• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

SMPS supplies are rubbish

OP
restorer-john

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,716
Likes
38,889
Location
Gold Coast, Queensland, Australia
I've built at least a dozen PCs over the past 25 years and never seen anything like it.

Looks like phenomenal regulation.

So you've built 12 gaming PCs in 25 years for an average life of 2 years or so? By my calculations, you were running a P166+, 420MB HDD, rocking 32MB of RAM, a SB Pro, a Cirrus Logic VGA card and playing Doom off 3.5" diskettes back then... :)
 
Last edited:

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
16,052
Likes
36,427
Location
The Neitherlands
Seasonic has a new group of PC power supplies that has amazing regulation. I have one of them in my gaming rig. In this review the 12V rail from 6.38A to 82.94A of load the voltage swing was from 12.263V to 12.266V.

https://www.tomshardware.com/reviews/seasonic-prime-ultra-platinum-1000w-psu,5397-4.html

On my particular example using the motherboard monitoring with anything from idle to the 600W I can load it up with on synthetic benchmarks I've never seen it change from 12.168V full stop. I've built at least a dozen PCs over the past 25 years and never seen anything like it.

PC power supplies are not really small and very well ventilated and not really cheap either.
Usually they run well below their maximum ratings.

Mostly the ones giving up the ghost are the ones in cheap DVD players, wall warts and some bricks. Certainly MOST of them keep working but not all of them.
It seems Restorer-John and I have both been in the repair business for over 30 years and have seen and remembered our share of failed parts.
The 'skewed' part may be that we only got defective equipment in our hands and often could already see from the type-number what would likely be the problem with certain devices and be proven right almost every time.
People come to us when something is not working, not when something works fine.
So when over a short time period a bunch of SMPS collect in the rubbish bin and almost every device these days comes with wall warts, bricks or crappy (cheap) built in non ventilated SMPS we sometimes may get carried away and call them all crap.

In my experience (repaired quite a few) the root cause is almost always the 'start-up' capacitor. This one is almost always mounted very close to a heat sink and is supposed to dry out and stop functioning before the rest of the caps dry out and can cause massive failures and perhaps even start a fire inside. Them being plastic doesn't help. So there is a time-bomb in there on purpose.
Once the device is switched on it will keep on working till it has been switched off or disconnected from mains and simply won't start up the next time.

To get it working again usually the secondary side caps (the bigger ones) need replacement, with the correct types, don't use generic ones but with the correct ESR ! and re-solder parts that became hot and the transformer as they may have 'solder-rot' and on the primary side the start-up cap.

Sure I can see many testimonies of SMPS working fine but that doesn't mean some cheapies are crap.
Basically not all SMPS are crap for sure but can see why Restorer-John is irritated with them bricks. It mirrors my experience.
 

maty

Major Contributor
Joined
Dec 12, 2017
Messages
4,600
Likes
3,170
Location
Tarragona (Spain)
Years ago, when I bought a PSU to my custom PC (I5-4400 -TDP 84 watts-, without graphics card), I choose the cheap but very good german bequiet! Pure Power L400W because the low ripple (and silent too), specially at 5V. € 48.

https://www.kitguru.net/components/power-supplies/zardon/bequiet-pure-power-l8-400w-review/6/

The new, big and expensive Seasonic Prime Ultra Platinum 1000W has half one third ripple to the same load!

https://www.tomshardware.com/reviews/seasonic-prime-ultra-platinum-1000w-psu,5397-9.html


PS: my PC, without graphics card, must be < 100 watts.
 
Last edited:

captain paranoia

Active Member
Joined
Feb 9, 2018
Messages
293
Likes
218
In my experience (repaired quite a few) the root cause is almost always the 'start-up' capacitor.

In my more limited experience, it has been the capacitor on the rectified mains. This is a high voltage location, and the devices never seem to be over-rated. Consequently, these capacitors fail, losing value so that the regulator tries harder to maintain the output, until it burns the switch out. If the regulator had better fault detection, it would shut down before blowing the switch off the PCB...

Another fault I've seen is using three parallel schottky rectifiers in one supply. Unfortunately, the schottky diode has a negative Vf tempco, so, as one device heats, it takes more current, so gets hotter, and thermal runaway occurs, burning the device out to short circuit. Replacing the triple with a single package, dual rectifier sorted that one.
 

Timbo2

Senior Member
Joined
Feb 27, 2018
Messages
497
Likes
396
Location
USA
Looks like phenomenal regulation.

