Rob's web

Home - Techniek - Electronica - Radiotechniek - Radio amateur bladen - QST - Improving your receiver with a frame-grid R.F. pentode

The many requests for information on the use of the EF183/6EH7 variable-mu frame-grid pentode mentioned by the author as an r.f. amplifier in an earlier QST article(1) have forced the following conclusion: there still exists that breed of radio amateur who would rather rebuild his present equipment than trade it in on a new model to obtain the best possible performance. This article is dedicated to that breed.

A knowledgeable reader of this article may correctly point out that atmospheric and manmade noise levels will normally exceed the low noise levels in the author's receivers. But that is how it should be! A quick method of checking receiver performance is to disconnect the antenna and note what happens to the audio-output noise level. It should drop when the r.f. gain is wide open and the receiver is set at its upper frequency limit. It should also be possible to peak up the antenna noise with the antenna trimmer control. If these two checks are marginal, then pull out the r.f. amplifier tube to cause a drop in noise. If the noise still doesn't drop, you know that the mixer or converter stage is contributing more noise than the r.f. amplifier or the atmospheric and man-made noise levels coming in from the antenna. Improving the receiver is now up to you.

Many articles on receiver improvements have appeared in the amateur magazines and there are sections in the ARRL Handbook on this subject, so no attempt will be made to give credit to all reference material used in this article. Some credit is due the old HQ-120X receiver which has served as a guinea pig during the past decade of rebuilding and experimentation. This particular receiver (a Navy version of the HQ-120X) went through the Second World War and was salvaged from a junk heap around 1950. The first attempts to improve it were started when W1NXY(2) discussed some changes in the postwar HQ-129 receiver. This consisted of changing the mixer stage to obtain lower noise operation. Revisions in the r.f. stage were already underway when W5UOZ(3) discussed this and other areas of the same receiver. Not all of the work through the years on the guinea pig resulted in improvements - the cathode-coupled triodes and cas-coded triodes were tried out as r.f. amplifiers and rejected for reasons discussed later on. A 68Z6 pentode was used for several years as an r.f. stage, but it was always unstable above 20 Mc. The present 6EH7 has been used as an r.f. amplifier in the guinea pig since early 1962, and it will continue in that role until a better tube is invented. This article not only discusses how to use the 6EH7 to obtain better performance, but it also discusses the principles used by the author to evaluate new tubes as they are introduced on the market.

The following list presents most of the factors which should be kept in mind when planning receiver front-end improvements.

  1. R.f. stages which were originally designed to operate with remote-cutoff pentodes should be rebuilt with better remote-cutoff pentodes and not sharp-cutoff pentodes or any triodes.
  2. The transconductance of the new pentode should be higher than that of the old pentode.
  3. The cathode current of the new tube should not exceed 20 milliamperes.
  4. The grid-to-plate capacitance of the new pentode should be equal to, or less than, that of the old pentode.
  5. The sum of the pentode's shot noise and partition noise should be as small as possible.
  6. The dynamic plate resistance of the new pentode should not be less than .5 megohm.
  7. The cost and availability of the new tube should be such that the average receiver can readily be modified.
  8. The heater voltage must match that of the old tube.

The factors given above are not all independent, so a compromise must usually be reached when considering an assortment of tubes. Various methods of weighing the importance of these factors are discussed in the following sections of this article.

Pentode versus triode

A number of articles have appeared on the subject of receiver improvements by using triodes to replace pentodes, but it is the author's experience and opinion that the resulting loss in r.f. gain and selectivity do not justify the reduction in tube noise. The loss in gain and r.f. selectivity is a result of the low dynamic plate resistance of the triode which swamps the Q of the resonant r.f. plate circuit. These plate circuits were originally designed to yield the desired Q when operated with the high plate resistance of pentodes. The dynamic plate resistance is in parallel with the plate circuit when considering the equivalent circuit of the tube, plus the plate circuit. For this reason, a plate resistance which is below 0.5 megohm is not desirable.

High transconductance

If there had to be just one criterion for evaluating receiver pentodes, it would be for higher trans-conductance. This primary characteristic determines the obtainable gain as well as the shot noise and partition noise. But in any practical application, higher gain can be utilized only if the stage is stable, and this is primarily a function of the tube's grid-to-plate capacitance. Thus, higher transconductance can be handled only if the Cgp is about equal to that of the old tube. Also, in practical applications, higher gain presupposes the end result of detecting weaker r.f. signals, but the minimum detectable signal is determined by the combined atmospheric, man-made and tube noise present at the front end of the receiver. We cannot do anything in the receiver to change the atmospheric and man-made noise levels. The tube noise of a pentode is usually considered to be the sum of the shot noise and the partition noise. The shot noise is reduced by higher values of transconductance. The partition noise is reduced by a combination of higher transconductance and a smaller ratio of screen current to cathode current. Some sharp-cutoff pentodes are on the market (the 7788 is one example) which yield low shot and partition noise by having both high transconductance and a low ratio of screen-to-cathode currents. But such tubes, even if a remote-cutoff version were available, would not be usable for improving a communications receiver because the total cathode current (45 ma. for the 7788) might cause heating in the Litz wire of the coils in the receiver. For this reason, an upper limit on the total cathode current was set at 20 ma. The price of the 7788 would also be a disadvantage for this particular tube.

