PSU efficiency stuff

[Hamster Homie]

Hey, quick question. My PSU is a CX550M and I wanted to see what the efficiency curve looked like and I saw different curves with different voltages attributed (screenshot below). Is it true that PSU efficiency can differ by input voltage ? I live in EU and I belive I have 220V comming out of my wall. Can smn explain to me wtf this all means ?

[LIGISTX]

3 minutes ago, Hamster Homie said:

Hey, quick question. My PSU is a CX550M and I wanted to see what the efficiency curve looked like and I saw different curves with different voltages attributed (screenshot below). Is it true that PSU efficiency can differ by input voltage ? I live in EU and I belive I have 220V comming out of my wall. Can smn explain to me wtf this all means ?

Higher voltage input to a PSU typically results in higher efficiency. Yes. 
 

I don’t remember all of my electrical engineering formulas…. And likely never will lol. But if you Google or Wikipedia it, I’m sure there are good answers. 

[IIIIIIIIII]

W = VA

 

At 110V, A has to be pushed twice as much as 220V, so that's less efficient.

[LIGISTX]

3 minutes ago, IIIIIIIIII said:

W = VA

 

At 110V, A has to be pushed twice as much as 220V, so that's less efficient.

Is this correct? I mean yes, that formula is correct, but isn’t the efficiency loss due to the internal switching of the power supply? Converting down to 12v is where the efficiency is lost if I remember correct…

 

Like I said, I don’t remember. But I didn’t think it was as simple as I’m America I’m drawing twice the current as someone on Europe. While that is true, is that the reason for the efficiency difference? 

[mariushm]

The AC voltage has to be rectified into DC using a bridge rectifier (or 2/several in parallel to reduce voltage drop and dissipate heat across rectifiers) 

A bridge rectifier consists of 4 diode, 2 of them are always operating ... so you get some voltage drop across the diodes as input voltage is rectified.

 

The power dissipated in the bridge rectifier will be higher at higher currents ... for example let's say to produce 400 watts to components, the power supply needs to draw 4A of current at 110v , but only 2A of current at 230v AC.

The losses in the bridge rectifier are approximately P = 2 diodes x 1v (voltage drop on diode inside rectifier) x Current   ... so at 110v you'd have 8 watts of wasted energy, at 230v you'd have only 3-4 watts.

 

After the AC voltage is converted to DC, that DC voltage with lots of fluctuations is boosted to around 400-420v DC using the active PFC circuit. The reason this is done is because you can store more energy in that bulk capacitor this way (giving the psu enough energy to last a few milliseconds in case there's a hiccup in the power grid) and because it makes possible to use much smaller transformers  and higher frequencies (the power supply sends tens to hundreds of thousands of 400v-ish pulses through the transformer and out comes nearly 12v DC on the other side.

 

That active PFC circuit has lower losses if it only has to boost from around 300v  to 400-420v  compared to boosting from around 180v.

Rectified AC has a peak of 1.414xVac so for 230v input you'll have a peak of around 325v DC and on 110v AC you'd have a peak of around 175v DC.

 

So the two above are just two of the reasons why running a power supply with a higher input AC voltage results in much higher efficiency.

 

 

Fun fact .. There are ICs which rectify AC voltage using mosfets instead of diodes, so the losses are much lower ... like orders of magnitude lower, depending on what mosfets you choose.

For example see LT4320 : https://www.analog.com/media/en/technical-documentation/data-sheets/4320fb.pdf

 

Problem is a lot of these including the above can't handle high voltages, the above only goes up to 72v DC so you can't rectify the 230v input of a psu.  But they're useful for example if you make a class D or class AB audio amplifier that run on 24-48v DC ... such chips help save 5-10 watts of heat which means you can use smaller heatsinks in such audio amplifiers or cheaper bulk capacitors.

 

There's also a class of power supplies that don't use bridge rectifiers to rectify the AC voltage to DC ...they use some other techniques to convert to DC and boost to high voltage... but it's hard for me to explain how they do it.

 

 

[Guest]

54 minutes ago, LIGISTX said:

Is this correct? I mean yes, that formula is correct, but isn’t the efficiency loss due to the internal switching of the power supply? Converting down to 12v is where the efficiency is lost if I remember correct…

 

Like I said, I don’t remember. But I didn’t think it was as simple as I’m America I’m drawing twice the current as someone on Europe. While that is true, is that the reason for the efficiency difference? 

Yes.  This is correct.

 

It's the CURRENT that creates the resistance.  Not the voltage.  So at the same wattage, if you're twice the voltage, you're half the current.   Remember the +12V output doesn't come until you're half way out the PSU.  You still have a bridge diode to go through at whatever the input VA is before it's boosted up to bus voltage then chopped back down to DC.

[akio123008]

18 minutes ago, mariushm said:

they use some other techniques to convert to DC

What techniques? I haven't heard of that?

[LIGISTX]

24 minutes ago, jonnyGURU said:

Yes.  This is correct.

