Need som info on power supply components

[mockingbird]

How off are we talking about? You said in the other thread that you were "not fully satisfied". Was it at least within spec? I imagine that the final 5V and 12V output also changed a bit when you swap the stock caps. I thought PFC was supposed to take care of all of this?

Also.....if you like to tweak.......add a 5vsb fan circuit such that when power supply is off but plugged into wall outlet, the 5vsb fan comes on and keeps everything nice and cool. Then when you turn the computer on, the 5vsb fan goes off and the main fan comes on.

This is a mod I had long since contemplated for my XBox... RBJTech posted an incomplete schematic to his (now defunct) site which no one ever got working. That was waaaay over my head back then and I doubt I could figure out how to do that now. I'm still trying to figure out how to implement your DM311 standby circuit to some PSUs.

Did you ever trace which caps were for which rails (Pi filters on the secndary)? Would be nice to know what's 3.3, 5, 12 so that I can have more flexibility when choosing replacements and fiddling with pi filter pairs. I've done this with my FSPs.

Thanks for all your help.

[mariushm]

You could easily do a 5vsb fan with a pic microcontroller or an attiny or any 0.5$ microcontroller.

A basic voltage doubler circuit would take care of starting fan at 10v ... to protect the input pins of the IC use maybe an optocoupler to detect when the 12v is gone (power supply turns off), use a npn transistor or darlington to connect the power to the fan...

Some other thing I thought about would be just using the 12v to keep a relay open blocking 5vsb power from going to the fan. When 12v goes off, relay closes, connecting the fan to 5vsb provided a npn transistor is conducting... which can be made using the charge in a capacitor that filled while the 12v was working. As you only need a few mA to make the transistor work, this should work like a timer... when the capacitors depletes its energy the fan stops.
Problem with this is default closed relays are expensive.

[momaka]

Originally Posted by everell

Also.....if you like to tweak.......add a 5vsb fan circuit such that when power supply is off but plugged into wall outlet, the 5vsb fan comes on and keeps everything nice and cool. Then when you turn the computer on, the 5vsb fan goes off and the main fan comes on.

Other than for fun, I don't recommend doing that. There are a number of reasons why:

1) It wears the fan more than needed. If you have a ball bearing fan, it wouldn't matter that much. However if you have a sleeve bearing fan you'll put considerably more wear on it.

2) It makes the PSU and its fan(s) clog faster with dust. (So even if you do have a ball bearing fan, what good would it do to you if can't push any air due to dust).

3) Draws extra power for no reason - seriously, most electronics like that don't need to be that cool. Take for example laptop chargers - they run just as hot (if not hotter) that these 5VSB circuits and even they survive for many many years. Unless crappy caps were used, I actually haven't seen a well-made one die of old age.

4) Adds noise in the room when the PC is OFF - this probably isn't an issue for most people. For me it is, though, because I have all of my PCs in my room and I don't like noise when I sleep. (I turn ALL of my PCs OFF overnight).

[c_hegge]

I'm still not totally convinced about the heat theory as to why C36 fails. On all the Bestec PSUs I've recapped, there was never any heat discolouration around C36, or anywhere else on the PCB, and bestec always use low quality caps, (usually OST or CapXon), or under-spec'd Jamicon general purpose caps, so of course they're gonna fail. Unless I see them killing pannies and rubys, I'm gonna stick with the junk cap theory.

I probably should add that I've only ever seen two where C36 failed. I run into heaps of bestecs and very few have bad caps

[Wester547]

Originally Posted by c_hegge

I'm still not totally convinced about the heat theory as to why C36 fails. On all the Bestec PSUs I've recapped, there was never any heat discolouration around C36, or anywhere else on the PCB, and bestec always use low quality caps, (usually OST or CapXon), or under-spec'd Jamicon general purpose caps, so of course they're gonna fail. Unless I see them killing pannies and rubys, I'm gonna stick with the junk cap theory.

Not all Bestecs have brown/tan glue, though - some have white glue - such as the ATX0300D5WC everell posted in the next generation Bestec thread. So not all Bestecs will have discoloration on the PCB (if I'm not mistaken). And as for why capacitor C36 fails, I noticed that that particular capacitor (a Jamicon?) is right under the roof of the heat sink, enough so that a top-down view of the power supply doesn't even show that capacitor (looking down from the top of the heatsink). We all know heat rises so perhaps the roof of that heatsink (as hot as it gets) is impeding the airflow of that particular capacitor, especially considering it's very close to the heatsink? I can't think of any other reason, other than what you pointed, that Bestec does not use good capacitors. Of course, I could be wrong. That's just my theory. But not all PSUs with underpar capacitors use heatsinks like Bestecs do (where the roof of the heatsink covers the capacitors near the heatsink). Bestec PSUs also don't have vents on the left and right sides - maybe that doesn't help either, though I'm not sure if that has anything to do with it.

You've seen very few Bestec PSUs fail from bad caps, even with the C36 capacitor in that very spot, with that very heatsink? If that's the case, then perhaps the placement isn't the reason.

EDIT: It must be the hot diode then as mockingbird and everell brought up. And it runs hot enough to cook a bad capacitor.

