that's crazy. Even with the increased ESR of lifting it off the PCB it still obviously did wonders for ripple suppression. Did yours power that system with the stock 20A 12V rectifier?! I wonder if it has working OCP, shutting it down when that rectifier is getting overloaded.

Nice work, as usual.
30A on the 12V should be plenty for just about anything. Maybe even a light duty welder?

Thank you and probably!

We share the same passion, I think They are very interesting beasts, aren't they?

We sure do. And yes they are, I don't know what it is about them but from the first time I popped off the top of an ATX PSU, I immediately found them infinitely fascinating, wanting to learn everything about them.

If power supplies were to be thoroughly inspected before being imported, such PSUs surely shouldn't make it to the market here. However, that is not the case. Also, there are a lot of "IFs" involved for a PSU with non-safety Y caps to become dangerous. First, your outlet/house grounding must be faulty. And secondly, the non-safety Y cap between L and G must fail short circuit in order for the computer case to become energized.

I guess that's a rare scenario, but a hazard nonetheless. I lived for a year and a half at an old 1929 house in the University District until the end of February, and that place had cloth covered wiring, and NO grounded outlets except for the electric stove! Could this scenario have been possible at my old house?

[kaboom]

Re: "Unapproved" input filtering

kaboom is back!

Aww, thanks.

and yes I did have fun, as usual. I like to sit in the corner at work and solder. My co-workers call it my "therapy" for the day

You are lucky to be able to work on those power supplies like that.

Quite the opposite around here. It'll always be some excuse. I'm not going to elaborate much, but let's just say that I've gotten so tired of the ones who think a "pre-emptive strike" on the badcaps is mere "justification to give me money." And a certain (used) ebay shitsta laptop (around two years ago) that was claimed to "be working fine" and I was "too enthusiastic" about upgrading it to XP pro SP3.

I'll post more sometime in VIP- I don't know how many "locals" know about BCN.

Are you talking about the first or second PSU I posted here?

Thanks for the heads up. What would be the "consequences" of said instabilities?

That's what I get for posting late w/o making sure I was clear... I was referring to the one right above/before my post.

If it starts complaining, there will be some combination of (lowish?) 5/12 crossloading that'll make it squeal and even be, in an extreme case, overly sensitive to step-loading. You'd have to scope the outputs, 3.3 probably won't show it, since the magamp effectively isolates it from the rest of the supply. Which, by the way, is why those Deers can have absurd ripple on 5/12, while 3.3 can still look good.

Edit to add: For adding a cap before a PI-choke where there wasn't any, don't go too low in ESR. Try a PW or FC. HE may be too low, and FM/FR are certainly too low. If it had one to begin with, all except FM are possible choices; whatever is closest to the originals- but even then, fake caps' datasheets could be "fudged."

Wow! Talk about a gutless wonder. The title of the thread is very appropriate. That thing got quite an upgrade, thanks for sharing! Did you do ripple tests before and after? Do you by chance have a solder shot too?

Sorry, no pix. I almost forgot to take some- it was back in the computer when I took some of them.

Even though it wasn't powering much, that thing was only ever topped by the Adam-puter and the redneck-puter, two of my favorites. Again, sometime in VIP...

I didn't have a scope then, but by DMM is capacity-coupled on the AC-V scale, so I made a quick-n-dirty check before and after on the AC millivolts scale. This was only a relative measurement, since a typical DMM won't be accurate.

I forget exactly what the loads were, which were the same for the before-n-after, but the millivolt readings (not measurements, uncal here) did drop. If there's nothing in the thread I linked, I couldn't tell you now...

Wow, that's crazy. Even with the increased ESR of lifting it off the PCB it still obviously did wonders for ripple suppression. Did yours power that system with the stock 20A 12V rectifier?! I wonder if it has working OCP, shutting it down when that rectifier is getting overloaded.

Thank you and probably!

I wonder how many people blamed it on windows or "broken motherboards" when the silly power supply was the problem...

We sure do. And yes they are, I don't know what it is about them but from the first time I popped off the top of an ATX PSU, I immediately found them infinitely fascinating, wanting to learn everything about them.

Use this to your advantage, since you seem to be around those who appreciate you and your knowledge and desire to learn.

I guess that's a rare scenario, but a hazard nonetheless. I lived for a year and a half at an old 1929 house in the University District until the end of February, and that place had cloth covered wiring, and NO grounded outlets except for the electric stove! Could this scenario have been possible at my old house?

