Noise & shielding, SDR Dongle

Question

First, I'm very new to SDR. This may turn out to be really obvious. I just want to verify my thoughts before I spend more dollars.

I have a USB dongle with the RT28xx in it. Works great. Side by side I have a 35 year old Azden 2 meter receiver. What I'm noticing is that the Azden seems to be able to pull in somewhat weak signals while the SDR doesn't even register a signal in the waterfall view. This is true even when I hook the SDR dongle up to the same antenna that I've got on the Azden.

To my mind there are two possible issues. First, the noise in the dongle is so high that the signal is way below the noise floor. Second, the dongle simply does't have the sensitivity to pull out the signal.

I'm leaning toward option 1 (noise in the dongle) and I see some 30 to 50 dollar shielding arrangements for the dongles. From my description of what I'm seeing, would any experts agree that 1 is the more likely option?

Answer 1

“Sensitivity” isn't all that well-defined. Insofar as it is, you might as well just think in terms of noise, because there is always a noise floor and it's always what determines the weakest signal you can receive.

The most direct way to deal with noise is to make the signal stronger than the noise. This means improving your antenna:

• making it more efficient
  • a full-sized antenna containing no lossy elements
  • tuned specifically for the band you wish to receive
  • away from obstructions, metallic or otherwise
• moving it to a better location
  • away from local noise sources
  • away from obstructions lying between it and the transmitter
• making it more directional, pointed towards the transmitter

This helps with almost everything. If you've done that and it isn't enough, let's start considering the categories of noise:

• Noise that is from outside sources and picked up by your antenna, because it's propagating just like the wanted signal.

  • You can't do anything about this other than use a directional antenna to avoid picking up noise that's not in the direction of the transmitter. And that means you have to pick a frequency and build the antenna for it, and aim it.

• Noise that is generated by your own electronic equipment and picked up by the antenna.

  • You can reduce this by moving your antenna (and receiver) farther away from it.

• Noise that is generated by your own equipment and radiates through the unshielded case.

  • As you note, this could be fixed by a shielded case. But it's often not the worst problem or the easiest to solve.

• Noise that is conducted from your computer to the rtl-sdr device along the USB cable.

  • Shielding can't help with this.
  • Try a different type of computer — brands vary widely.
  • Disconnect/shut off all other peripherals — in particular cases I have seen power adapters, Ethernet ports, Thunderbolt ports, and built-in speakers! be major noise sources that can be avoided. (Internal radios such as Wi-Fi and Bluetooth are usually comparatively clean, because they're actually designed to be.)

• Phase noise, from the internal oscillator's imperfection. This can be improved by hardware modifications.

• Internal analog noise from the RTL chip etc., and quantization noise, from the limited resolution of the ADC. These are not really very similar, but they are both addressed by

  • Setting the hardware gain controls properly, so that almost the full dynamic range of the ADC is used. (Do not use the built-in AGC (automatic gain control) — in my experience it always sets the level too high.) Turn AGC off, tune to a known signal, and adjust gain until the height difference between the signal peak and the noise floor is largest — this corresponds to maximum SNR.
  • For quantization noise only, increase the sample rate (as long as your computer can handle it; 2.4 MHz is the usual recommendation for RTL-SDR). This reduces quantization noise via oversampling.

Given that you are comparing two receivers with the same antenna and getting unsatisfactory performance from one, my recommendations are to do the following, as detailed above:

• Make sure the gain control is set optimally.
• Increase the sample rate to 2.4 MHz if it isn't already.
• Shut off/disconnect all unnecessary computer peripherals; try a different computer.

Answer 2

While the other answer is quite detailed, this one is much shorter and specific: I purchased and installed an aluminum RF shield for the dongle. The noise floor is significantly lower as a result! I was able to drop it even further by adding a ferrite bead to the USB cable. (I was already using a 10' extension to get it away from the computer.)

Update:

Watching the waterfall display while adding a ferrite bead to the USB cable and to the feed line where it immediately enters the dongle also reduced the noise visibly. Note that this was not a huge visible difference, but seeing a visible intensity change, even minor, in the background noise can really make a difference when looking for weak signals (which is what I do).

Answer 3

I can tell you what I did, and it worked pretty well. Since I didn't have copper wrap or a suitable little steel box, I wrapped my SDR in about 20-30 layers of alu foil. I soldered a copper filament from USB GND to this makeshift shield, wrapped a few turns of copper solid core wire to hold it in place, and then electrical tape. I should have put a mini heatsink on the tiny chips that get pretty hot, and drilled holes for passive airflow. Next I took two of those little antennas that come with the dongles, clipped the wire from one of them, and connected GND to the antenna bit of the clipped one. I just superglued it together, and covered the bases in electrical tape. Nice dipole, much better reception. I should use shielded antenna cable or add ferrite beads, but I'm not that bothered though. The little antennas are worth keeping, since they're easy to solder solid core copper wire onto. Just clip them, and add your desired 1/4, 1/2 or 1/1 wavelengths - lambda = 300/145 = 200cm for VHF, and lambda = 300/433 = 69.3cm for UHF. I would probably need at least 1/2 wavelength to make it worthwhile. I use a USB repeater cable to move my dongle far away from my noisy laptop PSU and other things. I believe it's 5M. Not much loss in that. I disable AGC in my SDR software and set it manually, since AGC usually sets gain too high. You should run rtl_test -p for about ten minutes to get your PPM value. It does vary, apparently due to heat. If I were to do it again, I would heatsink the SDR dongle and drill ventilation holes in it. Hope this is useful.