Evaluating new tubes

The data on the new tubes which are introduced on the market, seldom include specific mention of shot and partition noise, so recourse must be taken to compute this information from the data which are furnished. Shot and partition noise is usually spoken of in terms of the equivalent resistors which would give rise to the observed noise voltages. This analogy arises from the observable noise voltage which is present across the terminals of any resistor due to the random motion of the electrons which are present in the resistance material. This has the classical name of Johnson Noise. The equations which are in use for computing the values of these equivalent resistors are usually approximations instead of precise equations. This makes the computation easier. But it also results in a variety of approximation equations. The author here prefers to use two different approximations, one from the MIT reference 4 and the other from the Radiotron reference.(5) The results of both equations are then used to establish a ball-park figure for the equivalent noise resistance of the tube in question. The equations appear below:

Eq 1

gm = Transconductance in mhos
IK = cathode current in amperes
IP = plate current in amperes
ISG = screen grid current in amperes

A selection of both sharp and remote-cutoff pentodes is listed in Table 1 to show the results of the two equations. Keep in mind that a high-gain (high-transductance) pentode is desired which has the lowest possible equivalent noise resistance. The other columns in Table 1 are discussed elsewhere. The information on the sharp-cutoff tubes is provided as reference material for use when selecting a low-noise mixer stage. The reader can observe that the 6EH7 has the lowest computed equivalent noise resistance of the remote-cutoff pentodes listed in Table 1.

Table 1 - Tabulation of pentode tube data
noise resistance
gm/Cgp 1012
gm/Cgp 109
6JH612512517.614.03.68,000.0150.26 8221,140.533.648

Ratio of Transconductance-to-Cgp

As mentioned previously, a high-transconductance pentode will have merit as an r.f. amplifier only if the grid-to-plate capacitance is low enough to give stable operation. The 6EH7 does not have the lowest Cgps, as shown in Table 1, but it does have the highest gm. The obvious way to get a relative comparison between r.f. pentodes, is to look at their ratios of gm/Cgp. This is tabulated in one of the columns of Table 1. Again, the 6EH7 shows up as the best tube when the criterion is for the highest gm-to-Cgp ratio.

Ratio of gm/Cgp-to-Req

For the case where a new tube may not have the lowest Req but does show the highest gm/Cgp ratio (or vice versa), a relative comparison based on the ratio of gm/Cgp-to-Req would be helpful. This latter ratio is tabulated in the last column of Table 1. The lowest value of the two approximated Rea's was used for computing this ratio. Again, the 6EH7 is the best available tube for use as a gain-controlled r.f. amplifier when the criterion is for the highest gm/Cgp-to-Req ratio.

Fig 1
Fig. 1. Circuit diagram of a gain-confrolled frame-grid r.f. amplifier.

Selecting a mixer tube

With a few exceptions, most of the comments on selecting an r.f. tube apply to the task of selecting a mixer tube. Since the plate and grid circuits are not tuned to the same frequency, feedback by way of Cap should not produce instability. Some mixers are desired which have a broad-band response (such as in converters), so a lower value of dynamic plate resistance can be tolerated. This lower value cannot be tolerated, though, if the mixer is in a communications-type receiver where high-Q i.f. transformers follow the mixer stage. The 6EJ7 sharp-cutoff pentode (Table 1) is a good candidate for mixer service. The author uses four of them in this role, in four different receiving systems. The 6EJ7s have also been put to use as i.f. amplifiers in an f.m. tuner and a pre-i.f. noise-silencer.

An R.F. amplifier circuit

The 6EH7 has been used as an r.f. amplifier by the author in the HQ-120X receiver as well as an Eddystone 888A ham-band receiver and an RME DB-20 preselector. The circuit shown in Fig. 1 represents an r.f. amplifier which can be adapted for use in any receiver covering all, or parts, of the 0.55-Mc. to 30-Mc. range. The unspecified plate and screen-dropping resistors must be selected on the basis of the available supply voltages and the required plate and screen circuit currents. Please note the specified values of cathode and screen bypass capacitors. These values were selected to form series-resonant circuits where the inductance is in the form of the capacitor leads. Do not use larger values of bypass capacitors unless the amplifier is for use only on lower frequencies. A 0.01-µf. bypass can be used at 7 Mc. and lower, while a 0.1-µf. bypass can be used at 2-Mc. and lower. The unbypassed 22-ohm cathode resistor is used to compensate for variations in input capacitance and resistance which otherwise would occur when cathode or grid voltages are changed. Pin 6 is used to ground the tube's internal shield. An external shield is also recommended. The heat-dissipating style such as IERC's TR-6-6020B will assure long tube life. Something not shown in Fig. 1, but which is always used by the author, is a shield partition which straddles the tube socket. A piece of 1/32 inch sheet brass, which is about 3 inches square, is installed to pass between Pins 1 and 9 and between Pins 5 and 6. Pins 5, 6, and 9, the center post of the socket, and the grounded ends of the bypass capacitors, are then soldered to this brass plate.

