Last changes  26.12.2008

 

Preamplifier and Bandpass filter for SDR from  1.8 to 18 MHz

 

The SDR hardware in its simplest form consists of a wideband switched balanced mixer and low noise LF amplifier. This simple hardware has an amazing sensitivity and linearity which is sufficient for test and even regular use. Here are some of the problems in this simple hardware:

1.  The mixer is converting also the input spectrum around odd harmonics of the oscillator. For example the theoretical attenuation of the third harmonic mixing versus first harmonic is only 10 dB.

2.  The sensitivity of this simple SDR is somewhere around -124dBm to -128 dBm at 500 Hz bandwidth  (with good low noise preamplifier). But sometimes there are conditions, on higher frequencies, when the atmospheric noise is very low and the sensitivity should be above -130dBm.

3. The second order distortions are the weak point of this type of radio. To avoid these effects, an input filter  is needed for out of band strong stations and also a step attenuator.

4. The switched mixer has excellent parameters when the signal source has low and resistive internal resistance. The antenna as a signal source is not the best choice even if the antenna is tuned, not to mention the random wires. The image rejection and dynamic range are drastically reduced when the signal source has reactive component.

5. There is strong leakage signal of the heterodyne oscillator at the mixer input. This signal is heard as a strong carrier if a second receiver uses the same antenna.

 

 

Schematics

Fig 1 shows a simple preamplifier which solves to some extent the problems described above.

Fig.1

 

A parallel L1,C1 circuit with high Q-factor is tuned with variable capacitor in the range from 1.8 to 18 MHz. The signal from the low Z antenna is fed with inductive coupling  with transformation ratio 1:12. The FET follower transforms the high resonance impedance into low resistive impedance which is fed to the switch mixer. The voltage gain actually takes place by the 1:12 transformation.  The gain can be reduced  in 6 dB steps by switching the FET gate to different taps of the inductive coil.  The taps are ½ , ¼. and 1/8 of windings referred to ground. There is additional attenuator with single  resistor R1. When R1 is inserted, the Q-factor of the LC circuit increases due to reduced load from the antenna side.

 

Construction and components

The variable capacitor is a plastic type from MW receiver.  The two sections (5 - 235 pF) are connected in parallel. The inductance L1 is wound on toroidal ferrite core from HF low loss material (mu = 80). Iron powder high Q cores might be used also. The inductance must be 4.7 uH  which in our case is reached with 12 turns. The measured Q factor at 10 MHz was 200. L2 has only 1 turn on the same core. It is good for a L1 and  L2 to be wound on toroidal core since the magnetic coupling between two windings will be strong.

 I have tested a very lazy solution with a commercial 4.7 uH choke. The Q-factor at 10 MHz was 90 and L2 has 2 turns  wounded tightly above color markings of the choke. The results are quite satisfactory but the possibility of gain switching is missed.

The JFET is 2N5486 which is cheap and with low noise. Popular J309, 310 can be used also.  Fig. 2 shows the experimental prototype.

 

Fig.2

 

 

Adjustments

Nothing in this design is critical. Keep the 4.7 uH value of L1 if the entire range 1.8 -20 MHz is needed. For 1.8 MHz band additional capacitor of 1300 pF is added. The drain current is 9 mA but will depend from the type of the transistor used and the range might be 5 – 15 mA.

 

Results

This simple circuit was simulated with CAD simulator LTspice  http://www.linear.com/designtools/software/ltspice.jsp. This is a freeware simulator from Linear Technology.  For those who are interested I am giving the working file RF_preamp.asc. The results are presented on Table 1. The first column is the voltage gain. The second is the effective Q-factor which determines the selectivity and the third column is the attenuation for 3 times higher frequency.  Two cases with R1 on and short circuited are presented. The unloaded Q-factor of L1 was defined 200 on all bands.

 

Table 1

 

 

 

R atten. = 0 ohms

 

R atten. = 150 ohms

F MHz

C1 pF

Gain dB

Effective Q

Attenuation @ 3F dB

Gain dB

Effective Q

Attenuation @ 3F dB

1.8

1600

8.8

86

-49

0

138

-51

3.5

420

10

63

-46

5.5

125

-50

7

105

12.6

35

-42

8.7

95

-48

10.1

50

13

27

-39

9.7

78

-46

14

25

13

19

< -35

11

64

< -40

18

14

13.3

16

< -35

11

50

< -40

 

 

The frequency responses for 7 MHz tune frequency are presented on Fig. 3. The frequency response is not changed by the gain switching. There is some shift in resonance frequency when the taps are switched due to change in equivalent capacitance of the JFET input involved in the circuit. This can be corrected manually with the variable capacitor.

 

Ôèã. 3

            Real measurements were performed only for 14 MHz band. The measured gain was 11 dB (R1 = 0).  At this gain the measured sensitivity was -135 dBm (@ 500Hz bandwidth). The measured sensitivity of the SDR only with mixer was -126 dBm.

An experiment was performed to measure the level of the heterodyne oscillator leakage at the antenna input when two receivers work on the same antenna. The antenna input of the SDR was connected through 3 dB attenuator to antenna input of Icom 756PRO transceiver. Additional load of 50 ohms was connected there to imitate the antenna impedance. The transceiver was set to its maximal sensitivity with Preamplifier 2 on. The results are shown on Table 2.   The S-meter levels are not calibrated and the results are very approximate. 20 to 30 dB additional attenuation must be expected when there is a preamplifier in front of the mixer.  Generally speaking bellow 7 MHz the leakage signal is buried in the band noise.

 

Table 2

 

      Leakage S level

 

Freq. MHz

SDR Preamp on

SDR Mixer only

14

S9+12

S9+30

7

S7

S9+18

3.5

<S1

S9+5

 

Comments

The preamplifier gain is the native gain of this circuit.  The reduction of the gain for low frequency bands does not matter since there the gain is not needed at all – only the selective function is important.

Is the filter selectivity sufficient?  The most dangerous bands are 1.8 and 3.5 MHz. On these bands the third harmonics mixing is in the range where strong levels of probable undesired signals exist.  A simple calculation shows:

Filter attenuation  -50 dB; Mixer attenuation   -10 dB; Antenna mismatch attenuation - ??? but probably > 10 dB. The total attenuation is in order of -70 dB.

For example an ordinary 2nd order 50 ohms Chebishev band-pass filter with 2 capacitors and inductances will have 3d harmonic attenuation in order of 55 dB.

The dynamic range was not measured but probably is high. The drain current is high and the follower has a strong negative feedback.  I could not notice any adverse effects when using this preamplifier with SDR as a second RX in CQ WW DX contest.

Random wire antennas can be used with this preamplifier since the mixer is fed always with low impedance. Very short wires must be connected directly to the parallel LC circuit through coupling capacitor as is shown on Fig. 1

This preamplifier, attenuator &filter solves some of the problems described in the beginning of the article which is not bad for such a simple circuit. 

 

 

September 2008

73 Chavdar . LZ1AQ