Let's take the computer by itself first. Please note that nothing in a computer needs to be grounded, even the power supply. You can run a power supply board outside of its case with the two "current carrying conductors." You might be thinking "hot & neutral," but the white wire isn't a true neutral unless it's part of an MWBC- carrying the unbalanced current of two out-of-phase 120V ckts. That electric stove, or dryer ckt would be but two examples. A common 120/240V split-single-phase supply is another.

For those, the white, or bare, if before the service disconnect, is a true neutral. Also note that before a code change, electric dryers and stoves could use the bare (grey-area grandfathered) grounding AS the grounded or neutral. Dryer motors are almost 120V in these units, with the heater(s) being 240. So the older dryers would have the motor connected from either leg to the frame of the dryer, where it returned thru the (true) neutral, the same neutral that supposedly kept the frame of the dryer near 0 volts.

Now what if that neutral had a poor connection or even opened completely? Upon trying to start the motor, the dryer frame would go right up to 120V! This would not trip a breaker, since this was an opening of a normal current path, not a short. The dryer will appear "dead" (motor cannot start) yet anytime you'd hit the button, you'd get knocked on your ass (or worse)!

This ties in with something you asked, Momaka, about tying your grounds to neutrals. It is a very bad idea.

The above examples were required to help with grounding a power supply, or any other grounded item.

The reason we have the third wire is to provide a dedicated fault-clearing path should, say, the insulation in a transformer fail, or exploded capacitor guts bridge line voltage side to low voltage side.

That computer will work w/o out the ground. All that matters inside is that all commons be returned back to their source- namely the black wires on the LV side of the PS. The monitor, too.

But it's far more desireable to "anchor" the cases/chassis of said equipment, so when cockroach guts, capacitor pieces, bad/fake Y-caps, or failed insulation (anywhere) blows a fuse or trips a breaker. If it weren't for that, the incorrect Y-cap that shorts one day will instead bring the case up to 120V.

Now, I'll try to bring this all together. The "new" dryer plugs have four wires- L1, L2, W/neutral, GND/bond. It actually is more of a bonding/common that a "ground." We do not depend on soil resistance alone to "ground" anything at 120V- the resistance is far too high.

The reason for four wires in dryer and stove plugs is to make safe the double-contingency of an open neutral and and fault to the chassis at the same time. On the old 3 wire plugs, say there was a pinched wire dumping current into chassis/back up the neutral/"ground," which is making poor contact and burns open. Now, the "grounded," via the now-open neutral, dryer is a real hazard.

This works both ways. If you "ground" a computer, or anything else to the neutral, (strictly speaking, on 120V only ckts, it isn't- that's just vernacular) and the neutral opens somewhere upstream of the computer, like an open backstab, everything downstream of the open will quit working. And everything "grounded" to that (open) neutral will now be floating at 120V via the L-N load resistances! That includes your computer case, Momaka!

If that's not a "skystorm" on the "farms," tell me what is! We do not want these so-called bootleg grounds.

There are also other effects, due to Ohm's law. Wires have resistance, obviously. Let's say a given outlet measures 125V w/o load and 120V w/ 10A drawn. Loop resistance is (125-120) / 10 = 5 / 10 = .5 ohms.

Ignoring the small resistance in a properly functioning ckt bkr, we'll say that each wire contributes .25 ohms. A series divider in series with the actual load. The hot is down 2.5V from its original value, however, the "neutral" is now 2.5V above it's original zero! Wires have resistance.

That could be a space heater on the ckt where a computer was bootleg grounded. The case, and audio common, could feed signifigant current into a properly grounded audio mixer, for example. Severe hum will result- classic ground loop. The cause isn't balanced-vs-unbalaced audio cables, but rather the audio "common" being tied to a 2.5V, 60 Hz source- the neutral, with it's normal IR voltage drop, being used as a "common," "ground," or other such hack.

No bootleg grounds. Using a GFCI is better- they don't need any ground whatsoever as they detect imbalance between current leaving and current returning. If the difference between the two is greater than 4-6 mA (for people protection), they drop the load.

So in this case (punny..), the computer case will float at ~60V, current limited to a hundred microamps or so, since there is an L-case-N capacitive divider formed by the Y-caps. IOW, "Why we want good Y-caps."

The function of the Y-caps is obviously to augment the line filter- the "parasitic" effect of the divider is unavoidable. At most, it would give the most imperceptible tingle, and only if you had contact between the floating case and a grounded object- water pipe, radiator etc.

