The Cascode 7 Experimental Receiver
Introduction
This simple receiver was one byproduct of 4 months of experimentation with the various stages of a basic superheterodyne. The design goals for this basic receiver were 9 volt battery power, low current drain and to use single-ended, discretely built mixers. "Bench sweepings class" transistors (2N3904 and MPF102) were also factored to make this a relatively inexpensive and fun project.
The experimental receiver design presented on this page was chosen over others because of its' relatively modest sensitivity and low noise. Quiescent current drain is 28.4 mA at 9.0 volts DC power. Even with the battery degraded to 7.92 volts, this receiver has good sensitivity and draws ~26 mA.
Some outcast circuits and photographs can be found at Cascode 7 Supplemental Page
Cascode It Is
For mixing and product detecting in modern low-medium performance homebrew superhets, the NE602/SA612 doubly-balanced mixer is extremely popular.
In the 1960s up unto ~1990, the dual gate MOSFET reigned supreme for use as not only a mixer/product detector , but as an RF/IF amp in low current, simple homebrew receivers. In the 1980's -90s, attention was focused on improving the intercept point of receivers resulting in the popularization of the diode ring and Gilbert cell type mixers. While the dual gate MOSFET and cascode JFET stages, have their problems and limitations, I think they should not be overlooked as viable stages in certain homebrew receivers.
"Contrary to some opinions, dual gate devices are still around, still do a good job and really simplify a homebrew radio design" Mac Chapman, KI6BP, HBR Fifteen, Communications Quarterly for Spring 1999.
This part has become harder to obtain and more costly to purchase. In EMRFD, Wes Hayward discusses this issue and presents some text and interesting schematics using a practical alternative; a cascode N-channel JFET common source and common gate stage which may be easily configured as an RF amp, mixer or product detector in a method similar to the dual gate MOSFET. The cascode topologies share the high gain and lower noise of their dual gate cousins. The cascode JFET amp has been in the ARRL handbook since the 1970's, but it took EMRFD to get me to try this stage!
While no panacea, you can build a pretty decent receiver using cascode JFETs. The total cost all of all of the semiconductors in this radio was $4.90, which was cheaper than the DPDT switch used for the attenuator or a single SA602 for that matter..
To the right: A partial component and schematic of and nJFET cascode mixer stage using the humble MPF102. LO input is not shown.
Schematic 1: Attenuator, RF preamp and Mixer
50 ohms impedance RF enters via a rear panel mounted BNC jack and DPDT switch. This switch allows the operator to engage a -10 dB, 50 ohm attenuator pad. The pi-attenuator pad helps reduce interfering signals by reducing sensitivity and improving the mixer intermodulation performance. When listening to CW, I rarely adjust the IF gain control and prefer front end attenuators to adjust for signal strength on any homebrew receiver I use.
The input and output tanks on the RF preamp serve as a band pass filter. The Q of the inductors is relatively high as they are very lightly loaded. I could not sweep the RF preamp as my RF generator does not work between 6.8 and 7.5 MHz. The RF preamp has ~15 db gain and is not source bypassed to reduce gain by providing degeneration. This robust amp reduces mixer IMD performance, however is quiet and I preferred it to the common gate stage (shown on the supplemental page) for enhanced sensitivity on weak signal reception. The attenuator is available for preserving mixer IMD performance with strong signal work.
This strong RF preamp goes against proper design theory of a receiver front end, however, this is a popcorn 9 volt receiver which certainly does not fall in the high performance category. Omitting it or substituting a common gate preamp may be prudent for some builders. As shown, the attenuator is required for strong signal work or signal distortion may occur. Two attenuators or a variable type would be useful.
All of the front end and IF system tanks use the same inductor; 28 turns # 28 AWG wire on a T37-2 powdered iron torroid. All tanks tune very sharply. The tuned stages may be aligned by listening for maximum noise or signal strength on a CW signal or with a scope and signal generator.
My RF preamp coils were peaked with a crystal oscillator set to 7.040 MHz. There should be enough variable capacitance to peak it as low as 7.020 MHz or more without any difficulty. This will allow builders to set their receiver to monitor the bottom 28KHz of the 40 meter band. If not, more capacitance (5-10 pF), or trimmer caps with a larger maximum capacity might be tried. I used the Digi-Key SG30015-ND trimmer cap throughout this receiver as I have dozens of them in my junk box.
Many single-ended mixers using both BJTs and JFETS were tried. Some of the BJT versions gave tremendous conversion gain, however, strong signal distortion, noise and VFO harmonic energy leakage plagued these designs. In one BJT mixer I tried, the second harmonic of the VFO was 3.1 volts peak to peak at the mixer output! Ultimately, a cascode JFET mixer was chosen as it provides high impedance ports for the LO and RF and has moderately good strong signal handling capability. I measured the 5 MHz VFO leakage on the mixer output at about 40 dB down when compared to the LO input voltage at the gate of Q2. This is mostly due to the T2 tuned circuit which tunes very sharply.
