Message	Date	From	Subject
9726	2014-02-12 11:11:32	Ephemeral	Measuring noise figure (was Investigating a Norton feedback amplifie
>Hi Adrian, >Can you describe in a group email the procedure you're using the measure noise figure... To get absolute results you need a calibrated noise source. I looked at building one around a thermionic noise diode. The diodes are readily available and it is a simple circuit but you need a 100V power supply and that ought to be in a proper box for safety and by the time I had costed the parts it was less trouble to buy one off Ebay. Then you measure the ratio of the noise power with the source switched on to that with it switched off. If you know the excess noise ratio of the source you can easily compute the noise figure. If all you want is a relative noise measurement you can dispense with the source. If your power meter does not have enough sensitivity you need a low noise preamp. The book suggests a MAR-6 followed by MAR-3s. I used two MAR-6s, with a little resistive padding to achieve a broadband match. Actually the lowest calibration point of my source is 10 MHz. But the response seems to be flat to less than that. HP noise meters are probably nice but far too expensive. If you use a thermionic noise diode the procedure can be even simpler, according to this document: http://frank.pocnet.net/sheets/126/a/A2087.pdf Adrian \|

>Hi Adrian,

>Can you describe in a group email

the procedure you're using the measure noise figure...

To get absolute results you need a calibrated noise source. I looked at building one

around a thermionic noise diode. The diodes are readily available and it is a simple

circuit but you need a 100V power supply and that ought to be in a proper box for safety

and by the time I had costed the parts it was less trouble to buy one off Ebay.

Then you measure the ratio of the noise power with the source switched on to that with

it switched off. If you know the excess noise ratio of the source you can easily compute

the noise figure.

If all you want is a relative noise measurement you can dispense with the source.

If your power meter does not have enough sensitivity you need a low noise preamp.

The book suggests a MAR-6 followed by MAR-3s. I used two MAR-6s, with a little

resistive padding to achieve a broadband match.

Actually the lowest calibration point of my source is 10 MHz. But the response seems

to be flat to less than that.

HP noise meters are probably nice but far too expensive.

If you use a thermionic noise diode the procedure can be even simpler, according to this

document:

Adrian