I've recently received an RTL-SDR dongle, and I'm picking up a lot of noise that is probably coming from my PC, especially a lot of "spikes" on the 2M band. I've already started using a USB extension cable instead of plugging the dongle into the PC. What else can I do to reduce the noise floor and get rid of the RFI on 2 meters?
Here is an assortment of common ideas and recommendations for RTL-SDR devices:
Mount the device directly on your antenna, or otherwise minimize the amount of feed line used. (This does not directly remove noise but rather increases signal, but that's just as good if not better. It may also remove noise by moving the device further from your computer equipment.) Of course, if your antenna is outdoors you will need a weatherproof enclosure, and you must make sure the USB line meets USB specs despite its increased length (such as by using an active extension cable designed for the purpose).
Make sure you have the RF gain set correctly!
Too little gain will bury real signals in the noise floor; turn it up until you see the noise floor rising a bit so that you're seeing the noise from the antenna.
Too much gain will overload and produce duplicated signals across the spectrum; turn it down until those extra spikes go away.
I have found that the built-in AGC may err in the direction of too much gain.
Build a shielded enclosure around the device; they come in plastic cases with no RF shielding at all.
Add snap-on ferrite beads near the ends of cables, if they didn't come with built-in ferrites.
An RTL-SDR-specific suggestion I have frequently heard is to modify a USB extension cable by removing the shield at the female-A end. (One could also modify the device's male-A plug but that might interfere with the mechanical connection.)
The argument given is that the ground and shield are tied together on the PCB, and by disconnecting the USB cable shield from the ground you prevent noise carried on the outside of the shield from entering the RF section's ground.
I have not yet tried this modification myself, and a few comments here suggest it is not worthwhile.
1$\begingroup$ If disconnecting shields indeed does make a difference, an even better solution would likely be to snap ferrite beads on the feedline and USB cable. $\endgroup$ Dec 31, 2013 at 0:13
$\begingroup$ Well, I can say that I put ferrite beads on my USB cable(s) and it made almost no difference. (For my equipment and environment, anyway.) $\endgroup$– Kevin Reid AG6YO ♦Dec 31, 2013 at 0:17
$\begingroup$ I've experimented with cables which have shield connected at both ends and which have no shield connection at device end and it made no difference in my setup. $\endgroup$ Dec 31, 2013 at 13:30
$\begingroup$ How does minimizing feed line length increase signal without equally increasing noise? $\endgroup$ Oct 7, 2014 at 13:10
$\begingroup$ @PhilFrost It increases the incoming signal+noise from the antenna, while the noise from the dongle or computer does not also increase. So there should be a small effect. $\endgroup$– Kevin Reid AG6YO ♦Oct 7, 2014 at 14:20
I've just mounted mine in an eddy stone box. I put a BNC on the end and swapped the USB plug for a socket and made a hole in the box to fit the cable from a USB extension. Works a treat, rugged, mobile and no noise. 73
$\begingroup$ Any chance of quantifying "no noise"? I'm thinking of trying to do same and and interested in finding out how much improvement there is. $\endgroup$ Dec 31, 2013 at 21:39
$\begingroup$ Sorry I haven't measured but where I am I suffer a lot of breakthrough from the medium wave and other various broadcast stations. To be fair outside of the enclosure it isn't bad but having done some basic tests in the die cast box all is well. It really is a simple mod I bought the 92x38x31mm box and it fits perfect with ample room for connectors and still small enough to be mobile. Will uploadpixs when I get a spare 5 mins. $\endgroup$– user521Dec 31, 2013 at 21:42
$\begingroup$ The box was £7 delivered so worth a punt either way. You also don't have to make a permanent mod like I did and you can still test easily to see if it's of benefit. In fact an old biscuit tin will prob do for tests hi hi $\endgroup$– user521Dec 31, 2013 at 21:46
If you want to go to the extreme have a look here: https://www.youtube.com/watch?v=-mCRLfUnXTE At 8:40 into the video you can see a way to almost completely eliminate the problem. Look at the entire video to get an idea of the magnitude of the problem.