Practical Design of Mixer Converter Circuits
February 1941 QST

In the present era, designing a frequency converter circuit consists in most cases of picking from a catalog an IC or connectorized component that has the characteristics you need from a gain and mixer spurious product standpoint. Add a couple filters, a local oscillator (although in some cases the oscillator is part of the IC), and a power supply, and you're good to go. Of course there are special cases where you have to use a basic mixer and do everything yourself, but even that is simpler than designing a primary circuit using diodes or vacuum tubes as rectifiers. Obtaining match sets for good mixer spurious product cancellation is very difficult, especially in a large volume production environment. It really is amazing what engineers and hobbyists of yore were able to accomplish using point-to-point wiring and a slide rule. Here is a good article form the February 1941 QST magazine that discusses some of the considerations. Maybe you have an old radio to which this knowledge will apply.

Comparison of Tube Types and Checking Performance

By Curtis R. Hammond (W9PKW)

The design of an efficient mixer or converter circuit is often the one thing that prevents the amateur from building his own communications receiver. In application the amateur usually is unable to tell whether or not the stage is giving normal performance and, lacking equipment for checking gain, no attempt is made to find out if it is doing the job efficiently. However, there are simple ways of determining whether or not a mixer or converter is operating efficiently, and it is the purpose of this discussion to explain these methods and to give some theory on the operation of converters. The general characteristics of the several mixers and converters now available are also given, with a general discussion of the performance characteristics of each.

An elaborate mathematical theory of the operation of a converter or mixer¹ is of no great importance for our particular problems. Roughly, a converter operates as follows: Within the tube there is developed a current at oscillator frequency which is modulated by the incoming signal to produce an intermediate frequency. The ability of the tube to develop a current at an intermediate frequency is given by the" conversion conductance," which by definition is the ratio of an incremental change in intermediate frequency current to the incremental change in r.f. signal voltage that produces the current. This conductance in micromhos is published for all converters, and its use to calculate stage gain is analogous to the use of mutual conductance with r.f. amplifiers. The gain equation for a single tuned load is , where G_c is the conversion conductance, R_p is the plate resistance, and R_L is the tuned load resistance. Published values of plate resistance and conversion conductance can therefore be used to calculate conversion gain. The tabulation following gives a comparison of gain for a group of tubes now generally available. The gain figures were calculated for a tuned load impedance of 200,000 ohms, which is equivalent to the better transformers now available.

If gain was the only consideration the above would suffice for the selection of a converter tube. Tube noise is generally not a consideration when comparing converters simply because the converter is inherently a noisy device and most converters develop noise voltages of approximately the same magnitude. The noise output of converters of the 6A8 and 6SA7 type is approximately 4 times greater than that of an r.f. amplifier like the 6SK7 or 6K7. Where the ultimate in signal-to-noise ratio is desired it is necessary to precede converters of this type with an r.f. stage. Usually the selection of a converter is based on the characteristics of oscillator stability with regard to a.v.c. and terminal voltage fluctuation, pull-in characteristics, oscillator transconductance that determines the ease of oscillation especially at high frequencies, and other deleterious characteristics that cause loss in performance at certain frequencies. The chart on page 41 indicates some of the characteristics of the various converters. The gain figures and notes on stability and oscillator transconductance are of particular importance.

In general the converters perform equally well as mixers or as converters with the exception of the one characteristic of oscillator stability. Any of the converter tubes gives good stability if used with a separate oscillator and the circuits are isolated properly. Of the group the 6SA7 makes the best mixer because it gives high gain and has improved internal shielding of the signal and oscillator grids. The improved shielding is accomplished by using shielding plates similar to the beam-forming plates used in beam power tubes. These plates are attached to the side rods of the screen grid and confine the electron currents to beams which get into the outer regions of the tube where they are modulated by the signal grid. The sketch of Fig. 1 shows the construction of the 6SA7. The side rods of the No.3 or signal grid are mounted so that they split the' beam and make the electrons travel in radial paths. Electrons turned back by the signal grid because of a strong r.f. voltage do not return to the oscillator or No. 1 grid because they are caught by the collector plates. This reduces coupling between the signal and oscillator grids and improves stability. Simple structures of cylindrical grids such as used in the 6L7 and 6A8 do not have this additional isolation and are therefore not quite as good as the 6SA7. The improvement in stability evidences itself in the form of greater freedom from "pull-in " - that is, shifting of the oscillator frequency with signal-grid tuning or with a strong signal on the signal grid. This effect is usually not as serious as frequency shift due to terminal voltage fluctuation. The remarks relative to stability, given in the tabulation on page 41, refer to the stability with regard to terminal-voltage fluctuation.

