Issue #105 "Web Version"

AM Forum at Dayton April 24

On Saturday, April 24, another AM Forum will take place at this year’s Dayton Hamvention. The program will include a slide show "Amateurs and Their Broadcast Transmitters featuring photographs of ex-broadcast transmitters converted for ham use, which have become increasingly available as broadcast stations replace their older tube type equipment with solid state. The feature event will be the unveiling of the reactivated AM International, designed to be an active organization to promote AM and preserve its status in amateur radio. (See the cover story AK P/X issue #103.) This will be followed by a panel discussion on AMI and other relevant AM issues. We are hoping to be able include a little "getting acquainted" time after the formal program, to allow those in attendance to meet and talk with their follow AM’ers in person. As in years past, speakers at the AM Forum will include Dale Gagnon, KW1I, and Don Chester, K4KYV, editor and publisher of The AM Press/Exchange,

1993 promises to be an exciting year at Dayton far issues involving AM, following the recent articles in QST. Don’t be surprised if you see a lot more AM equipment at the flea market now that the mainstream amateur radio community has been alerted that there is a market for the boat anchors many hams have stored away for decades, thinking no-one had any interest in that sort of thing anymore. If there is more AM equipment than usual, we can expect much of it to be overpriced by vendors with dollar signs in their eyes and inflated ideas about the actual worth of the stuff.

This will be the first Hamvention since the widespread introduction of new no-code licensees into ham radio, thus there is the likelihood of a bigger crowd than usual, so prepare for the possibility that parking hassles and large crowds of people will make it a little harder to get around to see everything and handle the logistics of purchasing heavy items at the fleamarket and getting them to your vehicle. This is becoming a greater problem each year as the event continues to outgrow the facilities at the HARA Arena and we are faced with the problem of no convenient parking adjacent to what is probably the world’s largest electronics flea market.

With this in mind, we are already anticipating the thrill and excitement of this year’s event, and bracing ourselves for the grueling experience of running on little more than adrenaline for three days in a row.


HARD ROCK—or, Don’t Blame The Crystal

by George Watson, W0LOB

Let’s discuss piezoelectric crystals—design, function and application. First of all, I’m not a crystal designer or a piezoelectric scientist, but one, who over the years, has used many crystals. Out of necessity, I have had to pick up bits and pieces of information about crystals to understand why a given circuit did or didn’t function properly. Passing some of this information on, may in turn, help someone understand his particular crystal circuit a little better.

Most hams know that the heart of the crystal unit is a crystalline blank of natural or cultured quartz. Cultured quartz came into prominence after WW II and is used almost exclusively today. When the crystal blank is excited with an alternating potential of the proper frequency, it will exhibit mechanical resonance at that frequency. This frequency depends on the blank dimensions, primarily the thickness, and the mounting and blank contact techniques employed. The thicker the crystal blank, the lower its resonant frequency. Mechanical damping such as pressure plates, bonding wires, etc will alter this frequency slightly.

The ‘Q ‘ of this mechanical resonator may be a thousand times greater than that of a similar LC circuit so a quartz crystal is thus ideal for precise frequency control in feedback circuits such as oscillators or in filter applications- Losses in FT-243 type crystals can lower the ‘Q’ of the assembly considerably below the ‘Q’ of the crystal blank alone. The internal construction of the crystal unit is typically designed to minimize the effects of holder and lead wire ‘L’ and ‘C’ values, thus permitting the blank to be the sole frequency determining element.

The properties of the blank (temperature stability, activity, frequency of oscillation, etc) depend upon the angle of ‘cut’ through the parent quartz block, with a so-called ‘AT’ being primarily the cut for most Ham crystals (because of temperature stability. See Fig 1). This ‘AT’designator defines very accurately the angle of sawing with respect to the XYZ axes of the parent block.

Click here to see Fig. 1 - Typical Temp/Freq. Characteristics of ‘AT’ cut.

The contacts to the crystal blank are brought to the outside world through pressure contacts, such as the FT-243 spring and pressure plates, or through wires bonded to plated metal contacts on each face of the crystal blank. Each side of an HC-6 crystal, for example, has metal, vapor deposited or sputtered on, to which the wire bond is made. In the VHF range, non-plated, pressure mounted crystals are sometimes preferred because plating can dampen the high frequency response of the crystal.

The case is sized to hold the particular blank and interconnects. Some of the newer cases are as small as " square. See Fig 2. Compare this size with some of the 1930’s quarter pounders!

