Power supply build quality pictorial. part 2

[Behemot]

Re: Power supply build quality pictorial. part 2

Nothing like that. There is very mild wave-like ripple, but overall ripple is very good, as I stated. I would personally rather see 220+ uF just because of capability to withstand micro dropouts in the power network, but that's all. The talk in CM6805 datasheet seems to be legit as the PSUs results in both votlage control and ripple are very good for its price. You have to accept the fact technology is going further. I have been suspicious as well, but I checked and it works

80 % load, +12 V

100 % load, +12 V

[mariushm]

Re: Power supply build quality pictorial. part 2

They are selling it here for some time, I guess you have somewhat old infos

I didn't see it Romania.
Heard about this psu here: http://www.jonnyguru.com/forums/show...ighlight=vs450

Well what do you know, you made me look, seems they popped up on my favorite online store in the last week or so.

Corsair VS Series VS350 157.88 RON / 49 dollars / 36 euro
Corsair VS Series VS450 167.84 RON / 51.5 dollars / 38.66 euro

(this includes our stupid high 24% vat)

at these prices, i'd rather go for a Sirtec/High Power Element Plus 400w at 177 RON, at least that has Teapo capacitors.

http://www.mygarage.ro/teste-si-revi...-400-g12s.html

Yeah, technically it's not a better design, it's group regulated, but I know Corsair as a brand needs to have a much larger profit margin than Sirtec, so screw them.

[Behemot]

Re: Power supply build quality pictorial. part 2

I found them here starting at ~32 euros/40 dolars.

Crapxon or teapo, all the same shit, all will vomit itself inside out, it's just a matter of time. That HP is interesting only because of it's 80+ certification.

[tom66]

Re: Power supply build quality pictorial. part 2

If you only look at 10us/div you're never going to see 100Hz ripple.
Set scope to 2ms/div (giving two or three cycles) and ideally use a line trigger.

This is what 100/120Hz ripple on output looks like:


Low ESR caps do not like 100/120Hz. Their ripple current ratings are typically halved at that frequency. In addition as the voltage variation is much greater over time the current through the cap (I = C dV/dt) is greater. So seeing that basically means: goodbye output caps, no matter how good or bad they are.

I have seen plenty of SMPS fail like this, and it often takes out the output capacitors before it dies completely.

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

Crapxon or teapo, all the same shit, all will vomit itself inside out, it's just a matter of time.

This is just wrong. I have seen psus with good Teapos 8 years old or more, but only once I saw Capxon caps more than 3 years old and not bulging.

[Behemot]

Re: Power supply build quality pictorial. part 2

No, you are wrong. I have seen bad Teapos. (quite many of them actually) That tells me enough.

…

It has active PFC, is that even possible to get such oscillogram with active PFC? It should inject the capacitor with pulses around 400 V, the cap should never see power grid frequency ripple.

[Wester547]

Re: Power supply build quality pictorial. part 2

I don't see that thing doing 450W for very long. Those heatsinks are not very big.

The 180uF primary also means you'll likely see 100/120Hz ripple on the output near maximum loading. See it on a lot of PSUs with undersized primaries....

It was to my knowledge that it is power supplies that use full wave rectification (half bridge, full bridge, and push pull topology) that output a ripple frequency (or switching frequency) that is double that of the mains (half wave rectification's ripple frequency is equal to that of the mains). In simple terms, this does two things: the higher switching frequency allows rectification inside the PSU to happen at a higher frequency which means more can be done, but this also means twice the ripple (much more EMI) going across the internals and out to the load/computer. The PSU in question looks to use single switch forward topology (half wave rectification, along with two switch forward and flyback topologies, though flyback is terribly inefficient to begin with) so I'm not sure that it would have this problem (it is usually 100/120Hz ripple and not 50/60Hz ripple that causes AC line drop outs, unusual noise, and voltage regulation issues, 50/60Hz noise, etc), though APFC primaries are more stressed than non-APFC primaries and the lower switching frequency does undercut what they can handle a might bit. The PSU also might be able to get away with it with having the fan rev up to a much higher speed when the load/temperature increases. Actually, this, in non-APFC PSUs, is what passive PFC is there to facilitate and almost do away with problem wise as a 50/60Hz filter/inductor. That said, good PSUs will use APFC caps whose microfarad rating is within at least 33% range to the output watts~microfarad ratio. It also looks like the main transformer in the above PSU has some good windings or transformer taps.

Low ESR caps do not like 100/120Hz. Their ripple current ratings are typically halved at that frequency. In addition as the voltage variation is much greater over time the current through the cap (I = C dV/dt) is greater. So seeing that basically means: goodbye output caps, no matter how good or bad they are.

Well, actually, the datasheets I've seen indicate that their ripple current is higher at higher switching frequencies and lower at lower switching frequencies (by 10-50%, from what I've seen). However, what this also means is that they have to handle twice the ripple current at twice the switching frequency, so your point remains.

On the subject of Teapos, 10mm and higher Teapos are much better than 8mm Teapos.

[tom66]

Re: Power supply build quality pictorial. part 2

It has active PFC, is that even possible to get such oscillogram with active PFC? It should inject the capacitor with pulses around 400 V, the cap should never see power grid frequency ripple.

It'd be great if that were the case but it's not. Ripple across that is typically 20~30Vp-p under light loading. Beyond about 50Vp-p the SMPS controller may start encountering issues with regulating the output.