Hope it isn't too much at once and you all enjoy.

Somehow, I posted all that w/o hitting the char limit- amazing!

[SIDMX]

Re: "Unapproved" input filtering

...nice custom RS cap

Yeah I got my stash ...noticed you too.

Wow, that's crazy. Even with the increased ESR of lifting it off the PCB it still obviously did wonders for ripple suppression. Did yours power that system with the stock 20A 12V rectifier?! I wonder if it has working OCP, shutting it down when that rectifier is getting overloaded.

It's a somewhat long story, but in short I'm not sure if the shutdowns were trigered by P/S or motherboard because the CPU VRM of the board was cooked badly for 6 months. 20A rectifier? IIRC it was a crappy 16A fast.

Edit: It's the sme model as this one : https://www.badcaps.net/forum/showth...251#post189251 so must be a 16A fast.

Is that a shockingly good beer too?

If you mean like a Corona no it isn't IMO, but Corona is not a good beer honestly, it is so thin-flavored that it just goes well with almost everything, "Negra Modelo" or "Bohemia Oscura" those are really good beers, I prefer Tequila though .

[momaka]

Re: "Unapproved" input filtering

I guess that's a rare scenario, but a hazard nonetheless. I lived for a year and a half at an old 1929 house in the University District until the end of February, and that place had cloth covered wiring, and NO grounded outlets except for the electric stove! Could this scenario have been possible at my old house?

It sure is.
However, even if the L-G Y-cap shorted to the case and the case was ungrounded, you still likely wouldn't have "noticed" until you touched something grounded or stand barefoot on a concrete floor.

I like to sit in the corner at work and solder. My co-workers call it my "therapy" for the day

lol, that is indeed pretty cool. Most repair shops don't bother to have anyone fix their bad power supplies. They just keep a pile of crappy ones on hand and replace as needed.

This ties in with something you asked, Momaka, about tying your grounds to neutrals. It is a very bad idea.
...
This works both ways. If you "ground" a computer, or anything else to the neutral, (strictly speaking, on 120V only ckts, it isn't- that's just vernacular) and the neutral opens somewhere upstream of the computer, like an open backstab, everything downstream of the open will quit working. And everything "grounded" to that (open) neutral will now be floating at 120V via the L-N load resistances! That includes your computer case, Momaka!

Well, I'm glad you've been reading my "grounding" threat. Yeah, that's why I asked some questions there. Mainly, I grew up in a country where it has been almost standard (though I don't know if legal, but no one really cares over there) to tie ground to neutral. Makes me wonder how my friends and I (and even my parents' generation, for that matter) survived . Oh yeah, 230 V AC too .
I even remember my uncle showing me how to plug in some 3-phase motors in a 3-phase outlet in his shop, and these didn't have any plugs on them - just stick the motor's wires in the outlet .

If that's not a "skystorm" on the "farms," tell me what is! We do not want these so-called bootleg grounds.

Ignoring the small resistance in a properly functioning ckt bkr, we'll say that each wire contributes .25 ohms. A series divider in series with the actual load. The hot is down 2.5V from its original value, however, the "neutral" is now 2.5V above it's original zero! Wires have resistance.

That could be a space heater on the ckt where a computer was bootleg grounded. The case, and audio common, could feed signifigant current into a properly grounded audio mixer, for example. Severe hum will result- classic ground loop. The cause isn't balanced-vs-unbalaced audio cables, but rather the audio "common" being tied to a 2.5V, 60 Hz source- the neutral, with it's normal IR voltage drop, being used as a "common," "ground," or other such hack.

Actually, in most cases you won't really get a ground loop in audio equipment. Reason being is that most audio gear (at least the somewhat more oldschool stuff and older) are double-insulated and only use a 2-prong plug. So that 2.5V at the ground won't really make it anywhere.
...
But yeah, I'm not going to argue now that it's a bad idea to tie ground to neutral. It really is, now that so many examples were shown.

Somehow, I posted all that w/o hitting the char limit- amazing!

I never hit it, but I'm pretty sure my posts always come close.

...because the CPU VRM of the board was cooked badly for 6 months.

You shouldn't expect anything less from ASUS
You are probably wondering, "ASUS? What is this guy talking about, this is AsRock"... but look again at the caps on the motherboard, or in particular, their orientation relative to the shaded semi-circle on the board .

"Negra Modelo" or "Bohemia Oscura" those are really good beers

Yeah, my dad likes the Negra Modelo quite a bit. Whenever someone offers him Corona, Bud Light, or any of that stuff, he makes the funniest "hell NO" face ever .

