EEVblog Electronics Community Forum

Products => Test Equipment => Topic started by: 2nOrderEDO on February 26, 2022, 11:50:24 am

Title: HP3562A Vertical dB units
Post by: 2nOrderEDO on February 26, 2022, 11:50:24 am
Hello folks,

INTRODUCTION:
We are working on characterizing the noise floor of our LNA (output Vrms), and my friend has a nice HP3562A which we are using for this task. Since the thing is heavy as a boat anchor, usually my friend makes the measurement and sends me the data in .csv so that I can integrate the area under the PSD curve to get our RMS voltage noisefloor. For this I use python, numpy and matplotlib to generate the graphs (example below):

(https://www.eevblog.com/forum/testgear/hp3562a-vertical-db-units/?action=dlattach;attach=1425256)

The topic is, the HP3562A has several possible units for the logarithmic vertical axis (all units below will be in dB version):


(https://www.eevblog.com/forum/testgear/hp3562a-vertical-db-units/?action=dlattach;attach=1425244)

Up until know, I was receiving the data in dBV2, which yielded me much higher Vrms results when compared to Vrms noise I was observing in the time domain with my RTB2004. My first fail was converting the data in dBV2 to volts with 20 * log instead of 10 * log (We are talking POWER density Spectrum here, so it is a ratio of powers, not amplitudes). But even after amending this error, we couldn't match our measurement with the HP3562A to the time domain measurement and our calculated values (RTB2004 agrees wiht calculated noise floor).

The second fail was to set the instrument to use dBV2 instead of dBV2/Hz. Since the HP3562A is quite an old instrument, we thought dBV2 was shown instead of dBV2/Hz due to a display limitation. My friend, which is who operates the instrument, soon discovered the other units above mentioned in the manual. Then he sent me the same curves represented in dBV2, and dBV/√Hz. That is when I realiced that dBV2 and dBV2/Hz are not the same: after converting to linear units and squaring the data in dBV/√Hz, I was not getting the same result as the data in dBV2 units. That is when I went in to the HP3562A and found out about this different units.

WHAT I LEARNED:

I understand what dBV/√Hz and dBV2/Hz are, and that the former requires 20 * log and the later 10 * log to convert to the linear form.

QUESTION:

What are the other units? What are they used for? I specially wonder about dBV2 and dBV2s/Hz...

Thanks in advance for your insights!

Kind regards,
Enrique
Title: Re: HP3562A Vertical dB units
Post by: gf on February 26, 2022, 08:21:26 pm
V is voltage, and V² is proportional to power (since P =V²/R, where V is the RMS voltage (not peak)).
Power (or squared voltage) of different frequency bands adds up. Voltages do not add up.
If you want to integrate over a frequency range in order to obtain the total power in this range, then you need to integrate the power density (V²/Hz).
I also wonder what V²s/Hz is supposed to mean. From the units it seems to be an integral of power density over time. But over which time interval?