 

It's the CURRENT that creates the resistance.  Not the voltage.  So at the same wattage, if you're twice the voltage, you're half the current.   Remember the +12V output doesn't come until you're half way out the PSU.  You still have a bridge diode to go through at whatever the input VA is before it's boosted up to bus voltage then chopped back down to DC.

Right. I was just saying it was extremely simplified to just say power = VI, which means I must be 2x in a 110 vs 220 circuit thus there is your loss. It isn’t inherent to the fact your amperage is double, that does introduce heat due to increased resistance, but it’s more about the higher voltages that helps the efficiency curve.

 

27 minutes ago, mariushm said:

The AC voltage has to be rectified into DC using a bridge rectifier (or 2/several in parallel to reduce voltage drop and dissipate heat across rectifiers) 

A bridge rectifier consists of 4 diode, 2 of them are always operating ... so you get some voltage drop across the diodes as input voltage is rectified.

 

The power dissipated in the bridge rectifier will be higher at higher currents ... for example let's say to produce 400 watts to components, the power supply needs to draw 4A of current at 110v , but only 2A of current at 230v AC.

The losses in the bridge rectifier are approximately P = 2 diodes x 1v (voltage drop on diode inside rectifier) x Current   ... so at 110v you'd have 8 watts of wasted energy, at 230v you'd have only 3-4 watts.

 

After the AC voltage is converted to DC, that DC voltage with lots of fluctuations is boosted to around 400-420v DC using the active PFC circuit. The reason this is done is because you can store more energy in that bulk capacitor this way (giving the psu enough energy to last a few milliseconds in case there's a hiccup in the power grid) and because it makes possible to use much smaller transformers  and higher frequencies (the power supply sends tens to hundreds of thousands of 400v-ish pulses through the transformer and out comes nearly 12v DC on the other side.

 

That active PFC circuit has lower losses if it only has to boost from around 300v  to 400-420v  compared to boosting from around 180v.

Rectified AC has a peak of 1.414xVac so for 230v input you'll have a peak of around 325v DC and on 110v AC you'd have a peak of around 175v DC.

 

So the two above are just two of the reasons why running a power supply with a higher input AC voltage results in much higher efficiency.

 

 

Fun fact .. There are ICs which rectify AC voltage using mosfets instead of diodes, so the losses are much lower ... like orders of magnitude lower, depending on what mosfets you choose.

For example see LT4320 : https://www.analog.com/media/en/technical-documentation/data-sheets/4320fb.pdf

 

Problem is a lot of these including the above can't handle high voltages, the above only goes up to 72v DC so you can't rectify the 230v input of a psu.  But they're useful for example if you make a class D or class AB audio amplifier that run on 24-48v DC ... such chips help save 5-10 watts of heat which means you can use smaller heatsinks in such audio amplifiers or cheaper bulk capacitors.

 

There's also a class of power supplies that don't use bridge rectifiers to rectify the AC voltage to DC ...they use some other techniques to convert to DC and boost to high voltage... but it's hard for me to explain how they do it.

 

 

Thanks for the fantastic explanation. This is what I was alluding to, but would never have been able to come up with on my own. It’s not extremely complex, but it’s not as simple as power = Voltage x amperage. 

[H713]

On 7/4/2021 at 12:43 PM, mariushm said:

Fun fact .. There are ICs which rectify AC voltage using mosfets instead of diodes, so the losses are much lower ... like orders of magnitude lower, depending on what mosfets you choose.

For example see LT4320 : https://www.analog.com/media/en/technical-documentation/data-sheets/4320fb.pdf

 

Problem is a lot of these including the above can't handle high voltages, the above only goes up to 72v DC so you can't rectify the 230v input of a psu.  But they're useful for example if you make a class D or class AB audio amplifier that run on 24-48v DC ... such chips help save 5-10 watts of heat which means you can use smaller heatsinks in such audio amplifiers or cheaper bulk capacitors.

 

Synchronous rectifiers make a lot of sense for low voltage, high current applications. Hypothetically, let's say we were looking at around 60V and 10 kA.  At 10 kA, the 1.2 V drop across a bridge rectifier would result in 12 kW of losses. In practice, it would be even higher since diodes have some parasitic resistance. Even in a pulsed application, this is unacceptable. MOSFETs with an Rds_on of 500 uohms are available these days, some with current ratings in the 500 A range. For the above mentioned 10 kA application, you'd need at least 20 in parallel (realistically more, but let's just use 20 to make the math easy). At 500 A per device, with a 500 uohm Rds_on, we're looking at 125 W per device and about 2500 W total. Obviously with more devices, and better devices, it's possible to get that considerably lower.

 

This doesn't make a lot of sense for high voltages, however, because HV FETs have a much higher Rds_on, while diodes rated for 600 - 1200 V still have a pretty reasonable voltage drop at normal levels of current. Fortunately, bridge rectifiers in these applications generally aren't a huge issue, dissipating only a few watts. That's quite manageable and generally not a problem.