More about Noise
By Gerd, WB8IFMB
At the last meeting, we had a lively discussion about “noise”. This was because of my comments and arguments for choosing the low noise preamps for the “white boxes” in the February newsletter. Therefore, let me again try to make this a little clearer.
Dbs are not a good way to characterize the noise being generated in receivers. That is because db is just a number or, as somebody at the meeting pointed out, a ratio. Dbs can be added and subtracted helping in figuring out the gain or loss in an amplifier chain. Applying this general rule to a receiver with a 1 db noise figure and compare it to a receiver with a 4 db nose figure we would expect for the latter to produce twice as much noise. However, the real difference between those two receivers is 7.7 db or 6 times as much noise not two times as the three dB would indicate.
In the 1930s noise figure was associated to our room temperature, which was, at the time, a choice as good as any. But as frequencies were pushed up the VHF, UHF, and microwave ladder, room temperature became more and more irrelevant. There is, however, still some use for the noise figure. For cables and other attenuation (relays, circulators etc) between the antenna and the receiver input, those items being at room temperature, the noise figure is exactly equal their attenuation in db.
I suggested to use noise temperature instead of noise figure. If you do this, all calculations become real simple. You see, the noise generated by a device is directly proportional to the noise temperature. In exact equations you find the term kTB for noise power, where k is the Boltzman constant, and B is the bandwidth. The bandwidth B in our calculation is almost always the same, so we only need to consider the noise temperatures of the different components in order to compare our systems.
The equations to convert noise figure (NF) into noise temperature (T) is: T= 290 (10 .1NF -1)
And to convert noise temperature into noise figure: NF=10 lg ( 1/290 T +1)
Noise factor, by the way, is the number which the noise figure represents: e.g. 0dB = 1, 3dB = 2 etc.
Let’s go through an example to quickly calculate a system noise temperature:
Frequency 400MHz, sky temp: 350, LNA .4db, cable loss including connectors 1db, antenna relay .2 db Converting all the dbs into temperatures we obtain: 350, 280, 750, and 140. Immediately, the high temperature for the cable stands out and confirms my old saying, any cable is a bad cable! Try to do without. Adding up the individual temperatures we obtain a combined (system noise temperature) of 1520. Let’s say you eliminate the cable (mounting the preamp at the ant). Now we have 770, which is half the noise or a 3 db improvement. *)
One variable is important, the sky noise temperature, also sometimes referred to as antenna temperature. This expression is sort of accurate, but totally depends on what the antenna is looking at. At 400 MHz the sky temperature varies from a low of 150 to a high of 2800. For 136 MHz these temperatures are 2000 and 36000. How does this affect your antenna, which is usually pointed at the horizon? Well, the ground reflects the sky into the antenna (which explains the importance of a nice flat ground in front of your antenna) and since the earth rotates, different parts of the sky determine the noise temperature. EME programs usually provide sky noise temperatures around the moon, and satellite tracking programs give you the sky noise in the neighborhood of the satellite being tracked. Some programs can be manipulated to give you the sky noise in the azimuth you are interested in.
At higher frequencies the noise temperature gets much lower (between 1 and 10 GHz we talk about the “microwave window”), which explains that much less power is needed for communication. One last comment; although the sun is not a radio intensive star, but rather emits mostly light, it nevertheless radiates enough power across the radio spectrum to cause signal blackouts. Sun noise is often used to check and confirm the performance of a low noise microwave set-up.
*) Assuming the additional noise contribution by the then following rx is negligible. An example: rx NF 6db=86000 LNA gain 23 db=200. Additional noise at input of LNA 8600:200 = 4.30
|
Last updated March 27, 2003 |