Reduce noise and increase radio sensitivity with Frosty SDR

First of all, thanks to all at Outernet for their efforts to create a library in the sky- I have had a blast pushing the diy Outernet kit to its limits. In fact, this was my time using an E4000 based SDR and I wasn’t disappointed. I do a lot of radio work and I found that this dongle performed much better at receiving NOAA APT and Meteor M2 transmissions than its R820T2 counterpart. But there is room for improvement.

This dongle like most in its class has much better sensitivity when it’s cool. Once it warms up the SNR drops and signal quality degrades. When I first tried to receive Outernet’s Inmarsat-4 satellite signal I was unable to get a positive SigDet for more than a couple of seconds. I knew I was aligned properly and that the hardware was functional so something else had to be responsible for the failure. On a hunch, I decided to tape an ice pack to the dongle and the OuternetInABox software almost instantly received a frame lock and began downloading packets with a SNR of around 4 indoors.

It worked so well that I decided to make a permanent contraption utilizing a thermoelectric cooler, heat sink, and a fan. If you decide to build one for yourself, make sure you properly insulate the entire dongle and the end of the usb cord. Otherwise, you will have condensation issues. This unit gets very cold.

I’m powering Frosty SDR at 7v using a variable power supply. If I was doing radio astronomy or very sensitive radio work I would increase the voltage to 12v or somewhere in between.

Sorry, I had more pictures but the system would only let me post one. If you have any questions, feel free to ask.


How cool does your setup get? And how come your patch antenna is blue?

It’s probably around 35 - 40° F. That’s at 7v. 10 or more volts it will go below freezing and ice up without insulation. It’s not really necessary to be that cold for receiving the Outernet signal.

The stock antenna is silver gray, not blue


I really like your innovation.

@zoltan you’ve got to see this.

Nice work. Do you place it outside? I’d imagine at 35-40ºF that condensation might be an issue?

Did you use something like this?

Great pictures !

Thanks donde, Yes, that’s very similar to the model I used. I’m using a 12v, 60 watt peltier plate (TEC1-12706). I would prefer to use a 30 watt plate because 60 watt is overkill for an application like this, but this is what I had in the lab.

The insulation helps a lot with reducing condensation but I’m working on a solution to eliminate it altogether. The unit is operating indoors. I have it pointed towards a window with a clear line of sight to the satellite.

How would you characterize the SNR improvement by cooling the Tuner?

Well in this case, I went from having an SNR of less than 1 to a healthy 4. I wouldn’t be able to lock onto the satellite signal without the cooling. And I’m able to achieve this from inside the home which is a huge plus for me. I think once my CHIP arrives and I’m no longer tethered to the computer that the SNR will increase even more.

I plan to run some tests on some other frequencies I monitor over the next few weeks. I will have better info later. Right now, it’s almost like having a totally different dongle. I need to learn it all over again.

I’ve went down the same route a long time ago, you’ll run into condensation issues sooner or later. That will fry equipment.
I found that using oil cooling, and peltiers at the same time with lots of insulation works best - oil cooling is more efficient heat transfer medium than air, and the cooled objects can be placed in the freezer for pre-cooling or to test whether the whole idea works (it does).
Adding pumps to circulate oil is problematic at best, using peltier(s) on an actual well-insulated metal box containing gear, suspended in oil works better.
Electricity use and noise is way too much for residential use, daisy-chaining preamps will be cheaper on the long run.
Alternatively, use water cooling, less messy and easier than oil cooling / Peltiers.

I eliminated the condensation issue. The trick is to make sure all exposed surfaces of the dongle and the connecting LNA are covered with enough foam insulation to ensure that the cold doesn’t creep to the outer surfaces of the insulation.

Power savings is the next challenge. If you look under the usb connector in the photo below, you can see a small section of the plate I intentional left exposed to verify the peltier plate is working. That isn’t foam covering the plate- that’s frost buildup. In the future, I plan to later add an AtTiny85 based micro-controller to monitor and control the temperature. Peltier devices use a lot of energy and a thermostat will increase efficiency.

Here’s another photo showing additional insulation on the other side:

I even took the dongle apart to check for condensation intrusion. It was bone dry. I was pleasantly surprised to find a decent heat sink on the board.

There can be a significant improvement in the dongle’s performance when a heat sink alone is used. In one 12 hour test, I remove all the insulation and deactivated the cooling system. I received nearly 38000 packets with only 36 packet errors. The SNR, while low, remained stable at between 2.5 and 3.5 db. When I don’t use a heat sink, I receive about one packet error for every 20 or so packets- assuming I can even achieve a frame lock to begin with.

Today I will be adding a NooElec NESDR SMArt R820T2 based dongle and a new LNA I will be testing to Frosty. I am interested to see how the new dongle will perform with the 1.5GHz signal under cooling.

1 Like

What you are doing is very interesting. Please do not stop.

I am curious about extending the powered coaxial cable between the Outernet bias T LNA and the E4000 long enough that the e4000 can be inside a house and not getting sunlight etc to cause overheating.

Hi Seasalt. That could be done without shortening the antenna cable.You can use wifi or up to a 5 meter long usb cable. With some hacks, you might be able to extend the USB cable further. As long as the hardware is shielded from the sun it should be fine. A heat sink will keep the dongle cool enough.

I discovered that the Outernet antenna works well for FM radio. So I installed a FM trap between the antenna and the LNA. I got a 2 db boost using an old F type trap with SMA adaptors. Anybody else willing to try an FM trap and post their observations?


Please send pictures.

I had a small scare last night. When I was screwing the antenna onto the LNA, the female SMA fell off. Turns out it is a compression fitting and wasn’t broken. So today I reinstalled it and soldered both connectors to prevent this from happening again.

I now know for sure that FM and possibly airband frequencies are interfering with the much weaker Outernet signal. By using a FM trap and placing the antenna in a meshed pot I can greatly reduce this interference and tease Outernet’s signal out from the noise. But for a long term solution I will need to make a proper container to shield the antenna from RFI. I will also need to get a proper FM trap with SMA connectors to minimize signal loss. So I will post much later when I have developed a solution.