VOGONS


First post, by MMaximus

User metadata
Rank Oldbie
Rank
Oldbie

Last summer while restoring some old systems I opened all their PSUs to clean them. Since I don't know much about PSUs in general I took pictures in the hope of asking you guys for your opinion on these vintage power supplies - are there any that look to be better quality than others or are they all similar? For starters some are noticeably heavier than others but I don't know if it makes a difference.

#1 Seasonic SSA-200G

maKEGoSl.jpg

f0FnJBRl.jpg

#2 "Energy Casper" Taiwan YoungYear

8B6gr15l.jpg

sDfvzATl.jpg

#3 morexplus PPS-200

tbOxJvLl.jpg

Nw0Ma2Tl.jpg

#4 Morex 230w

hpZd8ggl.jpg

e2AWSogl.jpg

#5 Everbest 220w

bciArDjl.jpg

3SXuI5Vl.jpg

#6 Soletek LYP-200 B

4M4ai8ol.jpg

1mxeRcBl.jpg

Reply 1 of 12, by SW-SSG

User metadata
Rank Oldbie
Rank
Oldbie

The two that really reek of cost-cutting are the morexplus and the YoungYear. The others look alright enough; however, given how ancient some of those look (namely those last two), I would be suspicious of those capacitors even if they aren't currently showing any outward signs of failure.

Reply 2 of 12, by ODwilly

User metadata
Rank Oldbie
Rank
Oldbie

The Seasonic is particularly nice. Agreed with SW-SSG, although the heatsink in that Soletek screams cheap to me. The Morex 230 doesn't look that bad

Main pc: AsRock x370 Killer SLI a/c, Ryzen 5 2600, 1tb WD black nvme ssd, 24g ddr4 2400 @2933mhz, rx 480 8gb reference card, 2tb Hitachi Deskstar.
Retro PC: Soyo P4S Dragon, 3gb ddr 266, 120gb Maxtor, Geforce Fx 5950 Ultra, SB Live! 5.1

Reply 3 of 12, by MMaximus

User metadata
Rank Oldbie
Rank
Oldbie

Thanks guys that's very helpful. I had a feeling the Seasonic might be the best of the bunch, plus I had a similar one in the '90s that I gave away a few years ago so I'm definitely keeping this one.

May I ask how you arrived to your conclusions? In general, what does one need to look for to gauge the quality of a PSU?

Reply 4 of 12, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

#1 Seasonic SSA-200G
Good

#2 "Energy Casper" Taiwan YoungYear
Bad

#3 morexplus PPS-200
Bad

#4 Morex 230w
Good (strange layout though)

#5 Everbest 220w
Questionable (I'm always leery of the quad diode treatment)

#6 Soletek LYP-200 B
Bad (cost cutting on the heatsinks)

All hail the Great Capacitor Brand Finder

Reply 5 of 12, by SW-SSG

User metadata
Rank Oldbie
Rank
Oldbie
MMaximus wrote:

... May I ask how you arrived to your conclusions? In general, what does one need to look for to gauge the quality of a PSU?

The morexplus has missing input filter components, which are some of the first things to go with PSU corner-cutting, combined with small heatsinks and an overall relatively barren interior. The YoungYear (and the others) does appear to have input filter, but at the same time has no heatsinks and from there I start to worry about the quality of the other components, after being exposed to (I assume) more heat than they would usually experience in a PSU with proper heatsinking.

morexplus.png
Filename
morexplus.png
File size
579.33 KiB
Views
459 views
File comment
morexplus; no input filter
File license
Fair use/fair dealing exception
morex.png
Filename
morex.png
File size
365.43 KiB
Views
459 views
File comment
Morex; yes input filter
File license
Fair use/fair dealing exception

I can only assume that "morexplus" is some sort of value- or wallet-friendly sub-brand of Morex.

Reply 6 of 12, by SirNickity

User metadata
Rank Oldbie
Rank
Oldbie

Seasonic:
Seems pretty decent to me, and Seasonic is a known reputable brand. (That doesn't guarantee they've never manufactured a dud, though.)

YoungYear:
Wouldn't even power this one up on a dare.

MorexPlus:
Definitely a cost-cut PSU. Probably not dangerously bad (at least they invested in heatsinks) but seems to have omitted some of the AC stage, so I would avoid it.

Morex:
Doesn't look bad at all.