Click here to see Fig. 2 - Representative Crystal Holders

Transistor-type cans are also used with the blank mounted flat (parallel to the case bottom-surface), supported at four symmetrical locations around the periphery of the blank. These types will withstand high ‘G’ loading

conditions—shock, vibration and acceleration. All can types are evacuated and back-filled with dry air, typically, to preclude contaminants within the package so as to insure stability over life. Glass housings are also used and provide an ideal crystal unit, particularly for high vacuum applications (Oscar).

Since the frequency of oscillation goes down as the wafer gets thicker one can see why the package for a 100 khz crystal standard is larger than that for a 10 mhz unit, for example. Likewise, a thinner blank means a higher

frequency but because of fragility of the thin quartz blank, the upper frequency for fundamental operation is limited to about 30 mhz.

The purity of the output signal of the crystal blank alone depends upon the flatness of the grind of the faces, the parallelism of the faces and to some extent, on the level of drive applied to the crystal. Before we go any further let’s look at a few basics that may help clarify an item or two in the reader’s mind.

The crystal blank may be represented as:

Click here to see Figure 3 - Simplified Circuit of Crystal Resonator

As is typical in mechanical items, multiple resonances occur at approximate odd harmonics (overtones) of the fundamental frequency. Overtones are not exact multiples of the fundamental frequency so the blank is ground at the user’s designated frequency (rather than leaving it up to the user to select a given overtone, which in turn won’t be exactly on frequency). Keep in mind that the overtone operation of a crystal is not the same as operating the crystal on its fundamental and then tuning a following tank circuit to a given harmonic of that fundamental frequency

In Fig 3, (Co) represents the pin-to-pin capacitance of the holder and electrode. The parallel legs 1, 3, 5, -- represent the overtone modes within the crystal itself. (R) is the crystal series resistance, as seen by the circuit, at the series resonant frequency of the crystal. For example, if a 3d overtone (typically 10-75 mhz) is desired, then at (f3) the series resistance of the crystal will be (R3). The crystal manufacturer can supply this resistance data, if requested. A resistor of value (R3) can be substituted, in a series mode circuit, for the crystal (with capacitor DC isolation, if required) and the circuit should oscillate near (f3) if the components have been properly selected. In some cases, particularly at VHF frequencies, (Co) is resonated with a shunt coil at the desired crystal frequency of operation to emphasize operation at the proper overtone. (Co) is typically 5-6 pf in O/T crystals. Keep in mind that the crystal activity becomes progressively smaller as the number of the overtone increases.

The (L’s) in Figure 3 represent the motional impedance, or mass, and the (C’s) represent the bulk capacitance or stiffness of the quartz blank. (Ca) depicts the capacitance across the air gap that is present in pressure-type holders.

At a given frequency of operation one overtone branch will dominate (leg 1 or 3 or 5 or ---), based on the manufacturer’s grinding. The other branches will operate far enough from this resonant frequency, (fx), that their effects will be minimized. (But keep in mind that if the tank circuit has the right L/C values it is possible to tune the circuit to the wrong overtone in spite of what the dominant frequency may be.) The circuit then looks like this:

Click here to see the circuit

Tuning the oscillator tank circuit slightly below (fp) will make the circuit inductive and the crystal will be operating in a stable portion of its reactance curve (not necessarily maximum output power).

Figure 5 shows that at low frequencies, up to the series mechanical resonant frequency, (fs), the crystal is capacitive. Between (fs) and (fp) the crystal is inductive and at higher frequencies it is again capacitive. The frequency between (fs) and (fp) is small (a few hundred hertz) and this may explain why we may think a circuit is operating off frequency when in reality the crystal is ground for one mode (series or parallel) and is being used in a circuit designed for the other mode (parallel or series). It is for this reason that a crystal manufacturer must know the type of circuit (and preferably the exact circuit) in which the crystal will be used. For example, if the crystal is used in a grid circuit, ‘high impedance’ is implied and the manufacturer will typically grind for (fp) at a load of 30 or 32 pf.

If a 30 pf capacitor is connected in series with the crystal, the circuit will be series resonant at the nominal parallel resonant frequency. If the crystal is used in a Butler circuit (feedback K to K or Em to Em) or in a RF grounded grid or base circuit, then low Z is implied and a series resonant crystal will be required. Bipolar Oscillators are typically low Z (series mode) while FETS are high Z, like tubes, and typically use parallel mode crystals. If in doubt as to what type of crystal to order, send with your order a copy of your circuit or, model number of the rig in which the crystal will be used, or both. If critical, send an actual circuit board to the manufacturer in which to test the crystal. If you intend to operate on a specific overtone, also mention this in your order.

