Rotten Signals: How To Cure Them
A Talk About the Essentials of Transmitter Tuning

by George Grammer
Reprinted from QST for April, 1933 with permission of The ARRL

   George's article was aimed at helping the fraternity eliminate chirp and "r.a.c." (raw a.c., I presume) signals from their self-exited rigs. As the builder of a couple of Bruce Kelley 1929 QSO party Hartleys, I can vouch for the fact that George's very complete treatment addresses most of the questions that arise when trying to get one of these rigs to work well and sound good.

   The article was called to my attention by Bill Beckett, W2NBK, who sent me a good photocopy. It seemed quite appropriate to run it in this issue--where we are reporting the results of the December "party" (see the Amateur Radio Column). The article is reprinted here almost exactly as it appeared in QST, except that standard OTB typography is used and additional paragraph breaks have been inserted for easier reading. --MFE

   Fully half the letters intended for QST's Correspondence Department recently have dealt vehemently with rotten notes, especially of the 40- and 20-meter variety. In almost every case a self-excited transmitter is responsible--probably in most cases a transmitter using something bigger than a 210. Now a raspy r.a.c. note may be all right for the fellow behind the key; he doesn't have to listen to it. But it's all wrong for the other 90% who take some pride in putting out a clean signal; theirs are the ears that suffer. When the day comes that we all have single-signal receivers the rotten note problem will cease to exist--that type of signal is unreadable on a really selective receiver and loses most of its power to cause QRM. But right now the "mud-like r.a.c." is on the increase.

   The principles of correct transmitter adjustment were established long years back. After doing some listening to things flying around in the ether these days we began to wonder whether the old rules had lost their effectiveness. As a matter of curiosity we built a typical self-excited transmitter, using a tube which by all indications ought to be one of the least satisfactory at high frequencies, and rediscovered, with much satisfaction, that the old precepts were quite up to snuff. Evidently the answer is that they're not being applied.

What's in a Transmitter?    A self-excited transmitter has four divisions: an oscillator tube and circuit, a power supply, an antenna system and a monitor. Some people have tried to get along with only the first three with results which are evident. The fourth is just as essential to a good transmitter as any of the others. It doesn't matter much what kind of monitor it is--it may simply be the station receiver if the latter is well enough shielded--so long as the oscillator in it is steady and capable of giving a beat with the transmitter. Therefore the first recommendation is: Get a monitor. Obviously, it's impossible to tell whether a change has made any improvement if there is no way of listening to the transmitter.

Circuits    In performance, the different circuits--Hartley, tuned-plate tuned-grid, and so on--are practically identical. Whether or not the results are good is entirely a matter of how the circuit is handled. There are only two objects in the adjustments one makes to a transmitter - power output and frequency stability. The last is just as important as the first.

   Frequency stability is not a matter of a single adjustment or a single feature in the transmitter. There are at least four ways in which instability can get into a transmitter. The first is through changes in frequency caused by changes in plate voltage, or dynamic instability. If we plot a curve of oscillator frequency against changes in plate voltage it will be found that as the voltage is increased from zero there will be a continuous frequency change until the final voltage is reached. The extent of this change is a measure of the dynamic instability of the transmitter.

   On a small transmitter with but 500 volts on the plate the change in frequency can be 20 kilocycles or more at 7000 kc. if the set is poorly designed and incorrectly operated. Naturally it will be worse if the plate supply is 1000 volts or more. With a poorly-filtered plate supply the note from such a transmitter is going to be r.a.c. hash, because the frequency will be flitting gaily back and forth at the plate-supply ripple frequency.

   The remedy for dynamic instability is to be found in the use of a large ratio of capacity to inductance in the oscillator circuit, and particularly in the plate tank circuit if the grid and plate are separately tuned. In other words, High C. This does not necessarily mean an inordinately large tuning condenser. We know that few hams operating 203-As and 852s have high-voltage variable condensers with a maximum capacity of 500 mmfd. But nearly all of them have 220- or 250-mmfd condensers, and those condensers are big enough provided the coils are cut so that the band is hit with the condenser plates practically all the way in. Paring down the coil is the first step toward eliminating dynamic instability.

Other Considerations    Once the coil has been cut down so that about 200 mmfd can be used to tune it, attention should be given to the excitation and the grid leak. Both--and they are not independent of each other--have far more to do with the final condition of the note than most amateurs realize. If the tube is to work at reasonable efficiency it must have high bias, which in turn calls for a high-resistance leak and plenty of excitation. The dynamic stability is improved under the same conditions. The excitation must be adjusted with a load on the oscillator. The setting which gives the least plate current when the oscillator is not delivering power to an external circuit is invariably the wrong one; the excitation will be insufficient under these conditions and the stability will suffer.