EverBest:
I'm always leary of Guy-In-A-Basement brand names like this, but I don't see anything that looks obviously wrong with this one. The cables exit the side of the enclosure, which may not work in some tower configurations. If you own a Dremel, that problem is easily solved, though.

Soletek:
Obviously cost-cut, but doesn't look TOO bad. The important stuff is there.

Now, about how you would tell... This can be tricky, and it's really as much a gut feeling you get when you know a little bit about PSU design. Bear in mind, there are two factors that contribute equally to a good PSU: Good design, and good components. Cost-cutting can compromise either or both of those factors.

Let's talk about design first.

A typical switch-mode PSU will have a few stages, beginning with the AC section. Generally, the topology goes something like: Bring in the mains, go through a fuse, exit out to the front panel switch, arrive back in the PSU enclosure and go through a filtering stage (to prevent junk coming in and going out of the PSU -- the switching PSU topology is necessarily noisy by nature), rectify the AC to DC, and finally store the rectified bumpy DC in a couple of bulk capacitors.

Someone mentioned a distaste for the discrete diodes in the AC rectifier, vs. a bridge rectifier component. Personally, I don't really care -- it does the same thing. The only thing that matters is whether the diodes (discrete or single component) can dissipate heat well enough at their rated current. I would guess the discrete diodes are probably 5A parts, based on the somewhat obscured picture. That should be quite adequate, but maybe not as "heavy duty" as the case claims. 😉 The chunky bridge rectifiers on the other PSUs are usually elevated off the PCB, so they have more clearance for airflow which helps dissipate heat. Whether it's necessary or not depends on how much current is flowing through them, which is determined by the (designed) load. So... meh.

AC filtering is a sign of good design. It will work fine without, but if you don't see a boxy film capacitor across the mains, and/or disc caps from each leg to ground, and a choke (could be a toroidal coil, or a block that looks like a transformer) to filter common-mode noise, it's spewing junk into your house wiring, and that means it wasn't very well engineered. Move on to something better.

If there isn't a fuse between the mains input and the case switch, it isn't safe to use. Throw it away.

The mains caps take the brunt of abuse and are very important to the viability of the supply. Ideally they would be name brand (but often aren't), rated at LEAST 200V each (they're usually stacked in series across the mains, so the rating across both is really 400V), and have a capacity around 470uF. Numbers significantly less (e.g. 220uF) imply weak reserve capacity. Ideally, ALL capacitors, but particularly the mains caps, will be placed away from sources of heat like heatsinks and the bridge rectifier. There isn't a ton of room in most supplies, so just "not touching" sometimes is the bar you have to set. Minimum temperature rating is 85C, but 105C is better. (It's not just ambient temperature that matters -- it's also internal working temperature under load, and the higher the temp rating, the longer it will last at lower temperatures.)

OK -- next stage. The HV DC is chopped by a transistor -- which should be on a heat sink -- and fed through the largest transformer in the chassis. This converts the HV DC to high-current, low-voltage DC. From here, most PSUs will either have multiple transformer taps that provide the 12V, 5V, -12V, -5V rails, or some might be derived from others through additional switching stages. (In that case, you might see things on the label like "Max combined 3.3V and 5V = 150W" or similar.)

Small, light, janky-looking transformers are a sign of trouble. Tiny, pitiful heatsinks mean it won't be able to maintain stable operation under load, or it will just up and fail catastrophically. I don't even see the switching transistors on the YoungYear PSU. I'm not sure how that one even works.

Again, look for QUALITY name-brand capacitors on the output side, preferably with a 105C rating. The caps SHOULD be spaced away from heatsinks as much as possible. Size is dependent on a few factors, such as rated load, whether they're using multiple parallel caps per rail, the switching frequency of the DC-to-DC stage, etc. 2200 to 3300uF is a good baseline. If you see one 2200uF cap on the 5V rail, and a sticker claiming 15A, it's probably junk.

The output stage should have some coil inductors -- either toroidal or the kind that looks like a vertical post with thick wire wrapped around it. Ideally at least one per rail. This is a filter inductor, and helps make sure the output DC is smooth and relatively noise-free. It can be done with JUST a capacitor, but it will have considerably more ripple.