What if you are slightly off frequency when your new crystal is installed? This discrepancy could be particularly critical if you are multiplying the frequency several times. Most of the components In the oscillator circuit can affect frequency, even stray capacitance, if allowed to go to an extreme (remember that a parallel mode crystal is ground for operating into approximately 30 pf). Driving the crystal too hard (too much feedback) or loading the oscillator too heavily can ‘pull’ the frequency. Also remember that overtone frequencies are not exact multiples of the fundamental frequency. Looking back at Fig. 4, it seems logical that we can add some ‘L’ or ‘C’ to the crystal and change its frequency slightly since the Lx/Cx combination determines the operating frequency. Old-timers called such an operation "rubbering" The crystal. In fact, in the older crystal units that could be disassembled, a lead pencil could be rubbed on the face of the blank, affording some mechanical damping and thus lowering the frequency.

An inductor in parallel with the crystal (DC isolation?) will move (fp) higher but not affect (fs). An inductor in series with the crystal will lower (fs), add a second resonant frequency but have no effect on (fp). In turn, and simpler to use, a capacitor across the crystal will lower (fp) and a series capacitor will raise (fs). Be aware that in pulling the crystal one sacrifices the crystal stability, the total effect depending upon the slope of the crystal reactance/frequency curve. A Varactor or transistor junction can be substituted for the pulling capacitor and by adding a driving signal, limited angle modulation can be achieved.

What about AGING EFFECTS in crystals? After several hours of use the crystal frequency should stabilize and drift only a few hertz per year of operation. Much of this drift depends upon the quality and cleanliness of crystal manufacture-changes at the surface of the crystal blank. Negative drift typically indicates an unclean blank. For long life and optimum operation the environment in which non-hermetically sealed units operate should be kept below 40% relative humidity. For those concerned about low noise operation, the equivalent noise resistance of a quartz crystal is the same as the effective crystal series resistance at the operating temperature.

Keep in mind that in an evacuated holder the heat is removed from the blank by the attached leads and by radiation. The loss by radiation is much less (higher blank temperature) with a plated blank (e.g. HC-6 type) than with an unplated blank (e.g. FT-243 type). The FT-243 runs cooler.

What about CRYSTAL GRINDING or ETCHING to achieve a new frequency of operation? Ammonium Bifluoride is a chemical etch used today on crystals but I know nothing about it—although I assume it is as dangerous as the acid etch that was used some years back—hydrofluoric acid. Hydrofluoric acid is a common etching material but it is a DANGEROUS material. I don’t recommend its use unless the user is a died-in-the wool chemist. It is nearly the universal etchant—if you get it on your skin or under the fingernails I know of no neutralizer so you patiently wait and suffer until it wears itself out. Extreme care, rubber gloves, etc. are required. Years ago I bought some crystalline material to be mixed into a hydrofluoric acid solution and ended up never opening the bottle after finding out some of the properties of the acid. I may be overstating the hazard, but I do say, "Beware."

If one attempts to grind the blank to raise the frequency, care must also be exercised. This grinding, of course, applies to the old pressure-type mounting assemblies, such as the FT-243. These crystal types are readily disassembled/assembled where the blank has no metalized surfaces. The problem in grinding is in maintaining flat and parallel faces of the blank (not wavy or wedge-shaped). These errors will reduce activity and introduce unwanted birdies in the crystal output. A progressively finer abrasive is used (with interim frequency checks) followed by a final lap of 1000 to 1200 grain carborundum abrasive, used in a water-soap solution. Cleanliness is very critical in that even fingerprints on the blank can cause erratic operation. A soap cleaner is more easily removed from the blank surface than is a hydrocarbon cleaner such as carbon-TET. The pressure plates should also be clean. Ultrasonic cleaning can fracture the crystal blank, particularly if there are any crystalline fractures on the edges of the blank.

What is the effect of DRIVE LEVEL? In general, a higher drive level will raise the temperature of the blank (bulk resistance increases with drive level) and will tend to raise the frequency slightly. For good stability, minimize the crystal drive level, and, if a higher signal level is needed, obtain the level (plus some buffering) by adding an amplifier stage following the oscillator stage. The drive level of the crystal can be determined from the RF voltage across the crystal and the effective series resistance, (Rx). A vector voltmeter will do the job nicely (and include the phase) but few Hams have access to this piece of equipment. Crystals in vacuum tube circuits typically run at the 2-10 mW level.