VHF oscillations observed were nulled by using 33 ohm resistors at the drain of Q2 and the drain and at the gate of Q4. Wes Hayward, W7ZOI, once told me that values from 10 to 100 ohms may be used to to suppress parasitic oscillations in FET RF amps. Ferrite beads would also serve well and allow shorter lead lengths. The small (0.001 uF) bypass cap value at the Q2 gate also serves to suppress such oscillations. The 470 ohm source resistor for the mixer was chosen as it gave lower distortion when compared to lower resistor values which allowed more current to the mixer..
To the right: A photograph of the entire receiver during experiments with the mixer stages. This version has a double tuned filter (no RF preamp) ahead of a BJT mixer. Although, this receiver looks quite messy from the many experiments I performed, it works reasonably well. The IF and BFO/detector board from this version were also completely replaced.
Schematic 2: VFO
The heart of this receiver is this stable, BJT only, varactor tuned, variable frequency oscillator. Spare time over 4 weeks was spent experimenting with BJT oscillators and much was learned. This design is one of the most frequency stable VFOs I have ever built. On the version used for this receiver, my 1 hour frequency drift was +/-3 Hertz after warm up with the VFO set at about mid frequency. Markus Hansen, VE7CA, built a version with an air variable tuning capacitor instead of the varactor. This VFO drifted +/-6 Hertz over 1 hour. I built 2 other versions and the worst-case 1 hour drift was +/-14 Hertz, although experimentation was necessary to achieve this stability.
My friend and QRP colleague David White, WN5Y had some stability trouble with his varicap version and turned it into a standard LC VFO. " Just to let you know that I removed the varicap and the VFO was very stable. With the swamp cooler blowing right over it, no shielding and the temperature dropping 10 to 15 degrees, the frequency has only moved 60 hertz; quite incredible."
My experiments led to the following observations and speculations. My lack of engineering knowledge and limited test equipment means some of these conclusions are not scientifically valid, however, the goal of this site is build measure, change and learn.
1. The decoupling cap is part of the tank circuit (The 0.01 polystyrene temperature compensates)
2. The base bias influenced frequency stability.
3. The lore about JFETs being more frequency stable than BJTs is untrue.
4. The C to L ratio should be relatively high to enhance frequency stability, but this may increase the amount of
phase noise.
5. The use of air variable typetrimmer caps increases frequency stabilization.
6. Build a working version first. Temperature compensate after.
7. Low current increases the oscillator stability and reduces noise.
8. The BB104 dual
varactor in my VFO is more frequency stable at tuning voltages of <= 2.46 volts
DC. This is quite startling for me. I found from 0.40 to 2.46 volts applied DC,
the oscillator was extremely stable ( 3 Hz drift in 1 hour, uncovered).
Increasing the maximal tuning voltage above 2.46 degraded the frequency stability. The
solution was a voltage divider on the DC voltage control circuit. The 68K and
50K pot resistors set the maximum voltage at 2.40 v. Now it swings from 2.40 to
0.40 volts with the pot turned from one side to the other.
On another test version, I had to keep the maximum control voltage less than 2.20 volts to keep the VFO very stable at it's maximum frequency. Experimentation is the key.I believe that in the VFO shown, the Q of the BB104 was higher at control voltages less than 2.46 volts and this enhanced frequency stability. Output voltage is 3.4 volts peak to peak. The 4K7 resistor reduces oscillations at HF.
The 1N4005 diode is for temperature compensation of the varactor control voltage. Wire the tuning pot so that when turning it from left to right, the frequency of the VFO decreases. The frequency swing of this VFO is 28 KHz. This suits a simple linear taper tuning pot well as I have found that when VFO's have a large (30 or more KHz) frequency swing and a narrow IF filter band pass, fine tuning with a simple linear taper pot is difficult. 10 turn pots are better in these circumstances, however are costly for such a trivial receiver. My VFO is set to tune from 4.979 to 4.951 MHz. With an IF of 11.996 MHz, this receiver covers from 7.017 to 7.045 MHz.VR1 is an inexpensive, low current, TO-92 cased, 5 volt voltage regulator which I found provided excellent decoupling and voltage stability for this VFO. I recommend this over just using a simple zener diode voltage regulator.
To the upper right is a photograph of the VFO under development. The red control voltage wire is single strand 18 gauge wire to keep it still. I used a 1/2 watt 68K resistor to also prevent movement of the control voltage wire. In the final version, the red wire and 100K resistor lead length (seen over top of the trimmer cap) were shortened. The VFO output tank is fixed tuned, however, a variable trimmer cap may be substituted for the 15 pF cap used. As shown, the inductor had a 9 turn link for use with lower impedance LO ports of mixers tried during my experiments with different types of mixers.
