Power Supply Noise calulations

[themaccabee]

Hi,

I'm trying to understand the specifications of a power supply..

It is said that the output will be having noise levels < -110dBm/Hz in the frequency range 1MHz-100MHz


I was trying to find out what will be the maximum output ripple from that power supply. I did the following calculations, can somebody tell me if i ve made any mistake??

-110dBm =9.999999999999987e-15 W
Considering reference =50Ohms
P=V2/R
V=sqrt(P*R)
=sqrt [(9.999999999999987e-15 W)*50]
=7.0710678118654706478143659084881e-7
=707.1067nV

Since i began with per Hz noise im assuming what i got is
nV/√Hz

ie noise will be 707.1067nV/√Hz

Since this is noise energy in 1Hz

i took the the bandwidth square root √99MHz =9949.8743710661995473447982100121

Multiplying this with 707.1067nV/√Hz *9949.8743710661995473447982100121√Hz =0.0070356236397351397586930793424713
=7.035mV RMS noise voltage in the bandwidth

Am i right?
Thanks for any help
Regards

 

[FvM]

You can save a lot of superfluous typing by omitting insignificant digits from numbers.

In engineering and science it's actually expected that shown digits represent the precision of empirical measurement results...

In so far, < 110 dBm would be appropriately written as < 7 (or < 7.0) mV rms, following your 50 ohm impedance assumption. Because the < 110 dBm number must be understood as a limit line for a mixture of discrete frequencies, a lower wide band voltage can be expected.

 

[themaccabee]

Thank you Fvm for pointing out the unwanted usage of numbers..actually i copy pasted them from my pc calculator...right away..

Ok Let me put my question in another way..


One more question .Is n't the regulator power supply noise can be considered like the white noise??& wont the measured noise power vary with the bandwidth used to measure it??

I ve seen a lot of regulator power supplies specifying say 50mVpp ripple on their output..Measured with 20MHz bandwidth. in normal oscilloscope..Assuming a 50Ohm impedance in measurement, i want to convert this specification to dBm/Hz format..

So what is the correct way to do..
Is like the following

It is mentioned as 50mVpp in 20MHz BW

So is it like {[(50mV/ √20MHz)]² / 50 }= 2.5e-12W/Hz

So finally converting to dBm =10log(2.5e-12/1e-3)=10log(2.5e-9)= -86.05dBm/Hz



Thanks for any help

 

[FvM]

The calculation is correct and 50 ohm assumption is at least reasonable. Strictly spoken, a test configuration should be given with the specification. It may be implicitely specified by referring to a standard?

The other question is what you want to do with the 7 mV number?

 

[themaccabee]

Sorry i edited my previous post can you have a look at the same

Thanks,

Thank you Fvm for pointing out the unwanted usage of numbers..actually i copy pasted them from my pc calculator...right away..

Ok Let me put my question in another way..


One more question .Is n't the regulator power supply noise can be considered like the white noise??& wont the measured noise power vary with the bandwidth used to measure it??

I ve seen a lot of regulator power supplies specifying say 50mVpp ripple on their output..Measured with 20MHz bandwidth. in normal oscilloscope..Assuming a 50Ohm impedance in measurement, i want to convert this specification to dBm/Hz format..

So what is the correct way to do..
Is like the following

It is mentioned as 50mVpp in 20MHz BW

So is it like {[(50mV/ √20MHz)]² / 50 }= 2.5e-12W/Hz

So finally converting to dBm =10log(2.5e-12/1e-3)=10log(2.5e-9)= -86.05dBm/Hz



Thanks for any help

 

[FvM]

By reversing the calculation course, the numbers get a completely different meaning. Saying you have up to 50 mVpp in 20 MHz BW means, the spectral density will be at least xx dBm/Hz. If it's a discrete line, spectral density numbers become meaningless.

Usually EMC regulations specify the relevant measurement methods. I'm not aware of dBm/Hz quantities used in this regard.