Xtal Mode Typ Freq, Mhz

Old Xtals

Typ Freq, Mhz

New Xtals

Typ Res, ohms Drive Level, Max
Fund 15 To 30 20-50 10mW
3 o/t 10-45 10-75 2-20 1-2 mW
5 o/t 45-75 50-150 40 500 uW
7 o/t 75-105 100-200 60 50 uW

Table 1 Typical Crystal Parameters

Click here to see Fig. 6 - Crystal Filter

At f (well below resonance) the reactance of the crystal is capacitive and C2 is adjusted to that same reactance, (including the capacitance of the crystal holder). Thus the output signal through the crystal is equal in amplitude and opposite in phase (due to the CT Xfrmr) to that out of C2. The signals combine and cancel—no output. At circuit resonance, of), the reactance of the crystal and C2 are equal, but the crystal reactance is inductive. The crystal signal is shifted +90 while that through C2 is shifted by 90 . Thus both signals add at the output (signal peaked). Mechanical filters are limited to about 1 Mhz or less while crystal filters can be built up to a frequency of 250 Mhz.

A table of information on Selected Military Crystals is available.

Take care of that rock and it will take care of you!

Job Position Wanted

Warty Drift, WB2FOU wishes to relocate, and is looking for an engineering technician or electronics technician job in the states of PA, Ohio, N.Y. State, VA, NH or VT. If any AM P/X reader knows of a job opening or can provide a job lead as described above, please contact Warty at (817) 4976023.

Coming Next Issue

We are still seeking articles from our readers on any aspect of AM, technical or non-technical. Next month we will continue W2VRL’s series on audio pre-emphasis and filtering, and we plan to run profiles of two well known AM'ers in our "Meet the AM'ers" section. W4KYL will report on the Ft. Lauderdale, FL hamfest he attended on 6-7 February, and we will give a full report on the Dayton Hamvention including the An Forum. If you send it in now, your article or photograph may appear in the next issue.

The "B" Formulas

Readers have inquired about W2WLB’s continuing series "A Significant Readability Improvement on your AM Rig by Simple Pre-emphasis and De-emphasis." Each letter asked for clarification of the term "B Formulas", referred to in Part II (AM P/X # 103) and in Part IV (this issue).

These formulas are very simple. They cover grid inputs, cathode bypasses, screen bypasses, hot path bypasses, power supply output loads, stage decoupling networks, pre-emphasis circuits and modulated stage bypassings. They are used to determine the optimum values of resistors, loads, and capacitors for all the audio circuits in AM receivers and transmitters. This includes power supplies, filters, detectors, etc., anywhere audio may be involved. With them, we can optimize every circuit affecting modulation, including the modulated r.f. stage.

In these formulas we multiply the resistance, R, in ohms, times the capacitance, C, in microfarads, to equal near a value we need for particular functions. For example, a 10,000 ohm resistance with a 10 Mfd. capacitor would multiply to equal "100,000 B". "B" is simply a term coined by George to mean "RC". "1B" equals one ohm-microfarad.

The significance of the B Formulas will be further clarified by the excerpts which include schematic diagrams, reprinted below from an earlier article by George (April, 1981 Press Exchange) entitled *On AM Being superior to SSB - Formulas for Maximum Available Readability" (c)1981 by George Bonadio.

In Figure 11 we have our 100v power supply delivering a peak of only 10 ma., or 1/100 amp. From ohm’s law we find that 100v at 1/100 amp equals a Z, or load of 10,000 ohms. To satisfy a need of 100,000 B + or - 40%, we need 10 Mfd for C, 40%, that is, 6 to 14 Mfd. If we use less than 6 Mfd. we may have too much 120 Hz hum left. If this supply feeds a modulated stage, and we use smaller than the -40% value, some of the audio modulation low frequencies will develop across the power supply components C and L, instead of onto the modulated stage. Likewise, excessive residual hum left across C will show up on the air. If, however we make C much greater, above +40%, over 14 Mfd in Figure 11, we may over strain severely, a close fuse, a conservative diode or a hard working tube rectifier every time we turn the power on. I moved my Johnson Kilowatt, high voltage supply, from a 48,000 B to 130,000B. Now if I start in the high power position, the 20 amp circuit breakers drop. I have to start on half voltage, for a one second surge, then I can snap up to full power.

