The first consumer radios were called TRFs, meaning Tuned Radio Frequency.
1924 Atwater Kent #Model 4700
These early TRFs usually had three tuned circuits, each followed by a stage of gain. The last stage of gain doubled as the detector or demodulator stage. Two stages of pure audio amplification usually followed the detector. This was the basic formula for thousands of radios built during the 1920s.
The Superheterodyne mixes the output of a tuned RF stage with the output of an oscillator to create a fixed intermediate frequency, or IF, which is then amplified before demodulation.
1925 Silver Marshall Superheterodyne
The TRFs always tuned a little broader than the Supers. If a station was broadcasting high-fidelity audio, the TRF allowed more of it to get through to the speaker. If two or more stations were too close together, the Super had the selectivity to separate them. Variable IF bandwidth allowed the Superheterodyne to offer the best of both designs. In the broad position the full spectrum of audio was reproduced; while in the narrow position, maximum selectivity was achieved.
The best AM radios of the 1930s offered variable selectivity. In 1938 and 1939, most RCAs used a broadly tuned IF. It's always been a challenge to properly align one of these sets. If you "peak" the IFs of a late 1930s RCA, it will sound worse than it did when it came in to the shop. The only way to properly align one of these sets is with an oscilloscope and some type of frequency-modulated oscillator.
Clough-Brengle Model OMA Wobbulator
A frequency-modulated oscillator is usually called a sweep generator today, but in 1935, it was called a wobbulator.
The first wobbulators used a motor-driven variable capacitor to sweep (or wobble) through the desired frequency range. Later circuits used a Phantastron
oscillator that worked well, but still had no good way to add markers
A marker is a visual indicator on the oscilloscope trace that tells us where we are. We might know that we're injecting a 30 kHz wide signal into the IF stage, but without the markers, we don't know how much of it gets through.
Hickok Model 691 Marker Adder
Illustrations always show a sharp little marker "pip," but unless you've got a really expensive "commercial" marker generator, it just doesn't work out. We've tried everything from wrapping a turn of wire around the mixer tube to using a rebuilt Hickok 691 Marker Adder. Everything we tried was just too broad and swamped out the display.
Allen Lein's "XR-2206 Sweeper
If you've been attending the February Radio Workshops (sponsored by the Northland Antique Radio Club
) here at the Museum, you saw that Allen Lein has finally solved the problem. It wasn't easy; his latest design uses 19 integrated circuits plus all the associated resistors, capacitors and wires. But wow is it ever functional! Powered by eight "AA" cells, the whole package measures only 4 ½" by 5 ¾" by 2 ¾" and has a built-in frequency counter. It's continuously variable from 36 kHz to 825 kHz, so you can even sweep those 45 kc
IF coils on your 1925 Super!
Connect the output of Allen's "Sweeper" to the grid of your radio's mixer tube, the "Y" input of your scope to the diode load of the radio's detector, and the Horizontal Output of his device to the "X" input of your scope and you've got the most beautiful, 8 Hz, linear sweep you've ever seen.
Early wobbulator designs used a 60 Hz sweep derived from line voltage. Allen's choice of integrated circuits dictates the slower 8 Hz rate. This causes a problem on some radios as their AGC levels rise and fall with the slow sweep rate. To remedy this, he's put a variable bias supply into the device to clamp the AGC circuit of the radio under test.
To see more of Al's custom-built test equipment go to museumofbroadcasting.org/Lein.html.
But the real magic is the marker. It's only 1 ms (millisecond) wide and shows up like a textbook display on the oscilloscope. The old way of adding a marker into a sweep display was to inject a fixed frequency on top of the sweep frequency. Allen's solution is to generate a 10 ms gate pulse in the middle of the sweep. The pulse opens up a gate that allows a sample of the center frequency to go to the frequency counter on the "Sweeper." A 1-ms marker pulse is generated in the center of the 10-ms gate pulse. The 1-ms marker pulse goes to the "Z" input on the back of your scope and shows up as a perfect marker pip in the center of the display. Adjusting the tuning control on the front of the Sweeper allows you to move the marker and determine the actual response width of your radio. Best of all, Allen gave us schematics and wiring diagrams so you can build your own! For circuit information see xr2206.html
Something this wonderful begs to be used. Allen started on this part of the project even before he built the Sweeper. Using parts from a wrecked Philco 37-116 he built an AM tuner that outperforms my high-end, 1957 Fisher AM 80.
Fisher AM 80 "wide"
One can determine the bandwidth of a tuned circuit by moving the marker to the left and right of center until it has moved down either side of the response curve to a point that corresponds to 0.707 of the peak. This represents the point where intermediate frequency response has diminished by 3 decibels. Since the transition from uniform response to roll-off in a tuned circuit is never a sharply defined corner, engineers use the 0.707 or 3-dB-down point as the frequency at which flat response ends and roll-off begins. For example, if the IF curve’s 3-dB-down points fall 4 kHz either side of the IF center frequency, the audio response will be about 3 dB down at 4 kHz. Without showing the markers, these pictures are a little deceiving. It turns out that the Fisher's bandwidth is 3.6 kHz wide in the sharp position (compared to 3.1 in the Philco). This results in good audio response up to 1800 kHz, but poorer selectivity than the Philco.
Note that the Fisher's signal level in the broad position is dramatically less than it is in the sharp position, whereas the Philco's levels are almost uniform from sharp to wide.
Fisher AM80 IF stages
The Fisher has a 3-position selectivity switch, while the Philco is continuously variable. In the "Medium" position (not shown), the bandwidth of the Fisher is 13 kHz, while in the "Broad" position it's a full 20 kHz wide (compared to 10.2 in the Philco). The 10 kHz audio response of the Fisher in the "Broad" position may be the one advantage it has over the Philco.
The superior performance of the Philco is due to the physical design of the coils. The Fisher switches in extra windings to vary the amount of overcoupling needed to achieve broad IF response. The Philco physically moves the coils closer and further apart.
Philco 37-116 IF stages
One feature of the Philco that some may find disagreeable is their practice of "ganging" the tone control with their variable selectivity scheme, thus increasing the treble while increasing bandwidth, and decreasing treble while decreasing bandwidth.
Philco IF coils
We hope to make this survey of IF response curves a regular feature of the website. There are a lot of AM receivers out there with variable IF bandwidth, from McMurdo Silver and Scott to Wards Airline and from Heathkit to Collins. Thanks to Allen, we can now see how they really perform. To see more of Al's custom-built test equipment go to museumofbroadcasting.org/Lein.html.