Of course, the r.f. circuits will have to be realigned after the new tube is installed.

A mixer circuit using the 6EJ7 pentode

The mixer circuit which was used in the HQ-120X appears in Fig. 2. A separate local oscillator using a 6AK5 was installed on a subchassis underneath the main chassis. The 9-pin socket for the mixer was mounted on an adaptor plate which replaced the old 8-pin socket. The variable cathode resistor and the trimmer capacitor between the oscillator plate and the mixer grid are adjusted so that the mixer has high gain while remaining stable over the entire frequency range. Too much oscillator injection or too little mixer bias will produce "birdies." A triode local oscillator should not be used if pulling of the oscillator frequency is to be avoided on the higher frequency bands.

Fig 2
Fig. 2. Circuit diagram of an im3roved mixer and local oscillator for the HQ-120X receiver.

Dual conversion for the HQ-120X receiver

The image rejection of this receiver was not very good when operating above 10 Mc. because of its relatively low i.f. of 455 kc. A subchassis was installed under the main chassis of the HQ-120X which provided dual-conversion capabilities above 10 Mc. A three-pole, double-throw wafer switch was mounted behind the front panel with its shaft coming out just to the left of the sensitivity control and below the send-receive switch. The schematic of this dual-conversion unit is shown in Fig. 3. It uses a 6U8A triode-pentode, with the triode serving as a crystal-controlled oscillator and the pentode as the mixer. The frequency of this crystal can be between 1855 kc. and 2055 kc. The 1965-kc. unit was obtained from one of the surplus crystal companies. The cathode resistor of the pentode might have to be adjusted a little to obtain stable mixer operation. Of course, the first local-oscillator circuits had to be realigned when the dual-conversion feature was incorporated in the receiver. The fixed-padder capacitors which were in series with the oscillator coils, for the two bands above 10 Mc., were replaced with Arco 307, 350-1180 pf., variable padder capacitors. A signal generator which provided marker signals every 1 Mc., as well as a variable-frequency signal, was useful for the realignment work.

Fig 3
Fig. 3. Circuit diagram of the dual-conversion unit for the HQ-120X receiver.

Changing the HQ-120X over to dual conversion entailed a lot of work, but the elimination of the image problem was well worth it. The elimination of the images was especially rewarding when using this receiver as a tunable i.f. in the 14-Mc. to 18-Mc. range, which is the output frequency of the 2-meter converter used by the author. The noise present at the image frequency tended to degrade the performance of the converter by approximately 3 db. when the 455-kc. i.f. was used at 14 Mc.

Actual Results

Some reasonable questions come up every time the subject of receiver improvements is discussed. How can I check my present receiver's performance? What improvements can be obtained with these frame-grid tubes? The first question was answered in the introduction to this article.

The noise figure which can be expected with the 6EHF at 30 Mc. is approximately 5 db. The improved HQ-120X was able to copy a c.w. 0.03-microvolt signal from a 50-ohm calibrated signal generator at 30 Mc. before the dual conversion was installed. This 0.03-µv. level was where the c.w. signal dropped into the noise. The narrow-band crystal filter was in use during the c.w. test. An a.m. signal, modulated 40 percent with 400 c.p.s., dropped into the noise at 0.06 microvolts without a crystal filter. The S-meter circuit in the receiver was rebuilt and calibrated for use at 15 Mc. with the 2-meter converter. Table II shows the a.m. test-signal levels present at the antenna jack of the HQ-120X from a 50-ohm generator operating at 15 Mc., which produced the various S-meter readings. These measurements were obtained before the dual conversion was installed.

Table 2 - S-Meter Readings at 15 Mc.
S-UnitInput µV 50Ω Source

The Eddystone 888A ham-band receiver, modified by the author, was checked with the calibrated signal generator on c.w. at 29.5 Mc. The signal dropped into the noise at 0.01 microvolts. The Eddystone has dual conversion, a 1-kc. i.f. bandwidth and an 80-c.p.s. audio filter for use on c.w.

The author will be glad to answer general questions accompanied by a self-addressed stamped envelope, but specific requests to redesign particular receivers will have to be declined.


  1. Balogh, "A Low-noise 2-meter converter," QST, April, 1964.
  2. Santangelo, "Second guessing the experts on the HQ-129A," CQ, April, 1952.
  3. Stueber and Noe, "HQ-129X receiver improvements," CQ, May, 1959.
  4. Vacuum Tube Amplifiera, p. 625, MIT Radiation Lab. Series. Vol. 18, McGraw-Hill, 1948.
  5. Radiotron Designer's Handbook, p. 937, Fourth Edition, Distributed by RCA, December 1957.

Joel Balogh, K3CFA.