Now let’s look at the 100,000 B formula. It is for power supply outputs and far the bypassing of interstage decoupling resistors and for cleaning the d c. going into screens and in grid biasings. See Figure 12, Some of R1 will be in series with R2 so that a true B value will be higher, R2 may not be in the circuit, especially if the grids do draw grid current. Then only part of R1 is involved, so, then C2 must be larger. If any of bleeder R1 is used, tapped down from the top, there must be a C2 to fit 100,000 B, even though the manufacture did not out one in. Likewise, often the manufacturer used a very small screen C3 when the B+ came from a sliding A tap. Estimate or measure the screen tap resistance (always pull the power plug out first) from the tap on the resistor to chassis, and use that omage for your R3 value, to calculate your C3 needs. Too small values in these three places do allow "bouncing biases: a prime cause of "muddiness" in modulation, as well as somewhat more modulation trash above and below the regular sidebands. Stop making your own interference, See TABLE 100,000 B. It is for d.c. Its tolerance is a wide + or - 40%. The values shown will multiply to approximately 100,000B, using the standard number values. For changes of decimal points, any change on one side must be the opposite on the opposite number. Thus, you could oppositely change the decimals of 300 x 330 to 33 x 3,300 or to 3.3 x 33,000 and so on.

Click here to see the schematic

Use the TABLE 100,000 B to check all your power supplies, all your decoupling, all your d.c. filters, all your screen supplies. Write the present values on one diagram. If they are below -40%, below 60,000 B, then increase them to between 60,000 B and 140,000 B. If they are above1+40%, above 140,000B, recheck your figures and measurements. Manufacturers usually stopped at about 25,000 B in these circuits. Too much can lead to excessive fuse sizes, to motorboating, to too slow starts, and to wasted costs.



By George A. H. Bonadio, W2WLR 373 East Avenue, Watertown NY 13601-3829 A Series on Amplitude Modulation Improvements

Part IV


When we start to connect several stages to the same power supply we need to feed the DC through "decoupling networks". These are only one R and one C on each stage’s B+ lead. We see this in Figure 5. The first two stages are on the same B+ lead and separated by their R’s & C’s. Because those designs were for a steep roll off of the low audio tones, they did not need as much filtering as we need.

We need to prevent the gain of the first tube, which is in phase with the third tube, to develop on the DC line. There can be over 20 dB (100 X the W) of gain. If over 100th of that returns via the B+, we can have "motorboating", an oscillation of a few cycles per second.

Our filtering B formula applies here, just as it does, were the first tube with a screen R & C, as lightly sketched in. In each case the R X C = 100,000 B, +/- 33%. None of the sets I have studied has near that amount of filtering. You will so much improve your buss response that you will need to increase B here. Add enough C, to the C that is there, to satisfy the above B formula.

Usually the first two cathodes have C’s which are in a can remote from the heat of the cathode. However, by the third tube your cathode bypass C may be hung onto the hot cathode lead where it had deteriorated. Even when it reads the right capacity on a C tester, its series R, from aging heat, may be over 10 W . When I added 10 W to a good cathode C, the gain loss was significant on a 12BY7. Use new electrolytics or tantalums, but connect them remotely and away from heat. On cathodes try to use voltage rating new C’s which are at least twice the DC level expected or measured.


Two of the old popular sets have diode clippers, either tubes or solid state, with no easy means to set them to correct levels. Our compressor is going to do away with that problem. In practice you would need a scope to observe the trapezoid form, and modulate at the plate current that gives you the most output. This is the same current on any band. The RF efficiency varies, on different bands, but the maximum output, conveniently, always comes to the same DC level. Different DC levels would need a clipper readjustment. Of course, you should find, at what plate current that your SWR Forward meter indicates your maximum RF output. Try to load to near that current at each band tune-up.

After the diode clippers there is a Low Pass Filter (LPF) which is there to stop you from radiating much of the 2nd, 3rd, 4th, 5th, etc., harmonics of your clipped low and stronger tones. This clipping is done in square wave shapes, which will contain dozens of noticeable harmonics. You may leave the filter in if you readjust the C values to allow 3,200 Hz to pass undiminished, while reducing higher frequencies.

Clippers always put out trash sidebands, and for the fraction of time that they clip, they blot out any communications intelligence. Meanwhile, the modulator tubes, while clipping, are drawing their maximum current, wasting their lives, overheating the rest of the set and bothering neighbors on the spectrum. There is a one stage filter, which does little to heal the wounds.

As you bypass any clipper, remove any newly unused tube from its socket. Make any needed adjustments in B levels left there.


