rel: 351; Rel Rival - IF=1620kHz
I stumbled across this schematic and noticed the specified IF frequency is a very surprising 1620kHz.
My guess is that a decimal point is missing, and the IF frequency is probably 162kHz. However, I would like to know for sure.
This radio covers two bands:
1: 20-50m = 15 - 6MHz
2: 200-2000m = 1.5MHz - 150kHz
The corresponding Local Oscillator frequency ranges for a 1620kHz IF would be:
1: High side injection=16620kHz-7620kHz or low side injection=13380kHz-4380kHz
2: High side injection=3120kHz-1770kHz. Low side injection does not seem reasonable at 1470kHz-120kHz
Apparently, IFs around 1600 kHz were sometimes used in european practice from the 30s and 40s. It should also be noted that several communications receivers where it was not necessary to have complete coverage of the middle wave broadcast band, used IFs around 1200 kHz.
For instance, the following discussion can be found in the swedish book "Radioteknisk Handbok" of Eric Andersén, from 1946 (now this is a rather crude translation from swedish):
For the choice of IF, the following possibilities are available:
IF < 150 kHz
IF = 420 - 500 kHz
IF > 1500 kHz
If the IF is chosen around 110-130 kHz, this results in a high gain and good selectivity. To improve the image rejection it is however necessary to connect a bandpass filter in front of the mixer valve. It is also possible to use a separate HF amplifier stage, which improves the preselection and reduces the valve (mixing) noise.
If the IF is located to the internationally free frequency of 468 kHz, the signal and local oscillator frequencies come so far apart that the risk of image disturbances is eliminated. However, a trap, tuned to the receiver's IF has in this case to be installed in the antenna circuit.
At an IF of 1600 kHz, it is possible to use an untuned (aperiodic) input circuit and the receiver is tuned only with the local oscillator capacitor. This results in a, from a circuitry point of view, very simple miniature super, consisting of a mixer valve together with a combined detector- and LF amplifier valve (i suppose the author was thinking of a valve set like ECH11 + ECL11). When the detector is equipped with reaction to the IF transformer, this will also be useful as a bandwidth regulator.
With a high IF, the receiver can be tuned from LW to MW without band switching. Additionally, only a low pass filter in the antenna input is necessary, and a single variable condenser is sufficient for tuning.
- Einbereich in Europe (4 KB)
Thank you all for expanding, yet again, my appreciation of European radio design - this is what keeps me comming back to RM.
I had never seen a low end radio design with a high IF frequency. This is merely a reflection of the narrow scope my American radio experience.
The only time I remember seeing a high IF was in this special beginner's radio design for the Hollywood Radio and Television Institute - HRTI correspondence course.
The European circumstance of commercially active and contiguous LW and MW bands was a very good reason to use high IF for single-band, or "Einbereich", coverage in an inexpensive radio design.
The high IF may have been a good choice for a low cost design, but the sophistication of this design is evident. The large beautiful dial in this radio offered the convenience of continuous tuning, and the automatic ( mechanically linked to the local oscillator tuning capacitor) front end filter insured consistent performance across the very wide LW+MW band.
The high number of low valued capacitors around the IF transformers also suggests careful IF design for high performance.
I wonder what the differential variable capacitor at the cathode of the AF7 IF tube is for. When the capacitor turned to one end, there is just a coil at the kathode, which should reduce gain. When the capacitor at the other end, a series resonant LC grounds the cathode at the resonant frequency and should sharpen selectivity.
Was this differential capacitor under user control, for bandwidth adjustment?
From an original Schaub alignment instruction for models S229W, G/W, also with SW, a remark says:
... Connect test oscillator with 1620KHz (post war!) to grid of mixertube with 0.1MΩ resistor to ground. The value of C23 [this is the 4 - 18pF trimmer connected to the AK2] and reaction [this is the said differential capacitor and the coils 417 & 418] shall be raised until short before oscillations will occur. ...
The left shaft is marked with "Rueckkopplung" (reaction).
Due to the high value of the IF the amplification of the IF stages is low and the bandwidth is broader than the channel spacing of 9KHz. Therefore a reaction in the IF stage is appropriate in order to get the desired values.
REL sets made by Electra, Brno, Czechoslovakia really were using such high IF frequency. It is correct.
I forgot to include other examples of such high IF frequency supers from REL
Thank you Viktor and Prof Rudolph for the extra examples of high IF use. These examples add an engineering appreciation to the concept.