The final thing is overall sense of care in the design and parts choice. There should be some monitoring circuitry somewhere. It could be discrete components on the PCB (look for ICs near the output), surface mount stuff on the solder-side of the PCB, or a separate little card mounted vertically on the PCB (e.g., Soletek). This is typically the brains of the PSU, and will watch for things like over-voltage, under-voltage, over-current, excessive ripple, no-load conditions, etc. This is important because, if the PSU fails, it should shut itself down safely without taking your downstream load with it. The PSUs with barely any components on the board are engineered to do the bare minimum required to provide power. Not good.

Look also at cable gauge. Thin, scrappy wires imply an emphasis on cost. They don't need to be monsters, but they should feel adequate. If they're too thin, at the best case they'll drop some voltage as heat along the wire. At worst, they'll get hot under load and be prone to insulation failure under a fault condition -- particularly if the over-current protection is under-designed.

Now, at this point in the lifetime of all these PSUs, I consider re-capping a necessary maintenance procedure. The input filter caps tend to fail in a spectacular way after so many years, and the electrolytic (bulk, round cylindrical) caps aren't likely to meet their specifications anymore either. In some cases, they will leak corrosive electrolyte on the PCB, or may be bulging at the top. Even if they look OK, the internal resistance may be increasing, which generates more heat, leading to failure, and contributes to excessive output ripple. Replacing these parts is relatively simple, but it does help to know a thing or two about capacitor design to pick suitable replacements. I also replace the fan as a matter of course.

If you're testing these supplies on a bench, keep in mind they don't like to operate without a load. I've got some Delta supplies I recently re-capped, and they really wanted to be attached to a computer. Just having a HDD and optical drive attached, the regulation was terrible (11.2V on the 12V rail, and 5.3V on the 5V rail). One of them went into protection mode with just the HDD plugged in. That's actually a good thing -- if it hadn't, it might've destroyed itself or the HDD. A good solution is to use some automotive incandescent bulbs, or similar hefty load, to give the PSU something tangible to power. Load resistors are OK too, but you really should pull a couple amps, which will create some heat. Get those high-wattage sand-cast resistors and keep them in free-space, or at least off of paper and skin! 😀 Create a Molex test-harness that puts some load on both +12V and +5V. An old broken motherboard might work as well, as long as it's actually pulling current. I don't really like testing on good components, because they might turn into bad components if the supply is unhappy.

Hope this short novel is helpful. 😉

Reply 7 of 12, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
SirNickity wrote:

Someone mentioned a distaste for the discrete diodes in the AC rectifier, vs. a bridge rectifier component. Personally, I don't really care -- it does the same thing. The only thing that matters is whether the diodes (discrete or single component) can dissipate heat well enough at their rated current. I would guess the discrete diodes are probably 5A parts, based on the somewhat obscured picture. That should be quite adequate, but maybe not as "heavy duty" as the case claims. 😉 The chunky bridge rectifiers on the other PSUs are usually elevated off the PCB, so they have more clearance for airflow which helps dissipate heat. Whether it's necessary or not depends on how much current is flowing through them, which is determined by the (designed) load. So... meh.

Quad diodes imply the manufacturer was too cheap to use a FWB module and just grabbed whatever was available. It's not a definite indicator of lack of quality, but it's a warning sign telling me to check semiconductor ratings, features of the protection IC, etc. If they've cheaped out on the diodes, they might have cheaped out elsewhere. Of course, that's difficult to do without a closer examination.

SirNickity wrote:

AC filtering is a sign of good design. It will work fine without, but if you don't see a boxy film capacitor across the mains, and/or disc caps from each leg to ground, and a choke (could be a toroidal coil, or a block that looks like a transformer) to filter common-mode noise, it's spewing junk into your house wiring, and that means it wasn't very well engineered. Move on to something better.

If there isn't a fuse between the mains input and the case switch, it isn't safe to use. Throw it away.

Not just a fuse but a Zener diode or MOV should also be included. APFC designs have some capability of suppressing voltage transients without, but a complete design has a Zener or MOV anyway.

The input filter caps should also be checked to verify they're safety rated, both X and Y type in their intended applications. Cheap manufacturers use standard caps which present a hazard of shock or fire. Self contained input filter cans are another solution that I like even better.