Converter Circuits

Typical circuits for the six converters listed in the tabulation are shown in Figs. 2 to 7 inclusive. The 1A7G, 1R5, 6K8, 6A8, and 6SA7 can be used with separate oscillators simply by connecting the oscillator grid of the converter to the oscillator grid of the oscillator tube. The screen and other positive electrodes should be maintained at their normal rated d.c. voltages but should be by-passed to ground.

Fig.  2 shows connections for a converter circuit using the 1A7G and Fig.  3 shows connections for the 1R5. The 1R5 is one of the new miniature tubes for hearing aids and small portable receivers. The 1A7G has the conventional 6A8 construction, using an anode for feedback. The chart above indicates that the gain obtainable with either tube is approximately 34. The oscillator transconductance of the 1R5 is slightly higher and the oscillator stability is somewhat better. These two features are of advantage for high frequencies.

Figs. 4, 5, 6 and 7 show connections for converter circuits with types 6A8, 6K8, 6J8G and 6SA7 respectively. The high oscillator transconductances of the 6K8 and 6SA7 make them particularly suited for all-around usage. They oscillate strongly at high frequencies where Lie ratios are unfavorable. The 6A8 construction is not satisfactory for amateur usage because of instability in the oscillator. The oscillator electrode is a pair of rods located in the tube between the No.1 grid and the screen. These side rods collect electrons from the cathode stream and the electrode current is controlled by the No.1 grid. Unfortunately, changes in signal-grid or screen voltage also change the anode current. This conductance between signal grid and oscillator causes instability with variation in a.v.c, voltage. Fluctuations in screen voltage due to supply regulation also change the frequency. As a result, the 6A8 is subject to motorboating or "put-put" at high frequencies. Dial calibrations also drift with line voltage fluctuations. "Pull-in" is particularly bad with the 6A8.

The 6J8G construction incorporates a triode oscillator and a mixer section with a common cathode. This construction results in good stability insofar as screen and a.v.c. voltages are concerned. The 6J8G has two serious disadvantages, however, that have limited its application. The triode section shares a portion of the cathode area. The area used by the triode is quite small and as a result the oscillator transconductance cannot be made high. Also, at high frequencies a peculiar effect is experienced that causes a flow of current to the signal grid. This current causes a high negative potential across the resistance in the grid return, and this bias reduces the gain of the mixer. The effect can be reduced somewhat by using a high value of screen voltage, but it is then necessary to increase the bias to hold the cathode current to a safe value.

The 6K8 has been used extensively by the amateur and also the commercial manufacturer principally because it gives fair stability, and design problems are usually simple. The tuned-grid oscillator shown in Fig. 5 gives very little trouble and is easy to build. The oscillator frequency is not independent of screen and .v.c. voltages, but in most designs the frequency shift caused by one is offset by the other so that good stability is obtained. The 6K8 has an effect known as space-charge coupling which is experienced at high frequencies. This effect is as follows: The oscillator voltage on the No.1 grid causes a fluctuation in the number of electrons in the region of the signal grid. The electron density changes at the oscillator frequency and as a result a displacement current flows into the signal grid. At high frequencies where the signal grid and oscillator frequencies are quite close, the impedance of the signal grid circuit at the oscillator frequency is quite high and as a result the displacement current produces an a.c. voltage across the signal grid circuit. This voltage, when smaller than the bias, reduces the gain of the tube slightly. Under extreme conditions it overrides the bias and causes rectification in the signal-grid circuit, causing a serious loss in gain. The coupling can be neutralized by a small capacitance - approximately 2 or 3 μμfd - between oscillator and signal grids. Commercial practice is to use a condenser (known as a "gimmick") made by wrapping two pieces of wire together to give the desired capacitance. Neutralizing the space charge increases the gain and image ratio.

The 6SA7 construction has already been described. Using cathode feedback in the Hartley circuit shown in Fig. 7, excellent stability is obtained. The gain is quite high and the high oscillator transconductance makes a good oscillator.

The 6SA7 converter is tricky to use because the cathode returns through the oscillator coil. This connection, however, is the secret of the stability resulting with the 6SA7. The feedback is obtained from the total cathode current. A.v.c. voltage variations on the signal grid do not change the cathode current appreciably so that the oscillator frequency is almost independent of a.v.c. Screen-voltage variation produces a shift in frequency in the opposite direction and the two effects practically cancel. The frequency change with either variable is reduced by using the optimum tap on the oscillator coil. With average oscillator coils the tap should be adjusted to give a total oscillator voltage of approximately 10 volts grid-to-ground. Under these conditions the oscillator grid current measured in the grid leak will be approximately 0.5 milliampere. This current can be measured with a 0 to 1 milliammeter by connecting it at the bottom of the grid leak.

