BG7TBL noise source
Spectrum analyers are rather expensive devices, especially the one's with a tracking generator. Therefore old spectrum analysers without a tracking generator are most budget friendly. Without a tracking generator it isn't possible to align filters of desired. A tracking generator generates a steady amplitude signal at the frequency where the receiver end is "looking". Therefore in time each frequency is generated and measured and shown on the display. Another way to show behaviour of a certain filter is by using a wideband noise source. This noise source generates wideband noise, ideally at a steady signal level. This can be used as a cheap substitude of a tracking generator.
BG7TBL designed a rather cheap wideband noise source, or should I say "NOSE SOURCE"? (See picture.) This product can be bought easily on eBay. Altough I have a SA with tracking generator, I bought one (version 2016-03-06) due to curiousity.
The hardware design is quite nice. DC power input and four header pins as alternative power input and a gold plated SMA connector for signal output. There's a protective diode in the power in line and there's a power led placed on the board. The (surface mount) soldering and the silk screening of the board looks very nice. There are a lot of trough hole commections to the reference plane at the bottom. The quality of the product is visually great.
output signal strength
Altough there's a "pi" attenuator network on board, the output level is rather high. Don't feed the signal directly into your SA! Since the signal is rather high and broadband, the local mixer of your SA will be likely overloaded. It's safe to assume the noise source generates 0 dBm at some peaks. Use a 20 dB attentuator in combination with a Rigol DSA-815.
The heart of the design is a zender diode. This diode generates wideband noise. This noise is amplified in three stages and at the end attenuated a bit, likely for 50 Ohms impedance matching...
It seems that there are several versions, at least three. De oldest seems to be the 2013-12-18 version with two electrolytic capacitors. The 2014-08-20 design is even more complex and has six electrolytic capacitors. The 2016-03-06 is fully surface mounted and doesn't have electrolytic capacitors. This is a good thing since electrolytic capacitors will age and surface mounted caps perform much better!
It seems there's a (major?!) heat problem. After a short amount of time, the three amplifier semiconductor packages will become very hot. I'm not an expert in this area, but I guess the heat generation is suprisingly large. Assembly of a heatsink on top of the amplifier parts would be wise if operation for several minutes is desired.
The no(i)se source ;-) is rather cheap and does generate wideband noise. It would be possible to adjust filters using this noise source and a spectrum analyser. The build quality is very nice. The signal output isn't very linear, but for this price and purpose this shouldn't be a problem. If you have a rather old spectrum analyser without a tracking generator, this is a nice pice of equipment to have altough it never will be as good as an tracking generator.