Getting crystals to oscillate on their nominal frequency

[k4kyv]

I recently acquired a hand full of crystals for some of the popular 75m AM frequencies.  I noticed that nearly every one oscillated 300-400 Hz higher than their nominal frequency.  So I checked some of my other crystals and noticed the same thing.  That led me to believe the problem was with the oscillator circuit, not the crystals.

The oscillator is a standard pentode circuit using an old buzzard broadcast radio AF power amplifier tube, taking advantage of the poor internal shielding in those tubes to provide the feedback necessary for oscillation. There is no trimmer capacitor across the crystal to adjust the exact frequency.  The 47, 59, 41, 42, 6F6, 6K6, 2A5, 6V6 and other similar tubes will work about equally well in this circuit.

Since it is a simple straight forward circuit there was little to change.  But I decided to experiment with the components directly associated with the crystal. The circuit is a standard one shown in all the ARRL and Radio handbooks from the 30s through the early 60s. There is 2.5 mHy rf choke in series with a 50K grid leak resistor and the resistor goes to  ground.  The  crystal goes from grid to ground. There is a tuned circuit at the plate, which is coupled to the next stage through a coupling capacitor, and a dropping resistor  supplies about 100 volts to the screen, which is by-passed with a capacitor.

I noticed some of the circuits shown in the handbooks use the series rf choke with the grid leak, while others omit it.  I could find nothing in the text regarding the rf choke, but its only purpose I could think of is to prevent the grid leak resistor from loading down the crystal.  I shorted out the rf choke using a clip lead.  BINGO! The oscillator shifted  down to within 50~ of the nominal  frequency. A couple even oscillated slightly below nominal. I tried numerous crystals on 80, 160 and 40m, and each time, the rf choke brought the  frequency up several hundred Hz while shorting it out brought the frequency back down.  In no case did omitting the rf choke cause a crystal to stop oscillating.

I permanently removed the rf choke and now most of my crystals actually oscillate very close to the frequency stamped on the holder.  From what I could determine, the rf choke serves no useful purpose, but tends to pull the crystals off frequency in a positive direction.

[W3RSW]

Interesting Don.  I wonder if something similar hasn't been written before, at least from the superfluous RF choke 'sucking down power' standpoint.  I've notice that the later tube osc. literature omitted RF chokes in series with the grid resistor.  You certainly seem to have 'nailed it' on the freq. changing aspect.

I always thought that since xtals were in osc. circuits (that produced some power compared to a receiver amp, etc.) anyway, that a 47k load instead of several hundred k up didn't hurt much when noticing that the more recent literature eliminated RF chokes in the grid.  I think all the 50's handbooks had eliminated the choke in osc. circuits.

[flintstone mop]

I had a heck of a time with that. Mode seems to be in there somewhere in the brain cells

Fred

[KM1H]

Crystal control was still a new marvel, to most hams anyway, when that circuit came out. The old handbooks have a few pages devoted to the different circuits.

On another forum we were discussing that very circuit a few days ago with a "what the heck were they thinking" tone.

My own theory is that the early 30's resistors were fairly inductive and loaded the crystals down. One circuit had just a choke for the grid bias!

Use a #49 bulb in series with the xtal as both a fuse and a current monitor. Under some conditions current can be too much for the FT-243 style as the plate is tuned thru resonance.

Carl
KM1H

[k4kyv]

When that circuit was first developed, very few hams probably noticed if the xtal was a half kHz off frequency.  Some of the old ads offer crystals, guaranteed +/- 5 kHz of the requested frequency.  For more $$$ they offered them +l- 1 kHz.

The 1930's handbooks showed some circuits with the rf choke and others without it, but I have never seen an explanation of what the choke was for. Usually, the more theoretical circuit diagrams that go with the explanation of howthe circuit works, omit the crystal, while the practical how-to-build-it diagrams for an actual transmitter include the choke.  In the ARRL handbooks circa 1950, the rf choke was always there.  In later versions, late 50's and early 60's I recall the choke had disappeared from the circuit.

Something I do notice with mine is that without the choke, as you tune the plate circuit through resonance, the dip resembles behaves more like that of an oscillator working into a load, even if there is no load.  But the load the resistor presents to the crystal doesn't keep any of my crystals from oscillating nor does it significantly reduce the rf output from the stage.

[W3GMS]

Don,
I would sometimes use a small amount inductance in series with the high side of the xtal going to the input of a given oscillator circuit.  At times it was quite useful to net a xtal on frequency since it has the tendency to make the xtal a bit "rubbery".  Sometimes as xtal age they slide off frequency and its tuff to bring them back without adding the inductance.  In other cases with direct FM modulators it sometimes helped with producing the correct amount of frequency deviation.   
Joe, W3GMS 

[WD5JKO]

Don,

   Your analysis is interesting. I bet the old resistors had a lot of shunt capacitance, so the choke would keep the crystal seeing a mostly resistive/inductive load besides the class c grid current pulsing. So I wonder how the circuit would behave with a vintage resistor only? It might be fun to play with old resistors from the 30's, 40's, and 50's with an impedance meter to see how much shunt capacitance a vintage resistor like 47K has.

