Power Curves Formula

[lpax95]

Hello ,

I have taken upon my self to create a formula to explain the power curves of 80+ bronze or better power supplies, and I need help. I don't have the resources to test more than 3 power supplies so I am reaching out to this forum. If anyone has a power curve/power efficiency data and wants to share I would greatly appreciate it. All I need is your make, model of the power supply, and the data you collected. I am hoping that with enough data I can create a formula that will allow for some basic guesses when it comes to quantifying the curves. With all that being said if anyone has already found or created this formula feel free to post it. 

Thanks

-Luke

[Enderman]

Why can't you just look at the specs on the manufacturer website...

[Spotty]

Every PSU will have different efficiency and power curves. You can't really make a formula that will calculate it.

All you can do is look at the 80+ specifications and go off that, however that's only a minimum requirement of efficiency.
https://en.wikipedia.org/wiki/80_Plus

[lpax95]

Looking at the manufacturer shows only one. I have started collecting some of the more popular sku's but its not enough data to make a good formula.A community like this gives the chance to get bigger data set

 

[lpax95]

Just now, Spotty said:

Every PSU will have different efficiency and power curves. You can't really make a formula that will calculate it.

All you can do is look at the 80+ specifications and go off that, however that's only a minimum requirement of efficiency.
https://en.wikipedia.org/wiki/80_Plus

I know they are all different but with data, I can plot the points and extrapolate a curve that fits the data. It won't be perfect but it will give hopefully a rough estimate.

[Electronics Wizardy]

3 minutes ago, lpax95 said:

I know they are all different but with data, I can plot the points and extrapolate a curve that fits the data. It won't be perfect but it will give hopefully a rough estimate.

cause a lousy regression line will help someone?

 

And then were getting into measurement error. How do we all know all of our measuring equment is calabrated? And the error margin

[Enderman]

8 minutes ago, lpax95 said:

Looking at the manufacturer shows only one. I have started collecting some of the more popular sku's but its not enough data to make a good formula.A community like this gives the chance to get bigger data set

 

Uh, no, most high end manufacturers give a curve for each model of PSU, because every PSU is different.

Having an average of a bunch of PSUs is useless and says nothing about the PSU you actually have.

 

Also it's not like most people care if their PSU is 87% or 86% efficient at X load.

The people that do care will know how to use google to find the power curve for THEIR power supply.

[mariushm]

Techpowerup has the habbit of posting efficiency curves with their reviews, here's an example (scroll all the way to the bottom) :  https://www.techpowerup.com/reviews/NZXT/NP-1PM-E650A/3.html

The review database is here: https://www.techpowerup.com/reviews/?category=Power+Supplies&manufacturer=&pp=25&order=date

 

Cybenetics also has those efficiency curves (and loads of other useful information) for all the power they review - they're aiming to be a better replacement for the organization that tests and gives those 80+ Bronze/Silver/Gold/Platinum/Titanium marks, so they're way more thorough and do a lot more tests. 

 

Here's their review database : https://www.cybenetics.com/index.php?option=power-supplies

There's a PDF file for every model reviewed, and the efficiency curves are there.

 

[Spotty]

3 minutes ago, lpax95 said:

It won't be perfect but it will give hopefully a rough estimate.

If all you care about is a rough estimate of a PSUs efficiency, just look at the 80+ rating of the PSU then. If anyone cares about the efficiency of a unit beyond the 80+ rating certification, they should look at a review and efficiency test of that specific unit.
 

12 minutes ago, lpax95 said:

If anyone has a power curve/power efficiency data and wants to share I would greatly appreciate it. All I need is your make, model of the power supply, and the data you collected.

And where would these people get the data from? What's your methodology for collecting the data? Testing equipment? Are you measuring efficiency at 115V or 230V AC?...

[Rexper]

There's a bunch of data at Cybenetics, free to the public.

Perhaps you could reach out to them and ask if they could supply you with an easy to manage data format.

 

I've tried what you're doing. But you need data for different loads (10% increments and low loads) for multiple samples of different wattages and efficiency ratings. It's a pain to manually enter them all.

 

For example, I found the average formula for three random bronze 450w PSUs. Y=30.11 + 20.87x + 1.96x², with a Log₂ on X. 

But a much larger sample size would be wanted for reliable results.

 

It's better to keep your data from one reliable source, to eliminate deviations caused by environment or methodology.

 

Then you can find the relationship between the efficiency curve formula and wattage/efficiency rating/input voltage/temperature / whatever else may affect it.

 

This could be beneficial for an accurate power consumption calculator.

[seon123]

There's Cybenetics, which has numbers for low loads (≤100W), 10-110% in 10% intervals, and CL. You can also find the 80+ reports of the the tested PSUs, but you only get data points for 10% (only on newer PSUs), 20%, 50% and 100%. 

Though I don't know why you would want an approximate curve, instead of just finding testing of the specific PSU. And the efficiency is generally one of the least important aspects of a PSU.