[SIDMX]

Re: "Unapproved" input filtering

You shouldn't expect anything less from ASUS
You are probably wondering, "ASUS? What is this guy talking about, this is AsRock"... but look again at the caps on the motherboard, or in particular, their orientation relative to the shaded semi-circle on the board .

Yeah I don't like Asus boards (bad experiences in the past) but my friend just learned that in the past year , about the Asus/ASRock thing I'm aware that Asus established ASRock to compete in lowend market or someting like that I can't recall exactly but thanks nonetheless.

Yeah, my dad likes the Negra Modelo quite a bit. Whenever someone offers him Corona, Bud Light, or any of that stuff, he makes the funniest "hell NO" face ever .

That "face" probably means something like this:

[c_hegge]

Re: "Unapproved" input filtering

You shouldn't expect anything less from ASUS

Yeah I don't like Asus boards (bad experiences in the past) but my friend just learned that in the past year , about the Asus/ASRock thing I'm aware that Asus established ASRock to compete in lowend market or someting like that I can't recall exactly but thanks nonetheless.

Funnily enough, hot/discoloured VRMS are about the only problem I haven't had on ASUS motherboards. I've had it on some Intel boards, and on a couple of AMD Gigabyte boards, but at least they've been reliable otherwise. With ASUS, it's just everything else that goes wrong for me. 9 times out of 10, there's no visual evidence at all of anything wrong. They just stop working.

[Pentium4]

Re: "Unapproved" input filtering

*insane amount of good info*

Wow, thanks Paul for writing all the scenarios out for me. I sincerely appreciate it! You made it easy to understand for noobs such as myself I had to read it a few times but I feel wiser now thanks to you. I'm still glad to be out of that house though Also, it was still on a 10A breaker

You are lucky to be able to work on those power supplies like that.

Quite the opposite around here. It'll always be some excuse. I'm not going to elaborate much, but let's just say that I've gotten so tired of the ones who think a "pre-emptive strike" on the badcaps is mere "justification to give me money." And a certain (used) ebay shitsta laptop (around two years ago) that was claimed to "be working fine" and I was "too enthusiastic" about upgrading it to XP pro SP3.

I am pretty lucky I'm the only one in the shop who cares to fix anything, as the other techs have almost zero interest in component level repair. Those customers can't be helped, and they will never learn

Yeah I got my stash ...noticed you too.

It's a somewhat long story, but in short I'm not sure if the shutdowns were trigered by P/S or motherboard because the CPU VRM of the board was cooked badly for 6 months. 20A rectifier? IIRC it was a crappy 16A fast.

Edit: It's the sme model as this one : https://www.badcaps.net/forum/showth...251#post189251 so must be a 16A fast.

Yeah the RS do come in handy! I've used around 60 of them. If it's 16A, it's definitely getting overloaded!

If you mean like a Corona no it isn't IMO, but Corona is not a good beer honestly, it is so thin-flavored that it just goes well with almost everything, "Negra Modelo" or "Bohemia Oscura" those are really good beers, I prefer Tequila though .

I love Bohemia, it's an excellent cheap beer. It's my go to when I'm not drinking a tasty IPA, Stout, or red ale

lol, that is indeed pretty cool. Most repair shops don't bother to have anyone fix their bad power supplies. They just keep a pile of crappy ones on hand and replace as needed.

Very true. They only let me do the quick ones though, as in 30 minutes or less. A fair number of the "bigger" projects are ones I work on after hours or during my lunch break

[momaka]

Re: "Unapproved" input filtering

Funnily enough, hot/discoloured VRMS are about the only problem I haven't had on ASUS motherboards. I've had it on some Intel boards, and on a couple of AMD Gigabyte boards, but at least they've been reliable otherwise. With ASUS, it's just everything else that goes wrong for me. 9 times out of 10, there's no visual evidence at all of anything wrong. They just stop working.

Right you are.
When it comes to hot CPU VRMs, Intel motherboards easily win that crown. In fact, I just fixed one with a shorted MOSFET in the CPU VRM last week. Took it out to set up a computer yesterday, played around in the BIOS for about 10 minutes, and then the VRM shorted again - this time a different MOSFET. The worst part is, I noticed heat damage around the upper MOSFETs in the VRM before I even fixed it. So after I fixed it, I soldered a long piece of 14 AWG solid copper wire to server as a heatsink (ghetto mod thread forshadowing? ). But even with that, the VRM still shorted.