    They must have added that choke for a reason. Maybe it was a carryover from an idea that was not needed as you say. I wonder however if actual commercially made gear used that choke? With commercially made gear, every part had to be justified by an engineer being grilled by a room full of bean counters. Think of the "All American 5" receiver where there were few/none unnecessary components.

   I have been looking at the 9 Mhz oscillator in the Central Electronics 20a transmitter. This oscillator drifts quite a bit during the first hour as the rig warms up. Most of the drift seems to be the crystal since the warmup characteristic changes a lot with a change in the crystal. I do wonder though, going back to your point, if the oscillator stability is better/worse/same with and without the RF choke present in series with the grid resistor?

Jim
WD5JKO

[k4kyv]

IIRC, my 1957 handbook still shows the rf choke in some circuits.  All  the circuits in the 1950  handbook do.  By then, resistors were mostly carbon composition.

Carbon comps probably have less inductance than some of the modern day film resistors, where the film is in the form of a  strip deposited in a spiral round the insulative form.  Even the old tip-and-dot resistors from the 30's were carbon comps.  The only difference was the way the leads were attached.  The older ones had the wire wrapped round the ends of the carbon body and bonded, while the later ones had the leads somehow inserted into the ends of the body and bonded, covered with a protective layer of insulating material.  Maybe the earliest "dog bone" resistors were more inductive.

Regarding crystals, I have quite a  collection of the old round holder ones with pins that fit a 5-pin tube socket.  The ones that operate on 160 and 80m seem to be mostly within a kc or so of their nominal frequency.  But most of the 40m ones have drifted quite a distance away.  I tested out some the other day, and they had even drifted off the frequency I observed and recorded when I first acquired them and tested them out.  A data sheet from Bliley crystals says that a different holder configuration is (was) required for 40m crystals than for 80/160.   I  suppose in those  days, 40m crystals were still practically in the experimental stage and more flaky than lower frequency ones.  Most of the old ham rigs shown in the handbooks used 80m crystals and doubled to get to 40.

I have several of the old style crystal holders with a HC6/U inside, much like in the recent Electric Radio article.  I picked them up at a hamfest, all marked with an identical nominal frequency, somewhere in the 12 mHz range.  I opened one up and found the equivalent of a HC6/U with wire leads installed inside.  This looked to be a factory job, not a Hammy Hombone modification. 

I ordered some HC6/U's from AF4K for some of the popular 75m AM frequencies that I didn't have.  I cut the pins on the HC6/U's and soldered wire leads, and used them to replace the original 12 mHz ones in the round holders, and they work just FB OM.

I normally  run with a VFO, but it's fun to sometimes insert "crapstals" and run the rig crystal controlled.  I gave Derb my spare VFO, a highly modified VF-1 that I hadn't used for decades, so I don't have a replacement VFO to use when mine craps out till I  can get it going again.  But I have enough crystals to cover my most often used frequencies until I could fix the problem.  So maybe "crapstal" would be an appropriate term to use for mine. 

[KM1H]

From what I can see looking thru ARRL pubs is that they were obsessed with using that choke.

In the Radio Handbooks it was usually eliminated and the only reference is in the 1935 edition that says it can be eliminated if a carbon or other non inductive resistor is used. This seems to confirm my guess yesterday about the real early resistors being somewhat inductive.

As with many other circuits the ARRL rarely updated as better components became available or gave explanations for certain choices.  Maybe the original QST circuits went into more depth.

I still recommend a 60 ma bulb be used as a current indicator and even try with and without the choke and see which one results in less current.

Carl
KM1H

[DMOD]

I wonder however if actual commercially made gear used that choke?

I have a schematic of a Sintronics AM-1-K (1 kW BC rig) that I used to work on. [Would love to have that puppy right now]  This circa 1958 trx has two 4-400's in parallel modulated by two 4-400's in Push-Pull.


It doesn't use any chokes in the 12BY7A OSC grid. There is a 100k grid leak resistor with a variable and fixed capacitor in parallel with the crystal to ground.

There is a 2.5 mHy choke but is in series with the cathode and is used to keep RF out of the metering circuit.

Phil - AC0OB

[k4kyv]

How is a HV spike going to get to the xtal upon startup if it is connected from control grid to ground, with a 50K grid leak in parallel with the xtal, and no additional feedback or bias circuitry is connected to the grid?

I have seen fancier circuits that connect one side of the xtal to the screen or plate, but the beauty of the basic pentode  circuit is its simplicity, using the stray capacitance between the poorly screened control grid and plate in BCL audio output tubes for feedback, so no external feedback loop with rf chokes and capacitors is needed.  Xtal current should be very low except maybe if you are trying to get tens of watts out of a 6L6, since the stray capacitance is still very low.  Triodes can easily shatter xtals due to the high grid-plate capacitance, so more care is needed.

