Xtal Size = Xtal Activity?

[W1UJR]

Hello all,

Here is a question for the group.
Does xtal size have anything to do with xtal output and oscillator stability?
I've been working on a 30K-4 which uses xtals in the onboard exciter.
The xtal holders are an odd affair, not like our FT-243 holders, but larger, about 2" x3" with with three pins.
I'd prefer to keep the unit rockbound on 1885 and 1945, then maybe later build a small switch-box to switch between two different xtals.

The problem that I am having is the local oscillator seems to stop and start without rhyme or reason.
I removed the xtal from one of the large holders, and installed a rock out of a FT-243 holder.
Just for kicks I tested it with one with a homebrew TX which I use, its oscillates just fine, stable, and within 500 cycles of the indicated freq.
Now if I install that same rock and holder in the 30K, the unit does not run consistently.
I've tried cleaning the xtal and holder, still same results.
I've also "JSed" other xtals into the socket, still in their FT-243 holder, and the unit seems to be stable on the lower bands like 80 and 40, but still unstable on the 160 meter band.

What I suspect is that the exciter needs to see a certain amount of xtal activity or it will not run, does this make sense?
In reading online, I came across certain xtal manufacturers stating they need to know the capacitance expected in the oscillator.
This kind of puzzled me, but I found that reference more than once.
Is this a reference to internal capacitance between the xtal plates?
And if so, can this be "padded' with an external cap?

Anything else which I have overlooked here?

Thanks!


73 Bruce W1UJR

[AB2EZ]

Bruce

The equivalent circuit of a crystal is given in a number of places, e.g.

http://www.ecliptek.com/crystals/glossary.html  [Note, in particular, Equation 2 and Figure 5]

and

http://www.scapro.se/text/jauchkristallara.pdf [ Note, in particular, fig.8 and equation 7]

It is a very high Q tuned circuit, that will look like an inductance at frequencies slightly higher than its "series resonant" frequency.

When the crystal is in series with (or in parallel with) a capacitive load, the oscillator will oscillate at a frequency at which the inductive reactance of the crystal exactly equals the capacitive reactance of the load... thus producing a series (or parallel) resonance.

Note that a crystal in parallel with an external capacitor can be thought of in either of two ways:

a) the crystal and the capacitor form a loop, which corresponds to the crystal and the capacitor in series... and therefore forming a series resonant circuit (loop) through which current flows

b) the crystal and the capacitor are in parallel, forming a parallel resonant circuit... which has a large circulating current, but very little net current.

Either way, there is current going through the crystal at resonance

Therefore, to ensure that the oscillator will oscillate at a particular, desired frequency, the crystal must be manufactured taking into account the capacitive load it will be in series with (or in parallel with) in the circuit in which it will be used.

Likewise, you can change the frequency of oscillation ("pull" the crystal's frequency of oscillation), somewhat, by putting more or less capacitance in series with (or across) the crystal.

Normally, the capacitance across the crystal will consist of: the capacitance of the holder, the socket, and the associated wiring, inter-electrode capacitances of the tube, plus any capacitance that is intentionally placed across the crystal as part of the oscillator circuit design.

With respect to the observed intermittent operation:

When the crystal is operating in an oscillator, there is an rf current flowing through the crystal... whose size (r.f. milliamperes) depends upon the design of the oscillator. This current, if too large, will cause the crystal to vibrate too vigorously and/or to overheat... possibly damaging the crystal, or possibly just causing the oscillator to stop oscillating. Older crystals were designed to be capable of handling higher rf currents. Many crystals that come in FT-243 holders cannot handle such high rf currents.

There are a number of possible fixes for this including: redesigning the oscillator circuit to reduce the rf current that will flow through the crystal (i.e., in some cases, reducing the value of the grid leak resistor in parallel with the crystal will accomplish this), placing some additional resistance (around 100 or 200 ohms) in series with the crystal, building a separate crystal oscillator, and plugging it into the existing crystal socket (while opening the feedback loop in the existing crystal oscillator)...

Also, it is possible for the oscillator tube to have too low a transconductance... and that could make the operation of the oscillator intermittent... so you might try replacing the oscillator tube with a known-good tube.

Best regards
Stu



 

[W1UJR]

So what you are saying then Stu, is that "size does matter"? 