Particularly the DX 100 has either a bad match or not enough iron in some of its transformers. This occurs in several bargain priced sets. The usual source of difficulty is in the first audio transformer. Apparently the DC saturates the core enough that the low audio frequencies, which need more core, do run out of linearity.

On a scope you can see a nice lower audio frequency sine wave going into a set, but a bent sine wave coming out. When you approach 100 % modulation the distortion is severe. The harmonics of these distorted low tones are up into the more sensitive hearing ranges, producing "muddiness" which isn’t good AM.

There is some relief. A cathode resistor on the tube driving the transformer, may be increased by up to 50 % in ohmage. The tube current is reduced about by half this ratio, reducing the percentage of remaining distortion. There is room for a simple inverse feedback, which will directly reduce the ratio of distortion again.

If you have the usual gain to spare, you can tie an R between that tube’s plate and its driving tube’s plate. Both are at high voltages. No C is needed. Cheap! Try values between 220,000 & and 1 MW , with lower values doing more but costing more of the abundant original gain. If there is still distortion visible, on the deep tones, at high modulation levels, then have one stage of B value at less than 5,000 B. See line J of Figure 1. The DX 100 actually used 2 such stages, with small transformers, to produce line A.

I do not recommend changes of transformers. If you do insert a low pass audio filter, using toroids, within the rig, it is easiest put in between the plate and the screen to transformer B+ lead. The inverse feedback, apparently, is not spoiling this action, as I tried it, successfully, in my DX 100. More on filters, later.


Transformers are heavily coupled inductors. The coupling is so tight that usually we can ignore the L and the C in the device, and just think in terms of, say, "Z" to "z". However, when we start loading changes with extra C’s across the Z’s & z’s, our tubes start to see reactances in with their Z load lines. Then the load line becomes an ellipse. In a heavily driven stage, being pushed near its maximum output, this ellipse may cause extra distortion, especially if it cuts off current. These cut off are frequently a large part of the cause of "overmodulation way up and down the band" complaints. AM can’t afford this.

Fortunately, if we can estimate the Z, apparent load, of a winding, we can use the B (is less than) l0, formula. I need to warn you that many circuits were aimed to be lulling off’ some dB’s at only 1,000 Hz. This could be B values of, say, 40. If we are testing with a frequency response test (later) we can bring out test leads from that C, while passing 3,200 Hz.

If that B is over 10, adding an equal value C will reduce the output by more than one more dB. The replacement should be a C that reduces a part of a dB, but less than 1. The Ranger has one of those C’s on the secondary of its first audio transformer. It is used to control its inverse feedback loop, so all of it may not be removed.

In that same stage two triodes are in parallel, in one tube. That changes the net results of the Miller Effect, from about 30 pF to about 60 pF. With the drive being 470,000W , they went down to 150,000W , on the grids, as their escape. It passes.


When we get to the final modulation transformer we find the sources of much of our trash from overmodulation. The excess C bypassing is severe. What happened was that we needed a plate - to - tank C to pass the RF from a hot wire to a probably grounded tank (L&C). Then we needed a DC return circuit from which we needed to bypass excessive RF. Sometimes we used more than one RF choke and several "at least big enough" C’s.

After these designs, along came TV which brought TVI into being. Well, we must bypass that bad TVI, so we over bypassed 21 MHz TV IF with large C’s while TV changed to 42 MHz IF. Anti-TVI designers put in rows of .002’s(2n’s) and .005’s (5n’s) when 220 pFs (220p), on short leads, are more effective against modern TVI.

Meanwhile, the audio engineer needed to slope his audio down from 1,000 Hz, so he loaded a big C (the Ranger’s and DX 100’s .02, or 20n) across the secondary of the transformer, directly. All of these other C’s are also directly across the transformer secondary Z. When tank, RF filtering, TVI and Audio Shaping engineers each load down one Z with their choices, we get problems.

The Ranger final RF stage operates at about 500 V at 125 mA, or 4,000 W Z. All the above amount to about .03 X 4,000 = 120 B, when we would like to hold it down to as near to 10 B as possible. First, we can remove the .02, so we are down to 40 B. We should not change the .002 to the tank, for RF reasons. The other .002’s can be reduced to .00022 uF (220p) at an operating V of 1,000 or more (1KV). So, now we have reached B= 11.52, which we can still live with, reluctantly. If we used 330’s instead of 220’s we would have only B = 13.28, only once in the whole audio. No one will tell us about it.


