As soon as I post any video I always see something that I wish I had said. I would like to add the little fact that -174 dBm/Hz is the ambient noise only at 290 Kelvin. In the vicinity of room temperature it goes up (becomes less negative) by about 0.07 dB for every 5 degree Celsius increase in temperature. Also, at 10:40 I mention too briefly the need to account for cable loss in the gain measurement. So Let me re-emphasize it here. Careful that your gain does not include cable loss.
@@j.w.8663 "does not". Make sure that the number that you conclude for the amplifier gain has accounted correctly for the insertion loss of the cables. For example, if the amplifier has 30 dB of gain but a cable was included with the amplifier (sometimes necessary) that has 1 dB of loss, you don't want to conclude that the gain is 29 dB, even though that's what the spectrum analyzer is telling you.
I have a variable gain setup with an attenuator followed by an amp. For our "low gain" (0dB) setting, there is a -22 attenuator feeding the amp. In that case, I cannot simply terminate with a 50 ohm terminator, because then the amplifier wouldn't "see" the -22 attenuator. Therefore is it a fact that I can't use the "gain method" for this setup?
I will be using pyramidal horn antennas with an LNA and a spectrum analyzer to scan in a 360 degree azimuth to look for possible interference in the X and S bands (2GHz and 8GHz). One of the requirements is to "Calculate cold sky noise temperature at 10 degree elevation." Would you (or anyone) perhaps know how one does this? Is this simply measurable somehow? Google doesn't get me too far...😕
@@j.w.8663 I wish I could be more helpful. Your question did stir up a memory for for me, though. When I was in grad school I had some microwave measurement questions, and I contacted someone at NRAO (National Radio Astronomy Observatory) who was very helpful. I even went to visit since I was close. It impressed me how eager these guys were to be a resource for a student. I'm going to guess that there is well-established know-how for your specific question that resides squarely in the radio astronomy community. I don't work in that field, but it does have a wealth of knowledge about microwaves, the sky, noise, etc.
Use spec an often at work to verify Noise Figure or estimate noise power at DUT output. Calibration is critical. Always expect some error when testing via this method
You are quite correct. It is really hard to eliminate systematic error from a technique that does not rely on a calibrated reference. The quality issues of this method were anticipated at 11:25.
@@stephenremillard1 I do not recall a mention of noise floor (dBm/Hz) of this spectrum. Our spec an typically has noise floor of @-140dBm/Hz. For a good measurement I typically try and get the measured noise 16dB higher than spec an noise floor
@jason conerty Good point and I like your rigor. I no longer have access to the lab to get the noise floor without the DUT, however the CXA at 40 MHz using log averaging, with neither input attenuation nor preamp (the conditions of this video), has a DANL (displayed average noise level) of -150 dBm/Hz (typ). Without having given it much thought, I have always used a less conservative 10 dB higher rule, but 16 dB sounds like a good idea, and probably well thought out. Could you say something here about the importance of a larger noise difference?
@@stephenremillard1 regarding measuring noise figure in this type of setup with a DUT that has significantly lower noise figure? It's difficult to do this with a spectrum analyzer. Your best bet is to have two DUT available and allow one to settle and understand the noise power displayed on the spectrum. For instance if you have a a DUT that has 20dB of gain and 3dB noise figure you would expect the noise power to be at -174 plus the gain and noise figure of the DUT. Now this is specifically only the noise power of the spectrum with additional noise power presented specifically by the set-up. Adding the one Specific DUT you wish to acquire data for you would assume that total noise power is the sum of gain for both DUT plus the noise figure of the first DUT. It's messy and requires understanding of temperature and gain for both devices. It's a lot to take into consideration. Also if both devices are identical then order can be switched. This is usually something I only do for proof of concept. If I'm nervous about test equipment ENR values etc
![BLACK BAG - Official Trailer [HD] - Only in Theaters March 14](http://i.ytimg.com/vi/Du0Xp8WX_7I/mqdefault.jpg)
As soon as I post any video I always see something that I wish I had said. I would like to add the little fact that -174 dBm/Hz is the ambient noise only at 290 Kelvin. In the vicinity of room temperature it goes up (becomes less negative) by about 0.07 dB for every 5 degree Celsius increase in temperature.