I also find that other Intel motherboards regularly have problems with hot-running MOSFETs. Given that and their crappy BIOS options, I really dislike Intel boards - to the point where I actually prefer ASUS.

Speaking of which, slightly older ASUS boards usually have hot FETs around the chipset and/or RAM, but not the CPU VRM. Actually, to their credit, ASUS motherboards have pretty well-designed CPU VRMs IMO, with large copper planes to heatsink the MOSFETs and lots of cap spots, unlike some cheap brands (I'm looking at you Biostar ). Yet, why do they fail so much is beyond me.

Anyways, I think that's too far off-topic now. Sorry for that.

I'm still glad to be out of that house though Also, it was still on a 10A breaker

Hey, as long as the wires were 14 AWG, that low-rated breaker is actually not a bad thing - at least it will overload faster and (hopefully) shut off.

My current place has the plugs in the living room, bedroom, bathroom, and all of the lights combined on one 15A breaker. I'm not too happy about that, but I don't have a lot of high-power devices and I'm partially nocturnal , so it should be okay.

They only let me do the quick ones though, as in 30 minutes or less. A fair number of the "bigger" projects are ones I work on after hours or during my lunch break

30 minutes - dang! I can't take anything appart in 30 minutes, much less fix it .
So I am guessing that's what your co-workers are reffering to as the "therapy" time?

[kaboom]

Re: "Unapproved" input filtering

Right you are.
When it comes to hot CPU VRMs, Intel motherboards easily win that crown. In fact, I just fixed one with a shorted MOSFET in the CPU VRM last week. Took it out to set up a computer yesterday, played around in the BIOS for about 10 minutes, and then the VRM shorted again - this time a different MOSFET. The worst part is, I noticed heat damage around the upper MOSFETs in the VRM before I even fixed it. So after I fixed it, I soldered a long piece of 14 AWG solid copper wire to server as a heatsink (ghetto mod thread forshadowing? ). But even with that, the VRM still shorted.

A prescott fed by a two-phase converter? There were lots of cheep, and frankly, not-so-cheap intel boards with just two phases. The comparable Dell and Gateway spec'd boards of the time often were three-phase. But Intel liked to max them out. This fact, combined with the "novelty" of "quiet computers," often did those boards in- an example below.

You better believe those systems got quiet- right after the power supply shut down feeding into a short.

I also find that other Intel motherboards regularly have problems with hot-running MOSFETs. Given that and their crappy BIOS options, I really dislike Intel boards - to the point where I actually prefer ASUS.

Speaking of which, slightly older ASUS boards usually have hot FETs around the chipset and/or RAM, but not the CPU VRM. Actually, to their credit, ASUS motherboards have pretty well-designed CPU VRMs IMO, with large copper planes to heatsink the MOSFETs and lots of cap spots, unlike some cheap brands (I'm looking at you Biostar ). Yet, why do they fail so much is beyond me.

Because they were linear regulators, like 78xxs, except made with discrete parts. The buck converters for the CPU, obviously, are miniature SMPSes- their MOSFETs are either fully on or off.

Now on those intel boards, usually there are solder spots for copper "wing" heatsinks. The prototypes had them, silly, stingy intel left them out on the "production" models, just as they left out some of the caps and/or subbed in cheaper caps.

That was why those intel boards w/ 85 degree caps near the RAM, AGP, and chipset would become unstable over the years. Since those same MOSFETs, as pass transistors in linear regulators, were co-located near the slots and chipset, all those "unimportant" small caps went bad. Lots of 100u/16v.

There were some junky Compaq Presarios, back in the 90's, that used linear regs for the CPU! The regulators ran as hot as the CPUs did.

[kaboom]

Re: "Unapproved" input filtering

Wow, thanks Paul for writing all the scenarios out for me. I sincerely appreciate it! You made it easy to understand for noobs such as myself

You're welcome! Just remember, at the end of the day, after you learn anything new, you're no longer a "noob."

I had to read it a few times but I feel wiser now thanks to you. I'm still glad to be out of that house though Also, it was still on a 10A breaker

There was a time when you could bond to a cold water pipe, since the incoming neutral of the drop was supposed to be tied to the CWP with the GEC, even if the receps were ungrounded. It was common for washing machines to include their own ground/bond wire and clamp for just this purpose.

You'd have: washing machine chassis/cabinet -> (clamp to) CWP -> GEC -> service neutral.