At high frequencies it is necessary to keep the leads connecting the cathode to the coil, and the bottom of the coil to ground, as short as possible. The cathode lead in particular should be short. The inductance of this lead is not a part of the oscillator tank and oscillator voltage developed across it does not contribute to feedback. The voltage does bias the signal grid, however, and will reduce the gain of the converter. Under extreme conditions the voltage may be high enough to cause a flow of current in the signal-grid circuit. This current results because of high voltage between cathode and ground and because of phase shift of this voltage with respect to the voltage between grid and cathode on the coil. The cathode connection to the coil should also be made so that the lead pulls away from the coil at right angles. By pulling the wire away parallel to the winding the cathode-lead inductance may cancel a portion of the tap-to-ground inductance.

In band switching arrangements the circuit of Fig. 8 is recommended. It will be noted that the tap switch on the oscillator coil is located at the ground end of the coil. This puts the inductance of the switch and its connecting leads within the closed tank circuit. Since the tank currents flow through this inductance it contributes to feed-back and gives oscillation with a minimum of cathode-to-ground voltage. If the switch was between the cathode and the coil in the position of lead 1 the drop across the switch inductance would not contribute to oscillation, but would produce a high cathode-to-ground voltage. As mentioned above, this voltage is shifted in phase from the voltage in the tapped portion of the coil and may cause the signal grid to be driven positive and cause rectification.

The circuit of Fig. 9 shows the 6SA7 as a mixer. It will be noted that the neutralizing condenser C_n. is used to neutralize the space charge. The 6SA7 as a mixer gives an increase in gain over that realized as a converter.

Space-charge coupling is also experienced with the 6SA7, and a "gimmick" is required for neutralization. This coupling is characteristic of converter or mixer systems wherein the oscillator voltage is injected next to the cathode or filament. The 6J8G, although not having this coupling, has the transit-time effect which is just as bad and cannot be neutralized. The transit time effect is experienced with converters or mixers in which the oscillator voltage is mixed in the cathode stream outside of the signal-grid injection.

It might be of interest at this point to give the accepted theory on what causes the transit time effect. Electrons accelerated through the No.2 screen grid approach the No. 3 injector grid. At high frequencies, where the time of transit between cathode and No.3 grid is an appreciable portion of the period of oscillation, electrons accelerated by the No.3 grid on its positive swings reach the grid at a time when it is going negative and are repelled and turned back toward the screen. On the way back they are accelerated by the positive potential on the screen and by the increasing negative potential of the No.3 grid. Many of these returning electrons reach the screen and are drawn off as additional screen current. Some of the electrons, however, pass very close to the screen and are accelerated toward the No. 1 grid at high velocity; many of them obtain sufficient energy to overcome the negative potential of the No. 1 grid and flow in the external No. 1 grid circuit. This flow of current is d.c., and in a direction such that the drop in the external resistance increases the bias. If the tube is operated from the a.v.c. string as in the conventional case, the total return to ground is of the order of two megohms. A current of several microamperes increases the bias sufficiently to cause an appreciable loss in gain. The current can be eliminated for frequencies up to approximately eighteen megacycles by increasing the bias and the screen voltage.

The above information should be useful in determining the converter to be used for a particular job. Once the converter is built it is comparatively easy to ascertain whether performance is satisfactory. Of course in the laboratory the most satisfactory method is to check stage gain with a signal generator, but few of us have signal generators with which to make precision measurements. We usually rely on the sound of the set and whether it pulls in the signals.

The first check on any converter is to measure the electrode voltages with a high-resistance meter. The correct voltages are indicated for the various circuits. Next in order of importance is to check to see if the oscillator amplitude is high enough. The easiest method of checking this is to measure the d.c. grid current in the grid leak. This grid current increases directly with oscillator voltage and is so closely related to oscillator voltage that manufacturers, instead of rating the oscillator voltage to be used with a converter, rate the grid current as measured in a recommended grid leak. On each of the preceding circuits the rated oscillator grid current is given. In practice the grid current cannot be held to this value over the band, especially if a wide tuning range is desired as in commercial broadcast sets. In communications receivers where the tuning range is small the variation is not large. A 2-to-1 variation in a set having a wide tuning range is not bad. If rated grid current is obtained in the middle of the band the variation over the band is usually not excessive. The grid current is important because it determines the point of optimum gain, and other than rated value results in a sacrifice in performance.

Converters using the 6A8, 6K8, 6SA7, 1A7G, or IR5 should next be neutralized for space charge coupling. This is accomplished by connecting a "gimmick" between the oscillator and signal grids. If a gang condenser is used and the oscillator and signal grid sections are adjacent, neutralization can be accomplished by connecting the "gimmick" between the stators of the two sections. Commercial practice is to solder two small pieces of wire to the stator lugs and then to twist the ends together. About two turns is satisfactory. Note: Neutralization is done on the high-frequency edge of the highest-frequency band. Low-loss wire should be used. The capacitance should be adjusted to give maximum sensitivity.

