RTL SDR Upconverter for Shortwaves (0-30MHz)

This is a simple yet completely functional upconverter for the popular SDR donlge. The construction was build around cheap and easily obtainable mixer IC marked as NE 612, manufactured by Phillips Semi. Despite its limitations, it performs outstandingly well, at least when price-to-quality factor is discussed. This IC is used to upconvert 0-30MHz to a region acceptable by SDR dongle: 100-130MHz, all of that was achieved by employing a 100MHz 4-pin crystal generator.

Here’s the electrical schematic of the apparatus itself ;]


Modus Operandi

Since I am using an active antenna (MiniWhip) I needed to provide a power supply to the input port. This is accomplished by parts: J3 C15, C16 and L5 that acts as a RF choke. Received signal is passed through an elliptic (Cauer) filter that rolls off at around 27MHz providing full coverage to all bands of my interest. R1 is to terminate filter which was designed to be a 50 Ohm in both input and output ports.

We are dealing with a broadband input (frequencies are spread across multiple octaves), so it would be very difficult to achieve a good match to NE612 input impedance without using a transformer or an active device, and since I wanted to stick with standard components (no DIY coils, etc.) I decided to go with R1 despite it being a very “sub optimal” solution. Your approach to this problem may vary.

Q1 generates 100MHz periodic signal which is far from being sinusoidal, so I took the effort of calculating a simple series resonant circuit: L4 and C13. Since NE612 requires the LO to stay in 200mVpp-300mVpp range I’ve inserted a simple resistive ‘pi’ attenuator that gives about 15dB of attenuation. R2 is a parallel combination of 75 Ohms and desired load impedance: 50 Ohms (pin 6 has much higher impedance than required 50Ohms).

Output matching (here implemented by C8-L3) was calculated to perform impedance transformation form 1.5kOhm to 50Ohm @ around 112MHz, and gives a nice broadband match.

L4-C14 is optional, and servers the sole purpose of LO feed-through suppression which can be problematic (LO can easily saturate RTL dongle). One needs to experiment with C14’s capacitance, as the circuit is very touchy (please mind the component tolerances), due to the high Q (this is in fact a narrow notch filter that tries to kill the LO and leave everything else). I found that even PCB stray capacitance plays a major role here. Fine tuning can be done by observing the magnitude of LO tone on any SDR software as one increases/decreases C14’s value.

Room for improvement

Some steps that might be undertaken in order to the improve overall performance:

  • Use a transformer to transform input signal form 50Ohms to NE612 input impedance (~1.5kOhm)
  • Use another transformer or LC-CL balun at the output port
  • Applying both points from above will result in full double-balanced operation (this is what NE612 is intended for) and will lead to significant LO feedthrough reduction, so no more crappy and hard to tune L4-C14.
  • Don’t be a cheap bastard like me and use a goddamn LDO for crystal generator supply voltage!
  • Use a 125MHz crystal generator: this will help to avoid FM broadcast stations getting in the way, remember to redesign LO and output LC circuits.
  • Use a better mixer: look for something with a decent IP3


.. and a picture that documents my mad SMD skills :]