Another, not as obvious, parallel path exists: washing machine -> CWP -> backfeed into neighbors' GEC (remember, they share the street water main) -> up neighbors' neutrals, back to neutrals on poles -> finally back to the originating transformer.

In such a shared-pipe scenario, this parallel path causes a few amps to flow on everyone's water mains, even if all the services are proper! Electricity takes all paths.

Also, since the water main would feed other houses, and they also had the "water bonds," you could "ride out" an open service neutral with this bond in place. Your 120V neutral point (at the service) wouldn't float up nearly as much- they were less likely to be "240-ed." Of course, with PVC or "poly-pipe" water services, this wouldn't happen.

To debunk "path of least resistance," here's a neat experiment. You'll need:

A plugstrip
A lamp & bulb- CFL, LED, or incandescent, anything from 4 to 100W.
Cord connected, high power appliance- vacuum cleaner, toaster, heater, etc.

Connect plugstrip to power and connect lamp. Lamp comes on, right? Okay, now plug the vacuum in and turn it on. Does electricity really take the path of least resistance? The answer should be obvious, as both the lamp (high resistance) and vacuum (low resistance) are now running at the same time, directly in parallel with each other.

Hope that was fun.



House 2 and 3 "save" house 1 here:
http://forums.mikeholt.com/showthrea...t=95473&page=8

Some others:
http://www.mikeholt.com/mojonewsarch...s~20030409.htm
http://forums.mikeholt.com/showthread.php?t=160118

I am pretty lucky I'm the only one in the shop who cares to fix anything, as the other techs have almost zero interest in component level repair. Those customers can't be helped, and they will never learn

Use it to your advantage!

The ones who'll never learn? Let them break everything, then "amuse" yourself with their excuses of "Why you want so much to fix what I broke." They get uniblue and other trash on their computer, then think you, or your company's price is "unfair?" Walk away...

Oh sure, it'll be about how "I can't be w/o a computer," or "I haven't backed anything up," as if you should feel sorry. Well, you sure as hell didn't put uniBULL on there, did ya?

It's a two way street- they'll "complain" about not having their computer. You can also "complain!" You could be playing w/ power supplies, which is rather rewarding, but you don't have the time, you've got to do another "backup/fdisk/reinstall/driver search."

Give yourself an easy way out- tell 'em to "get vistafied." Those round corners and "special interface" are perfect snares for "catching sheep."

Very true. They only let me do the quick ones though, as in 30 minutes or less. A fair number of the "bigger" projects are ones I work on after hours or during my lunch break

Didn't you know power supplies can grow legs? Maybe a few will follow you home, where they'll be serviced properly...

[SIDMX]

Re: "Unapproved" input filtering

Yeah the RS do come in handy! I've used around 60 of them. If it's 16A, it's definitely getting overloaded!

I've used more than 200 already, . That rectifier surely is overloaded or very close (that FX is the 95W TDP revision) or maybe K-Mex upgraded that rectifier in later models IDK, anyways I warned him about the consequences but he didn't seem to matter so "Not My Circus, Not My Monkeys!"

I love Bohemia, it's an excellent cheap beer. It's my go to when I'm not drinking a tasty IPA, Stout, or red ale

There are like 3 diferent Bohemia's but "Obscura" is the best IMO.

Right you are.
When it comes to hot CPU VRMs, Intel motherboards easily win that crown...
...I also find that other Intel motherboards regularly have problems with hot-running MOSFETs. Given that and their crappy BIOS options, I really dislike Intel boards...ASUS motherboards have pretty well-designed CPU VRMs IMO, with large copper planes to heatsink the MOSFETs and lots of cap spots, unlike some cheap brands (I'm looking at you Biostar ). Yet, why do they fail so much is beyond me.

Yeah intel is definitely the winner specially if there is no option in bios to turn off FAN managment.

That was why those intel boards w/ 85 degree caps near the RAM, AGP, and chipset would become unstable over the years. Since those same MOSFETs, as pass transistors in linear regulators, were co-located near the slots and chipset, all those "unimportant" small caps went bad. Lots of 100u/16v.

There were lots of them in i845-i865 chipset era IIRC.

[SIDMX]

Re: "Unapproved" input filtering

Funnily enough, hot/discoloured VRMS are about the only problem I haven't had on ASUS motherboards. I've had it on some Intel boards, and on a couple of AMD Gigabyte boards, but at least they've been reliable otherwise. With ASUS, it's just everything else that goes wrong for me. 9 times out of 10, there's no visual evidence at all of anything wrong. They just stop working.

