Attenuator Nonlinearity

Hi,

We have a full duplex transceiver link whose power output is around 2W. For link tests when we connect a 2W fixed attenuator to transceiver output, it raises noise floor on receiver band very much so that receive sensitivity decreases. Is such a thing possible, is this a nonlinear effect? Is it problematic to use attenuator at its power rating?

Thanks,

A transceiver is a combination of a transmitter and a receiver connected to one antenna via a switch. You have connected AN UNSPECIFIED attenuator to the antenna port, so when you activate the receiver, the signal from antenna is attenuated by the attenuator by 'unknown count of dB'. No wonder that the sensitivity decreased.

The sensitivity have to remain unchanged. Please describe exactly how did you measure the sensitivity, the expected sensitivity and the attenuator value as weel as the bench connections. Did you use also variable attenuators ? Furthermore how did you measure the noise floor with and without attenuator ?

Two transceivers are connected to each other with certain attenuators. Transceiver has different transmit & receive frequencies. Transmit & receive parts are connected with a diplexer. Tests are done in this configuration.
Actually when attenuator is connected to 2W power amplifier, i see that noise floor in receive band increases dramatically. I checked that on spectrum analyzer.
What I am curious about is that :
1. Can this be a specific problem that is related to the attenuator I use? or
2. Any attenuator that is drived around its power rating can cause such problem?

Thanks,

The noise generated by an attenuator is exactly the same of that generated by a resistor having the same value. I suppose you are working with 50 ohm impedances the a 50 ohm attenuator, regardless to its attenuation value will generate the same noise of a 50 ohm resistor. This is why the noise figure of an attenuator is equal to the attenuation value.

You are measuring, instead, the noise generated by your TX amplifier.

The noise floor density is -174 dBm/Hz. If you connect a 50 ohm resistor as well as an attenuator terminated onto 50 ohm to a spectrum analyzer you will measure that value (normally is not possible to do this measurment due to the noise figure of the instrument itself). If we have an amplifier with 20 dB gain and
10 dB noise figure the spectrum will measure -174+20+10=-144 dBm/Hz. After a 10 dB attenuator we will measure roughly -154 dBm/Hz (the additional noise generated by the attenuator itself is negligible).

You didn't mention anything about the sensitivity.

Thanks, i know the bakground theory. When measurement is taken after PA, floor was -85 dbm/Hz. Then I placed a filter that filters receive band, nothing changed. Floor was still -85 dBm/Hz. Thats why I suspect attenuator.

OK, but the frequency band of the transmitter of "equipment1" must be the same of the receiver of the "equipment2" so a filter will have no effect in that band (neglecting the insertion loss). However you can check the attenuator by means of a network analyzer, or better you can connect a directional coupler before and after the attenuator to check forward and reflected power. If the attenuator is working properly, n both case the reflected power have to be very low (supposing the TX and RX are well matched).

Thanks,

I am going to apply some more tests. I hope I can solve the problem.