   As a general rule, the no-load plate current should be at least half the load plate current, although this depends somewhat on the frequency. The excitation is increased in the Hartley circuit by moving the filament tap nearer the plate end; in the t.p.t.g. by increasing the capacity of the grid tuning condenser. Since the adjustment is critical it is best to make changes in very small steps, watching the input and output and listening in the monitor. Listen especially to the character of the signal when it is keyed. If the circuit is High-C a key chirp is an almost certain indication of insufficient excitation, with the exceptions noted later on.

   Use the highest value of grid leak resistance that will permit the tube to oscillate stably with normal input--between 10,000 and 20,000 ohms for a single tube, usually. Too much leak is just as undesirable as too little. And remember that the higher the leak resistance the greater is the excitation voltage required, so every time the leak resistance is changed there must be a corresponding change in the excitation tap or condenser setting.

Efficiency    A high C-to-L ratio brings with it large circulating currents in the tank circuit, hence the tuned circuit leads should be short and of heavy conductor. If the oscillator coils are plug-in, it is also necessary to be sure that the joints make good contact and have low resistance. Coils may be bolted in place or heavy plugs and jacks, lately made available for transmitting coils, may be used.

   It is necessary at this point to make a distinction between tube efficiency and circuit efficiency. If the tube is running normal plate current under load and is correctly biased and excited, its efficiency will be as high as in any other circuit regardless of the L-C ratio. The circuit efficiency will be somewhat lower in a High-C circuit, however, because of the greater losses caused by the higher tank current.

   The distinction is important because it is necessary for the tube itself to operate at high efficiency if it is to stay cool in operation. And it is highly desirable for the heat dissipation in the tube to be well within the ratings because heat makes the tube elements to expand, which in turn causes the inter-electrode capacities to change. Since the inter-electrode capacities are unavoidably a part of the circuit, here results the second cause of instability--a slow frequency change or "drift."

   The greater the condenser capacity in the tuned circuit, the less will changes in tube capacity affect the frequency. Therefore the same things which give good dynamic stability will also minimize frequency drift--High C, a high-resistance grid leak, and correct excitation. In addition, the tube must not be overloaded. The plate should never show signs of color even when the tube is allowed to oscillate continuously for minutes or even hours at a time.

   Frequency drift also can be caused by heating of the tank coil and condenser, another reason why the resistance in the tuned circuit must be low. Drift from this cause will be less when power is being taken from the circuit because the tank current decreases with increasing load.

Mechanical Instability    When the electrical features have been taken into account there remains the third cause of poor notes--lack of attention to mechanical details. Cleaning up dynamic instability and drifting are not in themselves the guarantee of a good note. It is when these things have been done that the smaller--but nevertheless just as serious--causes of instability become apparent. Mechanical vibration of coils, tube elements and condenser plates can utterly wreck an otherwise pure d.c. signal. It's no job at all to pick out "mushy d.c. signals" on the air, which, when the key is held down for more than a few seconds, change into good d.c. And all because the key is mounted right alongside the transmitter on the operating table and the whole transmitter does a shimmy whenever a little brasspounding is in order.

   Self-excited transmitters should never be placed where they pick up every vibration set up by the operator's movements. Put the set on a separate table, jack it up with sponge rubber, hang it with shock-proof cord, anything you like--but protect it from vibration.

   It should be unnecessary to mention that the oscillator itself should be solidly constructed. Short, heavy leads, parts fastened securely so they cannot shake; in fact, every care that can be taken to prevent floppiness is worth while. Furthermore, it's not hard to do and costs nothing.

   One excellent way to turn a d.c. note into r.a.c. is to build the oscillator and power supply as one unit. Even the quietest of filter chokes and transformers will vibrate, and when the whole works is rigidly mounted on one frame or baseboard, the vibration is transmitted very efficiently to the oscillator tube and the tuned circuit. The power supply should be put off by itself, mounted on some sponge rubber or felt if necessary. If it is impossible to get a pure d.c. note with a power supply which by all the rules has adequate filtering, the chances are excellent that vibration is responsible.

Antenna and Feeders    After all these things have been corrected there is still the fourth possible cause of instability--frequency warbles caused by a swinging antenna or feeders. As soon as the oscillator is coupled to an antenna or feeder system and power is taken out, the antenna or feeders become a part of the tuned circuit. A Hertz antenna suspended well in the clear can swing a great deal before there is much effect on the frequency, but the feeders are another story altogether.

   Zepp feeders especially are likely to be bad offenders, because the wires are relatively close together and hence have fairly large capacity to each other, so that if they swing back and forth the oscillator frequency may change considerably. For this reason the feeder wires should be spaced at least ten or twelve inches and should be liberally supplied with light-weight spacers.