It's the first one I've seen with that extreme overheat, but the excessive ripple from P/S was to blame I think. I have pics and a video during overheat, IIRC VRM temp was above 100°C/212°F, I may be posting this in a new thread.

[c_hegge]

Re: "Unapproved" input filtering

Right you are.
When it comes to hot CPU VRMs, Intel motherboards easily win that crown. In fact, I just fixed one with a shorted MOSFET in the CPU VRM last week. Took it out to set up a computer yesterday, played around in the BIOS for about 10 minutes, and then the VRM shorted again - this time a different MOSFET. The worst part is, I noticed heat damage around the upper MOSFETs in the VRM before I even fixed it. So after I fixed it, I soldered a long piece of 14 AWG solid copper wire to server as a heatsink (ghetto mod thread forshadowing? ). But even with that, the VRM still shorted.

I've had hot VRM problems in particular with some 945 intel boards (D945GTP and D945GNT spring to mind but there's probably some others too). I ended up modding the VRM a bit on my spare 'GTP (see https://www.badcaps.net/forum/showthread.php?t=38947). The heatsinks I stuck to the FETs still do get fairly toasty on the workbench, but hopefully with a fan or two around the case it will be a bit better.

The other thing I didn't particularly like about it is that it only had a single 1200uF Nichicon HD cap on the VRM high side. I fixed that too.

I also find that other Intel motherboards regularly have problems with hot-running MOSFETs. Given that and their crappy BIOS options, I really dislike Intel boards - to the point where I actually prefer ASUS.

I'll still take an Intel board over an ASUS any day of the week. BIOS options never really worry me, and I've had far more ASUS boards fail for unknown reasons than Intel boards with cooked VRMs.

[Behemot]

Re: "Unapproved" input filtering

Edit to add: For adding a cap before a PI-choke where there wasn't any, don't go too low in ESR. Try a PW or FC. HE may be too low, and FM/FR are certainly too low. If it had one to begin with, all except FM are possible choices; whatever is closest to the originals- but even then, fake caps' datasheets could be "fudged."

This is the highest advantage of Samxon RS - they definitelly do not have too low ESR so you can place them almost everywhere without much concerns. On the other hand using NCC KYA or the upcoming KZN, you have to give it a well thought if it will enhance things or more like bring problems actually…

About current protection - don't forget to mention than SMPS are low-to-middle frequency antenna (as they operate around 50-100 kHz) so they radiate something. The metal cover catches that and conducts to ground. This is why GFCI has problems with SPMS and why it is mostly used in places where you can really get ugly electrocuted but there is not many SMPS, like in bathroom, but not in office or classroom full of computers. It would need to have so high tolerances for current flowing through ground instead of neutral than it would not help anyway, because the currents will be higher than deadly currents for man.

Yeah I got my stash ...noticed you too.

Don't push people, I have more than enough for everybody

[momaka]

Re: "Unapproved" input filtering

That was why those intel boards w/ 85 degree caps near the RAM, AGP, and chipset would become unstable over the years. Since those same MOSFETs, as pass transistors in linear regulators, were co-located near the slots and chipset, all those "unimportant" small caps went bad. Lots of 100u/16v.

Yeah, don't even mention that - ticks me off even more about Intel.
If not that, then they use those blue general purpose Teapo/Yageo caps.

There were some junky Compaq Presarios, back in the 90's, that used linear regs for the CPU! The regulators ran as hot as the CPUs did.

You mean as hot as the CPU without a heatsink?

I've had hot VRM problems in particular with some 945 intel boards (D945GTP and D945GNT spring to mind but there's probably some others too). I ended up modding the VRM a bit on my spare 'GTP (see https://www.badcaps.net/forum/showthread.php?t=38947). The heatsinks I stuck to the FETs still do get fairly toasty on the workbench, but hopefully with a fan or two around the case it will be a bit better.

I fixed a DG33SX2 (a c.c. of the DG33TL) with a blown upper FET. The (small) copper area connected to the upper FETs had clear signs of heat discoloration. Replaced blown FET, checked for others - all was good... so I soldered a thick piece of copper wire bent like a heatsink to take some of the heat away. This worked very well... for about 10 minutes in BIOS, and then another upper FET blue (on which I also installed a heatsink). Now it's waiting on the fix pile again, until I get more motivation and time to fix it again.

The other thing I didn't particularly like about it is that it only had a single 1200uF Nichicon HD cap on the VRM high side. I fixed that too.

