In her latest video Sarah from the SignalsEverywhere YouTube channel describes how it is possible to use an RTL-SDR to measure RF filters when combined with a noise source and the Spektrum SDR software. In the video Sarah comprehensively explains how to set all the various parameters in Spektrum, before demonstrating a filter measurement with a noise source. Sarah explains how the power measurements may not be entirely accurate, however it is enough to get some idea about the shape of a filter.
UPDATE: John wanted to add the following clarification:
In the article you wrote, you suggest that the two noise generators are equivalent. This is not the case. The noise generator you mentioned generates "Thermal Noise". Mine generates "Shot Noise".
The Zener in the generator you mentioned, contributes zero noise in the Gigahertz spectrum. Whereas mine generates Shot Noise roughly 20 db higher than the thermal noise floor all across the range.
The physics and the mathematics are as different as night and day.
Thank you to John Jackson from Jackson Research for writing in and sharing his Kickstarter for a laboratory grade RF broadband noise generator. John notes that he's discovered a semiconductor that is much cheaper than the expensive avalanche diodes used in other high end noise sources. This semiconductor has allowed him to create a noise source that works from 40 MHz up to 6 GHz. The target output power is 0 dB, with a flatness of less than 2 dBm over the entire bandwidth.
A noise source is useful for SDR/radio experiments as it is used in many applications such as scalar network analyzers, filter response measurements, satellite tuning, cellphone network tuning, antenna VSWR measurements, amplifier noise measurements and in jammers. If you're interested we have a previously posted tutorial about using an RTL-SDR and noise source for filter characterization and antenna VSWR measurements.
John is fundraising via Kickstarter in order to help fund development of the final product. The pricing is currently US$500. This might seem high, but John notes that comparable professional noise sources start at US$3,500 and go up to US$30,000. John writes:
There are a number of laboratory grade RF Noise generators on the market. They are all extremely expensive ranging in the thousands of dollars. This is beyond the reach of most individuals. In contrast, there are some low cost RF noise generator schematics floating around the Internet which are all based upon Zener diodes.
The problem with Zener diodes is the noise bandwidth. All the circuits I have seen have cutoff frequencies in the Megahertz range. The one noise circuit I saw used in the Gigahertz range was actually amplifier noise and had problems due to the several stages of amplification.
The alternative to Zener diodes is the avalanche diode specifically designed as a RF noise generator. These devices are difficult to acquire and have very high prices which often exceed the entire cost of all the parts and components needed to build the instrument.
Professor Jackson at Jackson Research has discovered a semiconductor that generates RF noise like an avalanche noise diode into the Gigahertz range, but at Zener diode prices. He has raw laboratory data and now needs to build up a professional grade instrument. To achieve this goal, he has launched a new KickStarter fund raiser to build up a number of units.
Over on YouTube RTL-SDR experimenter Adam Alicajic has uploaded a video showing how it is possible to use the RTL-SDR as a tool to measure the frequency response of an RF filter. To do this he uses a noise source circuit which produces wide band white noise connected to an LNA4ALL, connected to the RF filter and finally connected to the RTL-SDR. Then using the Touchstone spectrum analyzer software he does a 300 MHz bandwidth sweep over a section of the spectrum which shows the response of the filter.
The noise source can be built from a simple diode based circuit as shown in a previous post, or if you have the Ham-it-up upconverter you can buy the parts for the noise generator part of the circuit.
In his video he shows the frequency response of a 145 MHz helix filter, a coax notch filter and a 1090 MHz home brew bandpass filter.
Lacking both an expensive signal generator and spectrum analyzer, Hans used his RTL-SDR together with a simple diode based wideband noise generator to measure the frequency response of a 137 MHz bandpass filter.
To do this he used the RTL-SDR Scanner software which allows you to create a composite spectrum over a frequency range wider than the maximum 3.2 MHz of bandwidth the RTL-SDR provides. The wideband noise generator was cleverly constructed out of a diode operating in it’s reverse breakdown mode.
Apart from the obvious excessive spurs, this method worked quite well and the shape of the filter is clearly visible.