"Interesting take" on the importance of capacitor quality within a PSU.

[RyzenMaster]

"Interesting take" is just a nice way of saying I don't buy it one bit, but I also know very little in this realm.

I was talking to someone in another tech related forum and essentially they said this about power supplies - "if it [any PSU] has all rail protections it is good."

They said that if an EVGA W1 had all applicable protections on each rail, it would be good. They said it is bad since it is missing OCP on the minor rails. They tried to show that the PSU had only OCP on one rail by showing one chip on the PCB that only has OCP for one rail, but then I explained that that was just a chip to manage the single standby rail, so it only needs to support one rail. I said the supervisor IC definitely says it has support for OCP on the 3.3v, 5v, and 12v rails. They said that is impossible since the supervisor IC did not have enough pins.

Am I reading the datasheet wrong? http://www.weltrend.com/en-global/pr...tail/66/88/301

When asked about capacitor quality, they said "If you have all rail protections your PSU literally can't die unless by user error." Their reason was basically since the protections will keep everything in check so the capacitors will not be under stress enough to fail, which seems logical, but I don't think this is the case 100% correct. I said higher end capacitors last longer, and that even with protections, a PSU with cost cuts like cheap capacitors may not last long and can fail quickly. They called higher end capacitors "snake oil" outside of audio situations.

So, can you explain if they are right or if I am wrong?

This conversation did beg the question:

I know very little about what makes a capacitors good other than temp rating, so I really didn't know what to say when they said that. What makes a cap good? And why is one bad versus the other? I know we tier them by brands, but is that completely subjective experience, or is there actually some objectivity to it?


I don't trust anything this person says since I think they are factually wrong 90% of the time given experiences with them in the past.

[Orion]

"if it [any PSU] has all rail protections it is good."

Easily refuted by the fact that a protection against one of the most critical hardware destroyers, ripple, doesn't exist.

I said the supervisor IC definitely says it has support for OCP on the 3.3v, 5v, and 12v rails. They said that is impossible since the supervisor IC did not have enough pins.

Am I reading the datasheet wrong? http://www.weltrend.com/en-global/pr...tail/66/88/301

Your link doesn't contain a (it can be found here https://cdn.datasheetspdf.com/pdf-do...7_Weltrend.pdf), but the specifications page doesn't lie, and the Weltrend IC certainly supports OCP on those rails. Your interlocutor is welcome to read the linked datasheet and notice that OCP on 3.3, 5, and even *two* 12V rails is detected on pins (respectively) 11&13, 10&14, 5&12, and 7&8.

The rest of your post is once again refuted by the fact that there is no protection against ripple.

[RyzenMaster]

We did get on the topic of ripple actually, and they said it only matters for overclocking. They also said the chip is only capable of protecting one rail at a time since it "needs a control input." What would that be for?

[Jon Gerow]

"Interesting take" is just a nice way of saying I don't buy it one bit, but I also know very little in this realm.

I was talking to someone in another tech related forum and essentially they said this about power supplies - "if it [any PSU] has all rail protections it is good."

They said that if an EVGA W1 had all applicable protections on each rail, it would be good. They said it is bad since it is missing OCP on the minor rails. They tried to show that the PSU had only OCP on one rail by showing one chip on the PCB that only has OCP for one rail, but then I explained that that was just a chip to manage the single standby rail, so it only needs to support one rail. I said the supervisor IC definitely says it has support for OCP on the 3.3v, 5v, and 12v rails. They said that is impossible since the supervisor IC did not have enough pins.

Yeah. Don't trust anything this person has to say. LOL!

[Jon Gerow]

We did get on the topic of ripple actually, and they said it only matters for overclocking. They also said the chip is only capable of protecting one rail at a time since it "needs a control input." What would that be for?

It's not. He's talking out of his ass.

[RyzenMaster]

It's not. He's talking out of his ass.

Yea, I basically said "why would I trust you know that capabilities of this chip more than the manufacturer, when you couldn't even identify the chip to begin with."

Is it even common for PSUs to only have OCP on certain rails?

About OTP on this unit. The supervisor IC doesn't appear to have support for it, but EVGA claims this unit supports it. How would that be handled? By a thermistor and a separate IC?

[Jon Gerow]

Yea, I basically said "why would I trust you know that capabilities of this chip more than the manufacturer, when you couldn't even identify the chip to begin with."

Is it even common for PSUs to only have OCP on certain rails?

Typically, PSUs do not have OCP. Even if the IC supports it. They tend to use OPP. But if the PSU has DC to DC, then odds are the controllers on those D2D cards will have their own OCP. And, as you already point out, the +5VSB has it's own IC that also has OCP.

About OTP on this unit. The supervisor IC doesn't appear to have support for it, but EVGA claims this unit supports it. How would that be handled? By a thermistor and a separate IC?

The common means of having OTP without a Supervisor IC that supports it is to simply have the fan controller trigger the PWM IC to shut off if a condition exists where the fan would need to spin at full RPM.

[RyzenMaster]

Typically, PSUs do not have OCP. Even if the IC supports it. They tend to use OPP. But if the PSU has DC to DC, then odds are the controllers on those D2D cards will have their own OCP. And, as you already point out, the +5VSB has it's own IC that also has OCP.



The common means of having OCP without a Supervisor IC that supports it is to simply have the fan controller trigger the PWM IC to shut off if a condition exists where the fan would need to spin at full RPM.

I don't think I understand what you mean in the last line.

[Jon Gerow]

I don't think I understand what you mean in the last line.

Sorry. I meant to type "OTP", but type "OCP". Many PSUs have OTP even if the supervisor IC doesn't support it because the fan controller will detect the higher temps and tell the PWM controller to shut down.

[RyzenMaster]

Ah okay, makes sense.