Noise measurement and units conversion

I’m trying to replace an obsolete Lock-In amplifier with a newer model. My understanding is that a lock-in amp is essentially a single frequency spectrum analyzer. I’m having trouble getting readings from the new lock-in amp and I’d like to verify that the signal to it is good by connecting the signal to my SA44 to make the measurement. The existing lock-in amp shows a signal reading of a chopped (600Hz) signal around 1.7mV and a background noise on the order of 3.5nV/SQRT(Hz). After reading the SA44 Users Manual about noise markers I see that Spike can return a noise measurement using the following formula ????/??=10∗???10 Σ(?????????∗???)?2?1+???
Can I use this measurement to convert to nV/SQRT(Hz) and if so how do I recalculate it?

Hi Chuck,

Assuming 50 ohm, and the Hz in dBm/Hz and nV/sqrt(Hz) is 1Hz, you can simply perform the conversion from dBm to nV.

nV/sqrt(Hz) = 10 ^ ((dBm/Hz + 166.9897) / 20.0);

If you have access to Matlab/Octave, an anonymous function can make quick conversions

>> myconversion = @(dbm) 10^((dbm+166.9897) / 20.0);
then later

>> myconversion(-160)
ans = 2.2361 

Regards,
Andrew

Andrew,
Thank you for your response. I’m sure that this could work assuming, as you say 50 ohms and 1Hz. However, the lock-in measurement is at an unknown impedance and it is a 600Hz chopped signal. On a spectrum analyzer I think I’ll expect to see a signal spike at 600Hz and noise below the skirts where the whole trace would have a 50% duty cycle in time domain. The UUT is a biased PbS flame sensor with a blackbox radiator producing the stimulating signal. There is a fan between the BB radiator and the UUT that alternately passes and block the radiated signal at a rate of 600Hz. I was hoping that I could make the measurement in volts. Of course, this would require a high input impedance at the SA input. I think that if I could, then I could set the noise markers at some distance in the noise, take the square root of the difference in frequencies and divide the voltage measurement by the square root of the difference in frequencies to get my nV/SQRT(Hz) result. I don’t know if this is even reasonable, correct or within the capabilities of the SA44. What do you think?

Chuck,
If you are unsure of the output impedance, one solution is to add a low noise buffer (op amp), with a 50 ohm series output resistor and large DC blocking capacitor (passing 600 Hz would require about 22 uF to minimize impact on amplitude accuracy). You could also add some gain here depending on the expected signal level.

If this is not an option, you could try measuring the output impedance, or alternately see how much the amplitude changes when you add a e.g. 100 ohm series resistor to your output, and then use standard resistor divider equations to solve for output impedance (assuming a 50 ohm input impedance).

[Author]

Posts