WOW! Talk about cost-cutting. I hate Intel mobos.

BIOS options never really worry me, and I've had far more ASUS boards fail for unknown reasons than Intel boards with cooked VRMs.

At least ASUS boards fail and then you are done with them there and then. When Intel mobos are working (and most are), they are finicky - about RAM, HDDs, and just about anything you can think of. Oh, and then Dell takes them and changes their stock (and already-locked down) BIOS to an even more limited one. Sorry, but that's just a form of masochism. I rather not use a computer at all.

[lti]

Re: "Unapproved" input filtering

There were some junky Compaq Presarios, back in the 90's, that used linear regs for the CPU! The regulators ran as hot as the CPUs did.

My junky Compaq from the '90s uses a switching regulator that gets hotter than the CPU.

How did everyone start talking about CPU VRMs?

[Wester547]

Re: "Unapproved" input filtering

Yeah, don't even mention that - ticks me off even more about Intel.
If not that, then they use those blue general purpose Teapo/Yageo caps.

I think those would be either the SD or SX series (though I don't think those are general purpose as they were tested at a ripple frequency of 100KHz and not 120Hz). Can't say that's a good thing but I always thought Intel had a strict mandate that their motherboards have Japanese capacitors only, at least that's what PCBONEZ always said? I've never seen an older Intel motherboard without Japanese capacitors at least except on eBay, but for all I know some moron could have recapped it with Teapo. All my Intel motherboards, including the ones built for Foxconn, have all Japanese capacitors.

I can't understand their proclivity for 85*C capacitors either, though, other than for planned obsolescence. I think they actually started using them in the Socket 370/i820 motherboard days, but I guess that was less of a big deal as motherboards of that age don't generate as much heat, having much lower TDP CPUs, etc. That being said, the Intel motherboards I use the most, 2000-2002 vintage, do not have heat issues at all, even around the areas with linear regulators without heatsinks.

But I might be biased in that respect as my cases with those motherboards (either Dell clamshells or full size ATX cases with good ventilation) had good cooling, with the full size cases having 80mm front chassis fans, 70mm back fans, 70mm CPU fans, and 120mm or 140mm PSU fans all running at full speed save by the very latter. And the Dell clamshells had 92mm CPU fans and 80mm PSU fans of course, but the case was obviously designed well enough to make good use of them. Also, the Dell Pentium III boards I have without clam shell cases don't generate much heat at all (except the ones with RAMBUS of course, which aren't cooled well at all in those machines by default), and those are the boards that don't give a northbridge a heatsink.

I can understand how 85*C capacitors could be an issue in SFF machines or just poorly ventilated and cooled machines with their vents clogged, though, or just in later Intel boards that had heat issues by way of hellacious Prescotts and toasty VRMs.

WOW! Talk about cost-cutting. I hate Intel mobos. At least ASUS boards fail and then you are done with them there and then. When Intel mobos are working (and most are), they are finicky - about RAM, HDDs, and just about anything you can think of. Oh, and then Dell takes them and changes their stock (and already-locked down) BIOS to an even more limited one. Sorry, but that's just a form of masochism. I rather not use a computer at all.

Dell is indeed horrible about making the BIOS completely useless. But I wouldn't go as far as to say that I hate Intel motherboards or to peg them with ASUS boards. At least Intel motherboards last with good cooling and capacitors (excepting maybe the turbo fan syndrome issue of the SFF GX270s, etc). ASUS boards, on the other hand, never last, ever, ever, ever.

But yes, this is off-topic. As to the original topic, unapproved input filtering is of course a common sight in cheapo PSUs, either bearing fake Y capacitors or no Y capacitors at all...

[c_hegge]

Re: "Unapproved" input filtering

At least ASUS boards fail and then you are done with them there and then. When Intel mobos are working (and most are), they are finicky - about RAM, HDDs, and just about anything you can think of...

I've found ASUS to be the picky ones. Every Intel board I've used has worked fine with whatever other parts I happened to have on hand. That ASUS board which I polymodded recently is very picky, not only about what GPU you use, but also about what angle they are seated in the PCI-E slot. The card has to be seated with a very slight lean away from the CPU side and towards the other slots, or it won't POST. It's probably just a badly designed slot, but it's still annoying.

As for the whole unapproved filtering issue, I agree it's poor practice (and possibly even illegal in some countries) to use unapproved components. However, I can't think of an instance where I've ever seen an X or Y cap fail before - even on cheap junk PSUs.