Practically every RF final stage has been a tetrode, with a screen to be modulated from the plate modulation energy line, as a minor load thereto. In some large, KW finals, audio chokes have effectively produced a modulation effect upon those screens. However, the screen bypass C’s were designed for the lowest RF, in complete disregard to what would happen to 3,200 Hz energy there.

The net result is that the screen modulating treble audio going through the series screen resistor eventually is significantly bypassed to G by the size of the RF screen bypass C. The popular Ranger, again, uses 30,000 W to a C of .002 (2n). The B value should not reach 10 B. 30,000 X .002 = 60 B, a severe error. This has the treble on the screen out of phase with the treble on the plate, while the lower tones are substantially in phase.

Besides producing nonlinearity of modulation, distortion, it produces trash out beyond the normal sidebands, splatter, even before we hit 100 % modulation. We know that we can not change the 30,000 W R much without complications. All we can do seems to be to reduce the C by 6 or 7 to 1, say to 330 pF. But that is too little for screen RF bypassing at 1.8 MHz.

There is another way. We can take this load, which is only 13 % of the load made by the final, and bypass it enough, across the R of 30,000 W , so that the bypassing of that resistor has the same B value as the C at the screen has across its apparent Z.

Let us assume that the supply is 500 V. The screen is operating at 200 V. That leaves 300 V across the 30,000 W . Ohm’s Law tells us that there is 10 mA through 30,000 W to produce 300 V. This is the same A as the screen is drawing with the 200 V which are left. This suggests that the screen is acting like a 20,000 W R, but with a C of .002 (2,000 pF) across it. Then we multiply our 20,000 X .002 = 40 B. Now we find the C which, across R will = 40 B. This is 40 + 30,000 = .0013 uF (1,333 pF), using a .001(ln) paralleled with a .00033 or 330 pF, in the Ranger.

There is a similar screen problem in most of the manufactured rigs, including the DX 100 where we find treble modulation lost also in the RF grid return. This modulation of the grid is termed Self-Modulation, and comes from everything else swinging wildly.

The DX 100 RF grids, on the diagram, drain via 47 MMF, an RF choke of a few ohms, .005, then a meter calibrated R of 5.4 W , past another .005 through 2,200 W to the bias supply C. These add up to 2,210 W bypassed by .010047 uF, rounded to 2,200 X .01 = 22 B. Replacing the .005’s with .001’s will give us an R = 4.4, which is fine, under 10 B. After placing ceramic capacitors across the DX 100 two 10 K R’s, by formula B, the final will modulate treble tones much better. Rut the 4.4 B changes to 24.2 B when the meter, with its two .005’s are switched into the grid reading. Changing those to .001’s lets us have 8.8 R, fine.


The nature of speech is that it is very irregular, very much unlike a steady tone. However, some transmitter engineers seem to not have thoroughly checked out their sets with voice signals. If they would have put a scope to each bypassed cathode they would have seen the ones which had too little C across then. They would have seen bounces in the negative biases. Even the power supplies bounce when the audio modulation approaches 100 %.

This is for your own proof of performance. With a scope, get used to how your audio looks with a sweep of around 120 Hz. Often you can see repeating patterns. Now move your probe up from the first cathode to watch for speech effects on it bias though the single ended stages. Now go to the centertap on the first audio transformer secondary, and talk. Do not be surprised to see negative diving voltages coming from both positive and negative audio peaks.

A little rectified audio grid current, through a resistance to G, such as through the bias voltage supply, can muddy up your quality. Placing another equal C across any C there, may smooth down its intensity, to about half. If there is a great difference, the original C is needing a replacement. If there is no C directly connected to the centertap, but there is an R there, then use the filtering B formula of R = 100,000 33%, in placing a large C there.

Mentioned much earlier is the abandonment of regulation on the Ranger’s V on the screens of the two modulator tubes. With full audio neither a single tone, nor two tones) each screen will draw more current on the plus side of its wave. Thus, we see, on the scope, negative dives of V from each of both sides of each audio wave. This is rough frequency doubling.

As the gain of the tubes do change with the V of each moment of each screen, we can see the problem. Very undesirable distortions, of many harmonics, are generated. Much of the intensity of the even multiples of harmonics are balanced out, but this still leaves high levels of 3rd, 5th, 7th and 9th harmonics. Is this why the huge .02 uF 1.6 kV capacitor was put on the secondary of that modulation transformer? To shunt much of that trash away? The cause is on the diagram. Each screen comes out through 100 W , anti-parasitic R’s, to join with a C of .1 (100n), and up the wire with a .005 (5n). That is it. The source is a tap on the 20,000 W bleeder. If that tap is midway, the two halves are effectively in parallel to G, which gives us about 5,000 W Z. This is bypassed by these two C’s, which total .105 uF (105n). Then, rounded, 5,000 X .1 = 500 B when we need about 100,000! We need about 800 times as much. Thus, I used 22 uF and 33 uF would still fit. There is more undistorted audio output, and there is less trash up and down the band.