[c_hegge]

Originally Posted by Wester547

Not all Bestecs have brown/tan glue, though - some have white glue - such as the ATX0300D5WC everell posted in the next generation Bestec thread. So not all Bestecs will have discoloration on the PCB (if I'm not mistaken).

Heat discolouration has nothing to do with the glue used. I was talking about a darkening of the PCB itself (like http://www.badcaps.net/forum/showpos...&postcount=101)

Originally Posted by Wester547

And as for why capacitor C36 fails, I noticed that that particular capacitor (a Jamicon?) is right under the roof of the heat sink, enough so that a top-down view of the power supply doesn't even show that capacitor (looking down from the top of the heatsink). We all know heat rises so perhaps the roof of that heatsink (as hot as it gets) is impeding the airflow of that particular capacitor, especially considering it's very close to the heatsink?

Remember, most cases mount the PSU upside down, so the cap would be -above- the heat sink. I guess it is possible that the heat from the heat sink rises onto the cap, but the heat sink is cold when the PSU is off, and when it is on, it gets air from the fan, and other caps mounted under the heat sink don't seem to fail as often. Also, the 250-12Z has a very different heat sink shape, and the top doesn't cover C36 (http://www.badcaps.net/forum/attachm...9&d=1237938963), yet, everell and a few others have seen C36 fail lots of times on that model.

Originally Posted by Wester547

You've seen very few Bestec PSUs fail from bad caps, even with the C36 capacitor in that very spot, with that very heatsink? If that's the case, then perhaps the placement isn't the reason.

The ones I usually run into are 300-12Zs and 250-12Zs. My only other theory is that maybe the PCs which HP sell with them in the Australian market don't load the 5vsb as much as others, but I have my doubts as to that one, because the other PSU which they use interchangably (the Hipro HP-D3057F3H) always gets a bad 5vsb input cap.

[mockingbird]

Everell has mentioned before that the large diode next to C36 gets very hot and causes that.

[c_hegge]

^
I know what supposedly cooks C36, but I don't find that the diode gets hot enough to cook a cap.

[momaka]

Originally Posted by c_hegge

I'm still not totally convinced about the heat theory as to why C36 fails. On all the Bestec PSUs I've recapped, there was never any heat discolouration around C36, or anywhere else on the PCB, and bestec always use low quality caps, (usually OST or CapXon), or under-spec'd Jamicon general purpose caps, so of course they're gonna fail. Unless I see them killing pannies and rubys, I'm gonna stick with the junk cap theory.

I think I'll stick with that theory as well. The CapXon KM, Jamicon (SK?), and Teapo SEK caps they use are all general purpose. That being said, flyback topologies such as the one used for the 5VSB circuit put quite a bit of ripple on the cap so I'm not surprised if those general purpose caps overheat internally and fail because of that.

The 5VSB diode may run hot, but how hot are we talking about really? 75C? 85C? 95C? Under full 2A load, I would say all of that reasonable. Under a small load it should run cool, though.

[c_hegge]

Another mod would be to replace the diode with a TO-220 one, and mount it to the heat sink, using wires to connect it to the PCB. If you don't happen to have an appropriately sized drill bit, or have shaky hands like me, then that might me the better option.

[goodpsusearch]

Or just turn off completely the psu when the PC is off.

[mockingbird]

Originally Posted by c_hegge

^
I know what supposedly cooks C36, but I don't find that the diode gets hot enough to cook a cap.

There's a difference between heat that is convected through air and that which is convected through contact. The diode is practically touching the capacitor... I don't like that. Having said that, my cap wasn't bulged... Mind you, out of all the 21 capacitors (Minus the two primaries) on this 300Z revision , ALL of the capacitors are Jamicon "TK" Series (Some of which have bulged) besides this cap we are talking about which is a single OST "RLS" series. This is a highly rated cap at 0.022 ESR and 2150 ripple even if it is a substandard brand. It's not bulged, but I didn't test it otherwise. Would be interesting to note which series of caps in this position have bulged.

[c_hegge]

^
You may be on to something. On all of my bestecs, the diode doesn't touch C36. It's close, but it's usually bent outwards enough not to touch it. In any case, the bulged C36 on the two I've had to fix was CapXon IIRC.

[everell]

Originally Posted by mariushm

You could easily do a 5vsb fan with a pic microcontroller or an attiny or any 0.5$ microcontroller.

A basic voltage doubler circuit would take care of starting fan at 10v ... to protect the input pins of the IC use maybe an optocoupler to detect when the 12v is gone (power supply turns off), use a npn transistor or darlington to connect the power to the fan...

Some other thing I thought about would be just using the 12v to keep a relay open blocking 5vsb power from going to the fan. When 12v goes off, relay closes, connecting the fan to 5vsb provided a npn transistor is conducting... which can be made using the charge in a capacitor that filled while the 12v was working. As you only need a few mA to make the transistor work, this should work like a timer... when the capacitors depletes its energy the fan stops.
Problem with this is default closed relays are expensive.

Why go to that much trouble adding a fan. The Bestec ATX-250-12Z and ATX-300 12Z both have a place on the pc board to add one transistor, one resistor, and two diodes. That plus your fan and you have a nice functional 5vsb fan. See schematic and pictures of the pc board below.