   With light spacers the whole feeder system tends to swing as a unit in a wind, but heavy spacers, because of their greater inertia, cause the wires to whip back and forth. The antenna and feeders should be pulled up tight, of course.

   Some capacity coupling always exists between the oscillator coil and the antenna coupling coil, especially when the coupling is tight. This capacity coupling is no value in transferring power to the output circuit, but does help along the harmonic output and makes the oscillator particularly susceptible to capacity changes in swinging feeders. Therefore it is advisable to couple to the "cold" end of the tank because the r.f. voltage is low and little energy is transferred through capacity coupling. The antenna will take power just as readily as when the coupling is to the plate end.

Some Other Things    Even after these four causes of instability have been given the right kind of attention, it is still possible to have a modulated and chirpy signal if the r.f. is not kept where it belongs. Chokes sometimes do not do their duty as they should, allowing r.f. to get into the power supply, which certainly is not helpful.

   Sectional slot-wound chokes, sectional honeycombs, and plain single-layer chokes all are good when they have low distributed capacity and have no resonance spots near the amateur bands. A neon bulb is still about as good as ever for testing a choke. If the bulb glows when touched to the supposedly "cold " side of the choke some experimenting is in order. A poor choke decreases the efficiency as well as sometimes being responsible for a poor note.

   But r.f. in the power supply is not always a sign of a poor choke. Direct pick-up of r.f. from the oscillator circuit is often more than just a possibility. Power supply leads to the tube should be kept out of the r.f. field as far as possible. Shielding the leads when they come near the oscillator is worth while if the shielding is connected back to the filament center-tap at the filament by-pass condensers.

   A ground on the center-tap is advisable if it is short. At 14,000 kc., however, a ground lead sometimes does more harm than good if it approaches a quarter wave in length--a matter of but sixteen feet. The only way to find out is to try it, using the neon bulb and monitor.

   Under certain conditions there may be no sign of r.f. in the supply leads with the antenna disconnected, but it appears immediately when the oscillator is delivering power. This is the result of pick-up from the coupling system. It is particularly likely to occur with parallel feeder tuning because of the high voltage at the transmitter end of the feeders with that tuning system. If this happens, move the supply leads out of the field of the antenna coil and feeders or add enough length to the feeders so that series tuning can be used. With series tuning the voltage at the coupling coil and at the transmitter end of the feeders will be low and the electrostatic coupling consequently less.

   This brings up the subject of antenna coupling. The right degree of coupling is almost automatically settled if the tube is not going to be overloaded. Simply use the loosest coupling that will make the tube draw its rated plate current with the antenna or feeders tuned to resonance. The character of the note and the keying should be checked in the monitor when the antenna coupling is being adjusted. So long as the monitor says the note is good the coupling may be increased and more power taken from oscillator, for after all it is the signal that is the important thing and not the method of getting it. Put the antenna coil at the " cold " end of the tank for close coupling. Take as much out of the set as you can--but use the monitor constantly.

Power Supply    The remaining member of our four transmitter divisions is the power supply. A good power supply--well filtered, and having good regulation--is always desirable with any transmitter; it is absolutely essential with a self-excited set. A poorly-filtered supply cannot possibly. produce a d.c. note. And if the regulation is poor the signal is bound to be chirpy, because in spite of all the things we may do to improve the' dynamic stability of the oscillator there will still be some frequency change with changes in plate voltage.

   Use a separate filament transformer so that keying will not affect the filament voltage. There is only one way to get good regulation in the plate supply--use a filter with choke input and put in a bleeder which will drain off about 10% of the total current to be taken from the system. Besides having the advantage of good regulation a choke-input filter decreases the peak rectifier current, which increases the life of the rectifier tubes and makes it possible to draw considerably more output current from the transformer and rectifier than is possible with condenser input.

   The only disadvantage is that the output voltage is lower than with condenser input at light loads, a disadvantage that rapidly disappears as the load increases and which in the end turns out to be helpful because it cuts out a lot of the strain on the filter condensers. If you have only one filter choke put it next to the rectifier tubes and parallel all your filter condensers on the output side. The filtering will be just about as effective either way. And don't forget the bleeder.

   None of the things we have pointed out above are hard to do, in fact, it is no trick at all to get a thoroughly satisfactory note simply by paying attention to details. Possibly it may be necessary to sacrifice a little power output to get a good signal, but that is of little consequence. It takes a big change in power to make an appreciable difference in signal strength, and the small amount that is used up in the interests of having a 1933 signal will never be missed. Neither will the r.a.c. hash the good signal replaces.

 BACK TO CONTENTS