So you've built 12 gaming PCs in 25 years for an average life of 2 years or so? By my calculations, you were running a P166+, 420MB HDD, rocking 32MB of RAM, a SB Pro, a Cirrus Logic VGA card and playing Doom off 3.5" diskettes back then... :)

Pretty much! I was playing Doom after hours on a work network to try out a multiplayer FPS.
 

trl

Major Contributor
King of Mods
Joined
Feb 28, 2018
Messages
1,981
Likes
2,552
Location
Iasi, RO
8x24" LG LED monitors used in our offices. 4 of them on a single computer (quad screens) used 4-5 hours a day, the other 2 in 2 pairs on other PCs that see 1-2 hours per day.

12 months old approx. They started failing 6 months in and now 5 out of 8 PSUs have died. We are only talking 23-32 Watt supplies (max).

Here are 4 of them (dead):
[...]
Let me open them up and we'll do some autopsies...

(top bet: bad caps)

Open'em up and let's see what's failed in there. :)

BTW, just bought an Alibaba 8A/3...30V buck-converter recently based on XLSEMI chip. I don't like how it measures, but I know that it's barely warm, so very efficient. I'll see if I can blow it away. :)
 

Johnplayerson

Member
Joined
Dec 28, 2019
Messages
45
Likes
17
I must say I am impressed to see this thread by restorer John. This is the new world of making big money while still cheaping out on parts with power supplies. Smps power supplies in particular, require high grade long life 105c, low esr capacitors. It has never been a secret. Manufacturers of most products seem to want to produce SMPS power supplies, that will last just long enough to have the buyer rebuy the product down the road.

Planned obsolence, I like to call it.

With power amplifiers, it was almost insane to see $9000 Crown I tech amplifiers with cheapo, 85c teapo capacitors in the power supply. After years of me complaining of this issue, (last ten or so), Manufacturers at least started using 105c rated, lower esr capacitors in the power supplies. Still the power supplies tend to fail still, however not as bad as they used to, when they produced expensive QSC and Crown SMPS amplifiers with cheapo
85c capacitors.

Much like restorer john. I would never buy anything with an SMPS power supply, short of the computers that come with no other!!!. I was planning on making some Led entertainment lighting, and could have bought a rackmount smps supply for this. Instead I found a good deal on 6 middle atlantic pd 915r racks, and picked up about 25 1000ma 12 volt tranformer based power supplies from value village. Being a project type of person, since most of these working power supplies are already 15 years old, I open them up and replace the capacitor with UCC KYB. Now I have a power supplies that will pretty much last forever.

I look forward to the day some poster can say, MY 35 YEAR OLD SMPS POWER AMPLIFIER IS STILL WORKING.
 

knobtwiddler

Member
Joined
Nov 19, 2019
Messages
35
Likes
28
SMPS design and construction is not a job for junior engineers. That is the underlying problem, IMHO, SMPS can be made to last for decades (except maybe a change of e-caps). I have several SMPS-powered lab equipment from the early 90'ies that has been running heavy duty, never had any issue.

This is very true. If all SMPSs were inherently unreliable, there wouldn't be such a plethora of medical apparatus which uses them. The fact is, if you factor a large lump of copper out of the equation (i.e. large-ish transformer needed for linear PSU), then you will see that there is an 'underclass' of ultra-cheap SMPS PSUs that can be tapped into. These are the sort of PSUs that are designed to *just* meet their design criteria. If there is the occasional failure, it's not the end of the world. I am dubious about how well some of these bottom-feeder PSUs meet CE and FCC regs for radiated and conducted emissions btw (Dead-ringer's FCC fine didn't put them off!).

As you suggest, if you look into high-end lab equipment that's designed without excessive concern for the bottom line, you see robust, low-noise SMPS designs. These types bear no relation to the kind used to power your printer or tablet, other than the fact that they are of the SMPS variety.

It ought to be noted that making a good ole linear's DC output devoid of mains-related ripple (50Hz in Europe / 60 in USA) is quite an artform. A well-designed SMPS can have virtually zero artefect in this range off the bat. However, it will have demons in the 300KHz+ areas. This means that specialist knowledge is required in order to whack these moles (as well as the R+D budget and desire). If you're doing 'just enough' to make a serviceable commercial product, you may not worry about anechoic testing with high-end analysers.

NB - most common SMPS failure is Drain-Source short circuit in the MOSFET. Get a bleeper and see for yourself.
 

trl

Major Contributor
King of Mods
Joined
Feb 28, 2018
Messages
1,981
Likes
2,552
Location
Iasi, RO
I do have a bunch of defective SMPSs from DELL, HP, Lenovo (probably cables, MOSFETs, haven't investigated) and also a lot of no-name defective ones too (from lighting appliances).