SirNickity wrote:

The mains caps take the brunt of abuse and are very important to the viability of the supply. Ideally they would be name brand (but often aren't), rated at LEAST 200V each (they're usually stacked in series across the mains, so the rating across both is really 400V), and have a capacity around 470uF. Numbers significantly less (e.g. 220uF) imply weak reserve capacity. Ideally, ALL capacitors, but particularly the mains caps, will be placed away from sources of heat like heatsinks and the bridge rectifier. There isn't a ton of room in most supplies, so just "not touching" sometimes is the bar you have to set. Minimum temperature rating is 85C, but 105C is better. (It's not just ambient temperature that matters -- it's also internal working temperature under load, and the higher the temp rating, the longer it will last at lower temperatures.)

The output filter caps face more stress due to the higher operating frequency of the secondary.

SirNickity wrote:

The output stage should have some coil inductors -- either toroidal or the kind that looks like a vertical post with thick wire wrapped around it. Ideally at least one per rail. This is a filter inductor, and helps make sure the output DC is smooth and relatively noise-free. It can be done with JUST a capacitor, but it will have considerably more ripple.

Indeed, a proper pi filter is ideal.

All hail the Great Capacitor Brand Finder

Reply 9 of 12, by SirNickity

User metadata
Rank Oldbie
Rank
Oldbie

Hm, maybe I just don't have enough experience gained from tearing down dubious power supplies, so I'll grant you, anecdotally, there may be some correlation that implies you should scrutinize the rest of the design. That said, I use discrete diodes in all of my own designs, and it's absolutely not for reasons of cost. I've never taken the time to compare, actually. Sometimes I use axial, sometimes I use TO-220. IMO, and technically, a single-component FWBR is just a different package, and that's all. I guess it all contributes to that hunch based on trends you've experienced over time.

Point taken on the operating frequency of the secondary capacitors. It was just an off-the-cuff statement, but my thinking was that primary capacitors will see more ripple current (sum of all the rails, plus losses), are subject to any line transients that aren't caught by the input filters, and potentially suffer an imbalance when connected in series across the HVDC -- especially as they age. Also, they'll be subject to the same HF switching, but as a current source rather than sink. They'll be rated for higher ripple, of course, but still... life ain't easy for those guys. 😀

100% agree on the MOV, and X/Y ratings. At this age, the MOV is probably useless and should also be chucked, though. Arguably, you could leave it out since it would be easier to maintain a power strip with transient suppression than individual ones in a PSU, but then you forego that protection when plugging in to a wall outlet directly. Engineering is compromise.

Reply 10 of 12, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
SirNickity wrote:

Hm, maybe I just don't have enough experience gained from tearing down dubious power supplies, so I'll grant you, anecdotally, there may be some correlation that implies you should scrutinize the rest of the design. That said, I use discrete diodes in all of my own designs, and it's absolutely not for reasons of cost. I've never taken the time to compare, actually. Sometimes I use axial, sometimes I use TO-220. IMO, and technically, a single-component FWBR is just a different package, and that's all. I guess it all contributes to that hunch based on trends you've experienced over time.

TO-220 is far superior to both from the point of view of power handling. I would have no problem with a consumer SMPSU using that for the primary rectifier although it would be quite over-engineered in all likelihood. Surface mount discrete diodes can also be good as they mitigate some of the footprint issues of axial and TO-220 diodes and can also be readily coupled to a heatsink.

Every time I see axial diodes in a volume production SMPSU used as the FWB, I wonder why they went that route. A bumper supply from the parts recyclers? Common stock with the half bridge primary unit? I really don't perceive any advantages that have anything to do with quality.

All hail the Great Capacitor Brand Finder

Reply 11 of 12, by MMaximus

User metadata
Rank Oldbie
Rank
Oldbie
SirNickity wrote:

...

Hope this short novel is helpful. 😉

Thank you so much guys for your contributions 😀 I admit the discussion quickly got way too technical for me but I might come back to this thread for reference one day if I ever learn more about electronics.

I have to say I'm impressed about your level of knowledge though. Do you all have a background in electrical engineering?

Reply 12 of 12, by SirNickity

User metadata
Rank Oldbie
Rank
Oldbie

Just another hardware hacker learning what I can from discussions like these.

I know things got pretty deep, but there just isn't an easy answer to questions like "what makes a good PSU". Unfortunately, if you have to ask to begin with, the explanations often don't make a lot of sense. I know this from experience! 😀 BUT, I have found that if you understand 10% of what you read, that's a little bit more than you knew before -- and the next time you come across that same info, you catch a little bit more still. It all starts to sink in eventually. So just take what you can from it, and that's good enough.