Shown in the photographs below is a version of the VFO built by Earl, 4Z4TJ in Beer-Shavea, Israel. Earl's version uses a BB204 varactor. He connected the VFO along with a 16.05 MHz xtal LO to a dual gate MOSFET mixer to get an output signal in the 15 meter band. Earl used a Bourne 10 turn pot to get a 2 KHz per turn (or 5.5Hz / degree) tuning resolution and [most important] a genuine Collins knob. He reported good stability and that the frequency change/turn is linear over the whole range of 20 KHz. I really enjoyed learning about his perspective of the homebrew radio scene in Israel. Earl did a fine job of bread boarding the VFO as evident in the 2 photographs. Many thanks to 4Z4TJ for the feedback
Schematic 3: IF System
Shown above is the IF amps and crystal filter schematic. Stage gain is ~25 dB. The target input and output impedance of the crystal filter is intended to be 225 ohms. The center frequency is 11.996 MHz. My 6dB down frequencies during a sweep were 11.9939 MHz and 11.9988 MHz, making the 6 dB bandwidth 490 Hertz. It was difficult to sweep this narrow filter as the resolution of my RF generator is very coarse and only 12 plots were made. A limited graph of dB versus frequency from 11.99 to 12.10 MHz of my filter sweep is shown below. As you can see, the filter skirt response is not very steep, as only 3 crystals are used and the transformer links terminate the filter rather crudely. There is no ringing with this filter and filter selectivity is better when you are tuned on the low side of an RF carrier where the skirt is steeper.
Shown below is a photograph of the entire IF system, product detector, BFO and high gain audio preamp. In my receiver, I omitted the trimmer cap of the tank at the input of the second IF amp (Q3/Q4). I wound the T37-2 torroid with 29 turns and used a fixed 33 pF cap to resonate the tank. The peak to peak voltage of the Q3/Q4 amp was 0.40 volts lower than the peaked version built using the parts specified in the schematic. The copper wrap around island is for the B+. The crystal cases are connected together with a grounded piece of copper wire.
Schematic 4: BFO and Product Detector
Shown above is the schematic for the 12 MHz BFO and cascode JFET product detector. The output voltage of the BFO into the Q2 gate is 4.2 volts peak to peak. An audio transformer may be used on Q2 instead of the 1K resistor for more gain, however this adds cost and bulk to the project. A BJT product detector tried had 17 dB conversion gain and required only the final audio amplifier in Schematic 5 to deliver strong audio to the headphones. This product detector required an additional BFO buffer stage and strong RF amp to drive it however, and total current draw was 13 mA. It could not handle strong signals without distorting them and so was discarded. A low pass filter follows the product detector shown above.
"Device under test" is shown in the photograph above. This was a BFO/ product detector built with BJTs. The BFO waveform on the oscilloscope is shown in the inset photo. My Heathkit RF signal generator was recently fitted with a large tuning knob which increased fine tuning capability of the output frequency. Big knobs also improved the fine tuning of the VFO of some homebrew receivers I tried them on. I purchased three, 3 inch knobs at a flea market for $2.00.
Schematic 5: Audio System
Shown above. The audio and on/off control stages of this receiver. The Q1/Q2 high gain AF preamp is right out of Solid State Design for the Radio Amateur. The PA amp uses a Radio Shack 1K:8 ohm audio transformer and is out of EMRFD. I tried different bias resistors, but the resistors chosen by W7ZOI give a good bias voltage and clean signal amplification. With the antenna disconnected, there is almost no detectable audio in the headphones.
I was surprised when testing 4 sets of different headphones with this receiver. Two were identical in output, while one was very loud and the final set very weak. The LED is a super-bright, clear device that glows red with applied voltage. The 6K8 resistor dropped the LED current draw to 1.1 mA and it is still is bright enough to see well.
Chassis and Construction
The chassis was one of 5 various types I purchased for $5.00 U.S. each from the Boeing Aircraft surplus store in Seattle, WA . It used to be a digital switch box. I left a couple of the rear panel mounted RS-232 connectors in place out of laziness. The VFO is only shielded front and back, although normally I encase them in copper PC board or another smaller chassis. Small vertical copper boards scraps separate the various circuit boards from one another and make the 9 volt battery compartment. The SPST front panel switch is factory mounted on the 10K audio volume pot. The red B+ wires are all single strand 18 gauge. The black wires connecting boards and the 10K pot is RG-174 grounded at 1 end. I have not had any problems with AM radio detection at my QTH.
Experiment Conclusion
I encourage you to try your hand at experimenting with various superhet stages to learn more about this popular receiver configuration. The knowledge and satisfaction gained will certainly enhance your passion about building your own equipment. Avoid miniaturizing circuits when experimenting as this often will limit your ability to change a stage without great frustration and damaging components with part manipulation and soldering. You feedback is welcome. 73, de VE7BPO.