However, I do like that, at least in the audio field, manufacturers started to certify their SMPSs (FCC, CE etc.) and the newer SMPS designs started to measure pretty good too. Here's a 0.01% AC ripple & noise measured in 24V/1.3A load (real circuit load, not just resistive load!). About 2.4mV RMS of ripple & noise out of 24V DC (although, peaks are reaching to 24mV, so that would be 0.1%).

C3X_power_brick_ON_AC.png

Using a linear regulator as follower will further reduce the ripple to less than 0.5mV, so definitely good for audio related stuffs.

It's also dead silent while operating, and I'm usually so sensitive that I unplug the TV and the power stripe that feed my bedroom laptop (yes, I can't sleep due to the slightest SMPS noise in my bedroom).
 

ayane

Active Member
Forum Donor
Joined
Dec 15, 2018
Messages
183
Likes
687
Location
NorCal
I'm usually so sensitive that I unplug the TV and the power stripe that feed my bedroom laptop (yes, I can't sleep due to the slightest SMPS noise in my bedroom).
I'm actually the same way with my desktop monitor and RME ADI-2 Pro fs. They're both with fantastic quality SMPSs, but they put out a faint continuous whooshing sound that keeps me awake and alert.
 

mkawa

Addicted to Fun and Learning
Forum Donor
Joined
Sep 17, 2019
Messages
788
Likes
695
it really depends on manufacturer. delta, lambda, seasonic, etc. all make EXTREMELY high quality SMPS. agilent, lambda, tdk all make incredible instrumentation SMPS. just because the cheap SMPS that come with your monitors keep dying does not mean that SMPS modules are crap.

sigh.
 

Veri

Master Contributor
Joined
Feb 6, 2018
Messages
9,598
Likes
12,040
it really depends on manufacturer. delta, lambda, seasonic, etc. all make EXTREMELY high quality SMPS. agilent, lambda, tdk all make incredible instrumentation SMPS. just because the cheap SMPS that come with your monitors keep dying does not mean that SMPS modules are crap.

sigh.
How about mean well medical grade?
 

knobtwiddler

Member
Joined
Nov 19, 2019
Messages
35
Likes
28
How about mean well medical grade?

Buying an SMPS module off the shelf is a little like getting a microwave ready meal at the supermarket. How do you know where the ingredients came from? It only takes one 2-cent capacitor to fail and the whole PSU is dead. If reliability is an absolute priority, you roll your own from scratch, chosing each component from trusted sources. Having said that, I'd imagine that Hypex (for example) use good components. Open-frame PSUs enable you to look for trusted components, unlike potted modules.

Much as I've seen SMPS modules fail, in the years I've worked in audio I've also seen transformers fail due to manufacturing faults (if one fails, it's imperative to do a PM on it, unwinding it to see where the failure was).
 

gene_stl

Addicted to Fun and Learning
Forum Donor
Joined
Jun 14, 2019
Messages
867
Likes
1,200
Location
St.Louis , Missouri , U.S.A.
I prefer linears because in 1969 I learned their topologies and formulae. I have had several dead toroids come across my bench. Switchers depend on electrolytics to work at all and after several decades are likely to have failures. Caps drying up in linears might just increase the noise level.
Like LED lighting SMPS are an idea whose time has come. I do repairs on an instrument with an Emerson division Artysyn open frame power supply and on units built in the nineties the failure rate for power supply problems is 25 to 30 %. The power supply often killing the control board in the process. I can still buy the identical supply new. (about $50)
 
Last edited:

NTomokawa

Addicted to Fun and Learning
Forum Donor
Joined
Jan 14, 2019
Messages
779
Likes
1,334
Location
Canada
How about mean well medical grade?
I have no idea about their "medical grade" power supplies, however many of the LED luminaire suppliers I represent use Mean Well drivers. They're cheaper than Philips drivers and have a different set of strengths and weaknesses.
 

Veri

Master Contributor
Joined
Feb 6, 2018
Messages
9,598
Likes
12,040
I have no idea about their "medical grade" power supplies, however many of the LED luminaire suppliers I represent use Mean Well drivers. They're cheaper than Philips drivers and have a different set of strengths and weaknesses.
I see :) I asked simply since I bought such an SMPS to replace an older, buzzing one. It has yet to arrive but the 'medical' grade made me get the expectation it will be better than average. Will see!
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,663
Likes
241,002
Location
Seattle Area
How about mean well medical grade?
I tested one on behalf of someone else and was surprised they still had mains leakage (due to use of Y capacitor). This was confirmed by contacting them. I removed the cap and problem was solved although it would likely not pass conducted emissions test.
 

trl

Major Contributor
King of Mods
Joined
Feb 28, 2018
Messages
1,981
Likes
2,552
Location
Iasi, RO
Top Bottom