There are several phenomena that can take place that will upset performance after the above considerations have been observed. Parasitic oscillations take place in the oscillator section if too much feedback is used or if the values of grid coupling condenser and grid leak are too high. A 50-μμfd grid condenser is usually satisfactory for most circuits. Most grid-leak specifications call for 50,000 ohms. Battery tubes having low oscillator mutual are specified with as high as 200,000 ohms, and the 6SA7 with its high oscillator mutual or transconductance is rated with 20,000 ohms. If the oscillator and signal-grid circuits are not adequately shielded and isolated, severe coupling between circuits is obtained at some frequencies. The signal-grid circuit in extreme cases may load the oscillator enough to cause it to stop oscillating. This effect can be detected by observing the oscillator grid current as the set is tuned through the coupling point. A rapid dip in the oscillator grid current is experienced as the coupling point is passed. Shielding of coils and isolation of parts and leads eliminates this trouble. Motorboating on strong signals is the result of oscillator shift with a.v.c. and other element voltage variation. It was pointed out that the 6A8 was particularly bad in this respect, that the 6K8 was much better, and that the 6J8G and·6SA7 are very good. Motorboating can be experienced with the 6J8G and 6SA7 if power-supply regulation is bad and if the oscillator amplitude is not adequate. Stability is improved by operating at or somewhat over rated amplitude.

The major troubles experienced with converters produce a flow of grid current in the signal-grid return. This is true of the transit time effect with the 6J8G, the space charge effect with 6K8, 6SA7, 6K8, 1A7G and 1R5, and the phase shift of the high cathode to ground voltage in the 6SA7. The circuit of Fig. 10 shows how a check for signal-grid current can be made without the use of a sensitive microammeter. An electron-ray indicator tube such as the 6U5/6G5 will indicate any current flow in the a.v.c. return. Most returns have about three megohms total and a d.c. current of 1 microampere will produce 3 volts, which will make a noticeable deflection on the target. The voltage drop between the bottom end of the coil and ground should never exceed approximately 1.5 volts. This voltage can exist because of contact potential in the diode and other grids connected to the a.v.c. system, and does not indicate trouble.

Signal grid current with the 6A8, 6K8, and 1A7G usually results from space-charge coupling, as already described. A convenient test for its presence is to short the signal-grid tuned circuit with a condenser. This shorts out the voltage and eliminates the current. 'The "gimmick" when adjusted properly neutralizes space charge coupling.

Signal-grid current because of space-charge coupling is also obtained with the 6SA7 but in addition current can flow because of high cathode-to-ground voltage and phase shift of this voltage with respect to the oscillator grid-to-cathode voltage. If bypassing the signal grid does not eliminate the current, the trouble will be found in the oscillator coil and connecting leads. The cathode lead should be kept short and the circuit of Fig. 8 adhered to. The ratio of length to diameter of the oscillator coil should not exceed more than about 1.5 to 1. With long coils and small diameters there is appreciable phase shift with attendant troubles. As mentioned previously the cathode lead should pull away from the coil at right angles so that it does not couple to the coil.

Recently, certain manufacturers have used triodes for mixers. A typical circuit for this type of mixer is shown in Fig. 11. It will be recognized as similar to many of the circuits used in the older days. In commenting on this circuit it might be said that the chief advantage of the triode is that it develops very little noise. It is thus possible to add extra gain behind the converter in the i.f. and get high sensitivity with a good signal-to-noise ratio. The triode in this connection has serious disadvantages, however. It is necessary to use a special low-impedance primary i.f. transformer so that the grid-to-plate capacitance of the triode will not cause loading of the signal-grid circuit. In the practical case the tuning condenser required to tune the i.f. primary is approximately 2000 μμfd. The high cathode-to-grid capacitance causes severe coupling of the oscillator and signal-grid circuits. This evidences itself in the form of instability with a.v.c. variation, "pull-in " on strong signals, and oscillator shift with tuning of the signal grid circuit. In applications where stability is not of prime importance a pentode such as the 6SJ7 or 6AB7/ 1853 could be used to give good signal-to-noise ratio. The low signal-grid-to-plate capacitance in these types would allow the use of conventional i.f. transformers.

* Ken-Rad Tube & Lamp Corporation, Owensboro, Kentucky.

1 - In common terminology, a "converter" is a tube performing the dual functions of mixer and oscillator; a "mixer" does not incorporate an oscillator section. Any converter tube can be used as a plain mixer by providing excitation from a separate oscillator tube. - ED.

Posted September 4, 2023
(updated from original post on 6/16/2011