Also, at 10:40 I mention too briefly the need to account for cable loss in the gain measurement. So Let me re-emphasize it here. Careful that your gain does not include cable loss.
Re: "Careful that your gain does not include cable loss", -Do you mean 'does' or really, 'does not'?
@@j.w.8663 "does not". Make sure that the number that you conclude for the amplifier gain has accounted correctly for the insertion loss of the cables. For example, if the amplifier has 30 dB of gain but a cable was included with the amplifier (sometimes necessary) that has 1 dB of loss, you don't want to conclude that the gain is 29 dB, even though that's what the spectrum analyzer is telling you.
Ah, I see. The gain only of the amp. Thank you for the reply!
Thank you so much. You are a fantastic teacher
I have a variable gain setup with an attenuator followed by an amp. For our "low gain" (0dB) setting, there is a -22 attenuator feeding the amp. In that case, I cannot simply terminate with a 50 ohm terminator, because then the amplifier wouldn't "see" the -22 attenuator. Therefore is it a fact that I can't use the "gain method" for this setup?
The center frequency is 20MHz while measuring the noise power of around - 137 dBm. It should have been 40 MHz right?
I will be using pyramidal horn antennas with an LNA and a spectrum analyzer to scan in a 360 degree azimuth to look for possible interference in the X and S bands (2GHz and 8GHz). One of the requirements is to "Calculate cold sky noise temperature at 10 degree elevation." Would you (or anyone) perhaps know how one does this? Is this simply measurable somehow? Google doesn't get me too far...😕
I'm in the same boat as Google. Not really something I know how to do.
@@stephenremillard1 ok, thank you.
@@j.w.8663 I wish I could be more helpful. Your question did stir up a memory for for me, though. When I was in grad school I had some microwave measurement questions, and I contacted someone at NRAO (National Radio Astronomy Observatory) who was very helpful. I even went to visit since I was close. It impressed me how eager these guys were to be a resource for a student. I'm going to guess that there is well-established know-how for your specific question that resides squarely in the radio astronomy community. I don't work in that field, but it does have a wealth of knowledge about microwaves, the sky, noise, etc.
Use spec an often at work to verify Noise Figure or estimate noise power at DUT output. Calibration is critical. Always expect some error when testing via this method
You are quite correct. It is really hard to eliminate systematic error from a technique that does not rely on a calibrated reference. The quality issues of this method were anticipated at 11:25.
@@stephenremillard1 I do not recall a mention of noise floor (dBm/Hz) of this spectrum. Our spec an typically has noise floor of @-140dBm/Hz. For a good measurement I typically try and get the measured noise 16dB higher than spec an noise floor
@jason conerty Good point and I like your rigor. I no longer have access to the lab to get the noise floor without the DUT, however the CXA at 40 MHz using log averaging, with neither input attenuation nor preamp (the conditions of this video), has a DANL (displayed average noise level) of -150 dBm/Hz (typ). Without having given it much thought, I have always used a less conservative 10 dB higher rule, but 16 dB sounds like a good idea, and probably well thought out. Could you say something here about the importance of a larger noise difference?
@@stephenremillard1 regarding measuring noise figure in this type of setup with a DUT that has significantly lower noise figure? It's difficult to do this with a spectrum analyzer. Your best bet is to have two DUT available and allow one to settle and understand the noise power displayed on the spectrum. For instance if you have a a DUT that has 20dB of gain and 3dB noise figure you would expect the noise power to be at -174 plus the gain and noise figure of the DUT. Now this is specifically only the noise power of the spectrum with additional noise power presented specifically by the set-up. Adding the one Specific DUT you wish to acquire data for you would assume that total noise power is the sum of gain for both DUT plus the noise figure of the first DUT. It's messy and requires understanding of temperature and gain for both devices. It's a lot to take into consideration. Also if both devices are identical then order can be switched. This is usually something I only do for proof of concept. If I'm nervous about test equipment ENR values etc
Great video, thanks.
That was terrific
🌹🌹🌹