If we have too little C in the output of our power supplies our low audio frequencies will develop some of their output across the power supply filter L&C. Even if we use separate supplies for the RF and for the audio, the bass can still show up undistorted in the Final’s supply, and distorted into the audio’s supply.

Of course, the filter B formula applies. 100,000 B 33 %. The DX 100, on one high V supply, peaks around 333 mA, at 700 V. This is a load Z of about 2,100 W . This ends into two C’s of 125 uF each, in series, equaling 62.5 uF. So we see 100 X 62.5 = 131,250 B, which is excellent. However, the Ranger, at 500 V X 250 mA = 2,000 W and has only 10 uF there, or 20,000 B. It needs a minimum of 39 uF, up to a maximum of 75 uF.

Two, each, C’s, both new, of the same batch, can be put in series, if each is shunted with a 100,000 W W. As little as two of 80 to an ideal of 100, to a maximum of 150 uF’s, in series, is optimum. Look for the available room before you buy yours. I ended up with two 80’s, each at 450 V.

Electrolytic capacitors are not designed to operate as input C’s of power supplies. They can not long stand up to a full voltage range across them at 120 times per second. Hence we should not be too surprised, on the DX 100, to find that the filter input C’s of both low voltage supplies are testing very poorly.

I had very low bias voltage from that supply with 20 uF, 1K, 20 uF. Placing a paper C of only 1 uF, across the input C, raised the voltage to almost normal. MY replacements were both 50 uF at a voltage of more than twice the normal output. Then I placed the 1 uF across the input C to lengthen the electrolytic’s life.

With the low voltage supply, I compromised replacing the output 40 uF with a 100, at 450 V. The input value of 20 uF needs to be the same, to prevent over surging the rectifier, or changing the resultant V., however, at 450 V rating. As the bias voltage changes many other operations, no operating values should be assumed until the power supplies are fully corrected.


Our sets are precious. We cannot easily replace the ones which we have. There are great destroyers, time and heat. When the sets were new, quiet fans were too new to be built in, but not any more. Force some air through, preferably having cool air in at the bottom and hot air out at the top.

If you (I would not do it) were to overload your rig by 26 % over its 100 % normal power, and not get lower efficiency, you could reach 126 % of your former output, which is only + 1 dB. We cannot detect a difference of one dB in speech. Even 2 dB is often unnoticed. Run cool, run long. It takes + 6 dB, or 400 % of the power to make one "S" unit.

To Be Continued

KA2PYQ Running for FCC Commissioner

Fred Jody, KA2PYQ, Demurest, NJ, an AM enthusiast and subscriber to The AM Press/exchange, has announced that he is running for FCC Commissioner to fill one of the current vacancies on the Commission. Fred believes he has the engineering background that would qualify him for the position, and he is launching a grassroots campaign to get himself on the list of persons to be considered for the appointment.

Fred points out that in recent years, the commissioners have been primarily lawyers instead of engineers, and that the country has suffered as a result of having non-technically-trained legal specialists attempting to regulate the technical aspects of electronic telecommunications. This has resulted in rules drawn up by personnel hired as career civil servants, and rubber-stamped by the commissioners, who routinely defer to the "expertise" of these hired bureaucrats, even when their decisions are demonstrably based on faulty and sometimes arrogantly deceptive logic.

Fred is asking the amateur radio community to write to the President, the Vice President, and representatives in Congress, suggesting that he be given consideration for this important post. Talk this over on the air to let other amateurs know.

This is the AM PRESS:

An amateur radio publication dedicated to Amplitude Modulation.

This is the AM EXCHANGE:

Offering FREE ADVERTISING to enhance the availability of AM equipment and parts.

DISPLAY ADVERTISING: Rates upon request.

Edited and published by Donald Chester, K4KYV

NOTICE; The purpose of this publication is the advancement of Amplitude Modulation in the Amateur Radio Service, and there is no pecuniary Interest. Therefore, permission is hereby expressed for the use of material contained herein without permission of the publisher, with the exception of specifically copyrighted articles, provided that The AM Press/Exchange is properly credited.