I was scanning the forums for something else; but came across this post so I thought I would add my two penn'orth.
After buying one of the Chinese noise sources from EBay, I was quite disappointed with the measurements I obtained with it, I couldn’t get repeatable results, and as I couldn’t be certain of its flatness nor it’s ENR, I was never sure if my problem lay in the RSP or the noise source! I was also aware of the huge amount of noise that was coming from the source and left wondering if the combined noise power of it all may damage the RSP front end. I also noticed that the MMIC’s run incredibly hot from 12V. Within a few months all of the above became irrelevant when the output MMIC died the same as the original poster here ‘glovisol’.
I then purchased a noise source here in the UK from this company
http://g8fek.com/index.html. They not only sell calibrated noise sources but also general purpose ones which are suitable for amateur use. I originally brought a noise source to do simple tests of receiver noise figures by using the application note supplied on their website here (g8fek.com/uploads/9/4/4/3/94435411/rfdesign_an_-1_intro_to_noise_sources.pdf). I used their BBGen which has a flat +-1.5db output from 100KHz to 5GHz (datasheet here
http://g8fek.com/uploads/9/4/4/3/944354 ... 090211.pdf) which produced some repeatable results using an RMS voltmeter, but I've since moved on to using the RSP (I use an RSP1 now discontinued) with the AGC off to produce the two noise on noise off results for the y factor measurement. There is a good tool and information on this method of noise figure measurement on the pages of Owen Duffy here (
https://www.owenduffy.net/software/nfm/index.htm), I use the NFM application with manual entry to do the calculations.
To sweep filters I have followed the BBGen with a wideband MMIC based homebrew amp to sweep cavities at 177MHz and also filters at 28MHz using the RSP as the detector with manual gain adjustment. The same company sells a range of wideband amplifiers BBAxx (
http://g8fek.com/lna.html) which have predicted gain and flatness, I have brought their BBA1 as the amplifier following my noise source to get a higher output level to enable me to add a return loss bridge between the output of the source and the antenna port of the RSP, that way you can look at the RL of ‘devices’ that you put on the unknown port of the bridge.
I have just noticed that they now sell an "RXGen" which has a +/-0.5db accuracy from 2MHz to 2,4GHz. Somewhat better than my original BBGen!
I have been to a few rallies and talked with the designer of these products, we even tested the RSP against a calibrated noise source at the Newark Hamfest last year, and all the numbers stacked up.
I eventually brought an RFD2305 calibrated source so that I could make meaningful measurements. For this post, I did a quick and dirty measurement using the RSP1 and the NFM app to show the gain and NF of a simple MAR8 MMIC amp in a milled box (Images at end of Post)
As you can see, the results are pretty good, the image below was taken from The MAR8 datasheet:-(Images at end of Post)
You just have to bear in mind that we are not talking about several thousand pound systems here; we are talking about low cost tools that will comfortably give you some meaningful and reproducible results within the ‘Amateur’ budget. As long as you keep the RSP device in its linear range, and make sure that you don’t overload the front end by adding too much amplification, the results that I have obtained have always stacked up with professional test kit within a db or so.
I do not work for RF Design UK, the only affiliation that I have with them is that I now know the designer well after discussing many of his products over the last few years. Over those few years have brought noise sources and broadband amplifiers from them. Not only have I been impressed by the quality of product, but the support I have received is excellent too.
Enjoy experimenting..
Martin