The active PFC keeps current/voltage in a nearly resistive ratio. To do this means you'll have ripple across the main cap. No free lunch. A very good PFC controller can minimise this ripple. From what I have found, Quasi-PFC supplies, with typically 0.9-0.95 PF, can manage with smaller primary caps.

[Behemot]

Re: Power supply build quality pictorial. part 2

This thing managed PF of 0,97 starting with 20 % load and moved to 1 at 40 %, where it remained till the maximum load. So I think there really IS something about the CM6806+CM03X combo. As I already said, it explicitly states it has advanced driving which can work with weaker silicon and smaller input cap. Go on, check the datasheet for yourself.

[c_hegge]

Re: Power supply build quality pictorial. part 2

I may be getting myself a VS350 to review at some point. I can tyell you now that with Jun Fool caps, it's not going to do any better than a 7/10 (assuming everything else is perfect, but knowing CWT, I'll also have to take an extra point off for soldering).

[tom66]

Re: Power supply build quality pictorial. part 2

This thing managed PF of 0,97 starting with 20 % load and moved to 1 at 40 %, where it remained till the maximum load. So I think there really IS something about the CM6806+CM03X combo. As I already said, it explicitly states it has advanced driving which can work with weaker silicon and smaller input cap. Go on, check the datasheet for yourself.

Get an isolating transformer and measure the ripple if you want. You'll find there is a considerable 100/120Hz component. I highly doubt a PF of 1.0, because that's perfect, and even the best PFC frontend could only really reach 0.995~0.999 under ideal conditions. Quasi PFC is fine, too. It's a low cost solution and it's absolutely fine to have a PF of "only" 0.9. This isn't some competition to get the highest PF and you aren't paying for it, so why does it matter?

I don't like PFC for a few reasons:
- Added cost.
- Decreased efficiency (losses in preconverter) massively outweigh wiring losses.
- Reduced reliability.
- Primary cap under greater stress. Easier to cheap out on without obvious consequences until the primary cap bloats. Even good brand will die with poor PFC circuit.

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

No, you are wrong. I have seen bad Teapos. (quite many of them actually) That tells me enough.



It has active PFC, is that even possible to get such oscillogram with active PFC? It should inject the capacitor with pulses around 400 V, the cap should never see power grid frequency ripple.

I have seen bad Rubycon more than 3 times. That tells me enough

[Behemot]

Re: Power supply build quality pictorial. part 2

Your bad. If I'd see them dying before rated time under rated conditions in datasheet, I will move them into bad caps list. So far I haven't.

[c_hegge]

Re: Power supply build quality pictorial. part 2

Any cap will fail if you abuse it, rubycon included (heat, over-voltage, etc). Teapo, however, don't require any abuse or harsh treatment to make them fail. They will fail after 3-4 years of perfectly normal use, Just like Fuhjyyu and CapXon. I'm yet to see a Rubycon do that.

[Behemot]

Re: Power supply build quality pictorial. part 2

That's what I'm talking 'bout brother…

[japlytic]

Re: Power supply build quality pictorial. part 2

TPI Electronics KHS160-3 (from Philips PM3585)

Old-school computer power supply style.
Pair of Elna 330uF/200V and a pair of Rubycon USP 220uF/200V after high voltage DC inductors on primary side.
All other electrolytic capacitors are Nicon TN Series.
2200uF 25V x2 before output inductor to +12V regulator with 2200uF 16V after output inductor to +12V regulator.
2200uF 16V x2 after +12V regulator.
4700uF 10V x2 before +5V output inductor with 4700uF 10V x3 after it.
2200uF 10V before output inductor to -5V regulator with 2200uF 10V after output inductor to -5V regulator.
2200uF 10V x2 after -5V regulator.
470uF 35V for the B+ for the control section.

[ben7]

Re: Power supply build quality pictorial. part 2

TPI Electronics KHS160-3 (from Philips PM3585)

Old-school computer power supply style.
Pair of Elna 330uF/200V and a pair of Rubycon USP 220uF/200V after high voltage DC inductors on primary side.
All other electrolytic capacitors are Nicon TN Series.
2200uF 25V x2 before output inductor to +12V regulator with 2200uF 16V after output inductor to +12V regulator.
2200uF 16V x2 after +12V regulator.
4700uF 10V x2 before +5V output inductor with 4700uF 10V x3 after it.
2200uF 10V before output inductor to -5V regulator with 2200uF 10V after output inductor to -5V regulator.
2200uF 10V x2 after -5V regulator.
470uF 35V for the B+ for the control section.

Wow, that is the most over-capped PSU I have ever seen!

[tom66]

Re: Power supply build quality pictorial. part 2

Then you have not seen this.

HP 54501A oscilloscope power supply.

All Chemicon capacitors.

~20000uF on the 3.3V digital supply.
~8000uF on the 15V CRT+fan supply.

After 18 years, I had to fix a cold joint on the power connector for the -12V analog rail (it had been hand reworked), the trace was going crazy and scope was randomly resetting. That was all; all the caps were still fine.

[Behemot]

Re: Power supply build quality pictorial. part 2

Well for scope it is a must I'd say

[japlytic]

Re: Power supply build quality pictorial. part 2

I forgot to mention about the power semiconductors in the Philips PM3585 power supply:
2SD1062 regulator transistor for +12V/2A rail.
2SD1062 regulator pass transistor for -5V/3A rail.
SF164A rectifiers for +12V/2A and -5V/3A rails.
CTB34 rectifier for +5V/20A rail.
RS605 primary bridge rectifier.
2SC2625 primary switcher pair.