The use of reaction at the cathode of the IF stage in the Schawb S229W peaked my curiosity; in particular, how much reaction could be obtained from a tuned series LC circuit at the cathode. The inductor seen here just below the cathode is a high impedance choke to pass the DC bias current, and can be ignored for this post.
I looked up Frederick Terman's analysis of grid-to-plate Miller capacitance in Triode RF amplifiers. This analysis is for a common cathode Triode amplifier stage with a tuned LC circuit at the plate. The important result of his analysis is that the input conductance at the grid is negative below resonance, as shown in this figure that I extracted from Terman's 1955 Electronic and Radio Engineering page 427. This negative grid conductance is what causes instability when driven by high impedance (=low positive conductance) of the previous stage.
I ran the following simple comparative AC simulation in LTSpice freeware from www.Linear.com using a U309 JFET instead of the triode. The schematic shows the JFET wired in common source with parallel LC load at the drain, and two versions of cathode loading, one with a series LC and another with a parallel LC load. The version with the parallel LC load at the drain corresponds directly to Terman's triode circuit with a parallel LC load at the plate.
The simulation conducts a frequency sweep from 1MHz to 2MHz with 1Vrms applied at the gates of all three stages.
I chose 100uH and 100pF because they resonate near the 1.6MHz IF frequency.
Click to enlarge.
The text below the schematic shows the basic operating point of the fets with 10mS transconductance, and 3.5pF and 1.2pF of Cgs and Cds respectively. These capacitances correspond to Cpg and Ccg in a triode, and are of a comparable scale. The pentode in the Shaub can be compared to the JFET circuit in the center with the series resonant LC source load.
The bottom waveform plot on the right shows the voltage (solid) and phase (dash) transfer functions from the gates to the Drain and Sources of the three circutis. A peaking response is visble in the Red trace for the drain load of the common cathode stage, and a notch function for the series resonant cathode load in the black trace. The blue trace appears flat because the parallel resonant impedance load at the cathode is too high to have a visible peaking effect on the 100 Ohm source impedance.
The central plot on the right shows the magnitude and phase of the input gate currents for the three stages. Note again that the current is highest for the common cathode circuit with parallel LC load at the drain.The dashed lines also show the phase reversal above and below resonance that indicates the presence of negative conductance below resonance.
The top plot on the right corresponds to the input conductance (=1/resistance) curve in Terman's figure 12-16 above, but with a important style differences: Negative values of conductance are represented with a dashed sign curve. The portions of the Red and Black trances to the left of resonance are to be interpreted as negative values. The Blue curve is negative above resonance. While Terman seems to use a linear conductance scale, I used a log conductance scale.
(I did not produce a capacitance plot as seen at the top of Terman's figuer 12-16. I could have done this by plotting the inverse of the Imaginary component of the gate currents, just I got the grid conductance by plotting the real component of the grid currents)
The important result from the comparison of the three circuits is that a parallel LC load at the drain produces a negative conductance below resonance at the gate, just as a series LC load does, at the cathode. The other important difference is the magnitude of the negative conductance. The drain load case shows -1mS of input gate (grid in a triode) conductance just below resonance, while the series resonant load at the source only produces a -10uS conductance at the input gate. This is 100x weaker negative conductance.
The drain load case will be stable for input resistances driving the gate less than 1kOhm, while the series LC source load case will be stable for input resistances up to 100kOHm.
As we know from the 1920's triode era, LC plate loaded RF amplifiers were hard to stabilize, and the strong negative input conductance of this example illustrates the reason.
The series resonant load at the cathode is much easier to stabilize. Not every resonant circuit driving the gate/grid of this stage will have a high enough Q for an impedance higher than 100kOhms, and thus cause oscillatioins. But oscillations are possible for Higher Q circuits. Recently I measured some 455kHz IF transformer coils to have a 200kOhm internal resistance at resonance. If I drove a stage with -100kOHm=-10uS of input gate or grid resistance/conductance with this transformer, I would have oscillations.
Now I can see that the series resonant load at the cathode of the IF amplifier could be used to adjust reaction at that stage. It is also necessary that the impedance of the IF transformer driving the control grid of this stage be high enough for an adequate amount of reaction.
Also note that the highest gain of the stage occurs when the cathode load is a it's lowest impedance, and this occurs at resonance for a series circuit.
Thank you Prof Rudolph for providing the stimulus for this post. Thank you Hans for introducing me to the concept of "einbereich" and thank you Torbjörn and Viktor for your prespectives and radio examples.