Tubes like the 5763 and 6AG7 usually won't work without a small external feedback capacitor, because those tubes are very well shielded at radio frequencies.

[flintstone mop]

Check this link for a little peak of a typical tube osc circuit.
Not near any high voltage.

Fred

http://faculty.frostburg.edu/phys/latta/ee/6cl6xmtr/6cl6schematic.html#ecoscillator

HEY we're on the same channel c aaaawmawn

[k4kyv]

Is it the circuit below?

That's the circuit I use.  The oscillator is turned on by switching on the power supply to the oscillator.  It drives an 802 buffer, which drives the 211 which drives the pair of HF-300's. To run CW, I cathode key the 802 stage.  I had to neutralise it to  get rid of a backwave when I key it on 40m.  I tried an 807, but it was very squirrelly and also required neutralisation for the same reason.  The 802 is much more stable.

Taking out the rf choke shown in the diagram in series with the grid resistor is what brought the oscillating frequency to nominal.

The tuned circuit at the plate has to be tuned slightly above the resonant dip at the operating frequency, so that the plate of the tube sees a slightly inductive reactance to get reliable, stable oscillation. This also reduces rf crystal current.

[WU2D]

Don,

The choke decouples the grid leak resistor no doubt, but why? Could it help poor crystals oscillate?  Does it help keying or chirp? See how it effects the turn on time and keying. Perhaps it is just a leftover from the old days when you keyed the oscillator. The Paraset 6V6 oscillator uses the choke too...

Mike WU2D

[k4kyv]

An incandescent lamp would have the opposite thermal characteristic to what is needed to do that.  When the filament is cold, the resistance is very low, close to a dead short.  When it heats up to full brilliance the resistance increases to its normal operating value.

The same is true with vacuum tubes.  That's why we have the inrush current problem that may shorten the life of tubes.

I suspect regular incandescent lamps would last a lot longer if some kind of slow start could be applied to them to allow them to warm up gradually without excessive inrush current.

I looked through my 1935 Radio handbook and I did see one reference to the rf choke besides the apparently random presence and absence of one in the diagrams.  They said that if a choke is used, the grid leak resistor can be reduced to as low as 20K.  That seems to confirm my theory that the purpose of the choke is to prevent the grid leak resistor from loading down the crystal, and that the choke is optional.

I'm going to try making a temporary adaptor to insert a small pilot lamp in series with the crystal just to see how much it  lights up, and whether or not I need to worry about too much crystal current.  I have used some of my crystals for years and have never been concerned about excessive current since I run only about 250 volts on the plate of the pentode and 100v on the screen, enough to drive the 802 buffer, probably less than 1 watt rf output from the xtal oscillator stage.

One indicator of too much xtal current is when the frequency substantially drifts for the first few seconds after the oscillator is switched on.

[k4kyv]

With a 50k resistor in series, you would have to have over 1000 steady volts just to get an incandescent pilot lamp to barely glow.

[KM1H]

Is it the circuit below?

I would either use a lamp like Carl suggested to act as sort of ballast or change the value of R1 completely.

The #49 lamp was a pretty common recommendation to save crystals from being fracturered.

The circuit I dont understand is when they eliminate the grid resistor and just hang a 2.5 mh choke to ground. For DC thats like using a 40-60 Ohm resistor.

However the circuit as used by Don was used in the very first xtal oscillator article in QST, way back in July 1924. The only difference being a triode. Back then you were given a hunk of quartz and had to make your own crystal....and that was real iffy. That went on for several years before packaged rocks were offered....at a pretty high price.

A 6L6 can easily crack a FT-243 crystal and a tri-tet circuit can crack all but the hardiest of the older style as currents can be in the 70-100 ma region.

A 6AG7 or the later 6CL6 is arguably the best xtal oscillator tube for transmitting as far as current, heating, and stability are concerned.

I never built the RCA 813 circuit but Ive used BC-610 rocks on a 6146 and 6550 many decades ago with no damage.

Carl
KM1H

[k4kyv]

I have used HC6/U xtals with a plug-in adaptor in my rig and they worked just fine.  I'm not so sure I would trust those little ones similar to HC6/U but only about 1/2" X 1/2" though.

With pentodes and tetrodes you can shatter crystals when you try to run power with the oscillator.  With triodes, it is necessary to be very careful when running any oscillator circuit.  I have always avoided the tri-tet like the plague.

I never could see any point in running a high power (20-100 watts) one tube crystal oscillator transmitter.  The additional components and circuitry are too few and simple not to run a low power crystal oscillator stage driving a separate PA.  Besides, the separate PA is usually more efficient than a one-tube circuit.

[KM1H]

Using a high power oscillator was more of an experiment than anything useful for on the air. It was in the early 60's mostly aboard ship as a way to relieve boredom.

The last time I seriously used a crystal oscillator for transmitting was in 1955-1956 as a Novice. I havent had any use for them since except in receivers and low level transmitter stages to drive mixers. Building a stable VFO was more fun and worthwhile

Carl
KM1H