All kidding aside, thanks for a most through and helpful treatment of the subject.
I can see that my attempts to "pad out" the xtal may have been in vain.
Its funny how it will run sometimes for 15-20 minutes with no issue, but at other times stop oscillation in a few seconds.

I'm back into the rig this evening, and plan to actually sub in one of the original xtals in the holder, something which I did not try.
When I first obtained the rig, I just gutted the holders and installed the FT-243 sized xtals, thinking all would be fine.
If the oscillator is stable on the original xtals, then I may try your suggestion of adding a series resistor to the xtal, or perhaps having the proper size blank commercially cut and ground.

Of course, I could just break down and drive the unit with an outboard exciter. In the real world, a small Icom 703 or 706 would be a excellent, and stable VFO, and perhaps more practical. On the other hand, I've been greatly enjoying the engineering challenge of making it work!


73 Bruce W1UJR

[k4kyv]

I don't think the physical size of the crystal matters so much, but the activity of the crystal can make a difference.  Maybe the 30K-4 oscillator circuit has insufficient feedback to maintain stable operation of the crystal. Check the oscillator tube, and its associated components.

My HF-300 rig is capable of xtal control.  It uses the old style Bliley or Valpey type crystals with the round holder about the size of a half-dollar, and two pins that fit into pins 2 and 4 of a standard 5-pin tube socket.  I have noticed that when I acquire those xtals at hamfests, etc, very often they either wouldn't oscillate at all, or are extremely unstable and flaky.  I would take the xtal/holder assembly apart and carefully wash the crystal and the metal plates that the xtal is sandwiched between, using denatured alcohol, and then wipe them with a clean, soft, lent-free cotton cloth.  90% of the time that would do the trick and the xtal oscillated normally.  Sometimes the xtal plate would be obviously damaged, such as having a visible crack through it, or there would be be a discoloration that would not wash away.  A crystal can be just plain bad.

The xtal oscillator in my transmitter uses the classic pentode circuit, in which a tube, primarily intended for use as the audio output stage of a broadcast receiver, is used for the oscillator tube.  The crystal is in parallel with a grid leak resistor, inserted between grid and ground.  The tuned circuit goes between the plate of the tube and +HV.  The plate circuit is tuned for the most stable note, not the greatest output.  The feedback to the crystal is accomplished by taking advantage of the less-than-perfect shielding between the plate and grid, typical of screen-grid audio output tubes.  Typical tubes for this circuit would be the type 41, 42, 47, 59, 2A5, 6K6 or 6F6.  Beam power tubes like the 6V6, 6AQ5 and 6L6 will also work.  A well shielded tube, such as the 6AG7, 5763, 802, 12BY7 or 807 may not work in this circuit because the shielding between grid and plate is too complete.  But it is easy to add a small capacitor from grid to plate to give it enough feedback that oscillation will start.  Usually a "gimmick"  capacitor, a couple of short pieces of hookup wire twisted together, is all it takes.

So I would try cleaning the xtal and checking the oscillator circuit for bad components.  If worse comes to worse, maybe the circuit could be modified to increase rf feedback to the xtal.  It might be wise to temporarily insert a #47 pilot lamp in series with the xtal.  If the bulb shines brightly, there may be too much xtal current.  This can eventually lead to failure of the xtal. If  the bulb doesn't glow at all, there is too little xtal current.  It should glow a dull red or orange.

Most of the time I use a highly modified T-368 master oscillator unit as an external VFO.  But I do like to run the old xtals for the fun of it every now and again.  Unfortunately I don't have xtals for popular AM operating frequencies.  Maybe I'll try my hand at grinding a few to move them.  I have successfully moved xtals from other types of holders to the round types that fit my transmitter, and they worked just fine.

[Tom WA3KLR]

HI Bruce,

From what you have said, the crystals and holders have been cleaned and you tried other blanks in the holders.  So by now you should have been running with active crystals, but you still have the unreliable oscillator.  It appears that the problem is in the oscillator circuit, not inactive crystals.

Most oscillator circuits in HF gear are parallel resonance circuits.  The manufacturer needs to know the load capacitance for frequency correlation only.  The “20 pf. load” crystal and “32 pf. load” crystal are no made any different, except are on slightly different frequencies. 

Adding more capacitance across the crystal in a parallel resonant circuit pushes the parallel resonant lower and towards the series resonant frequency.  A small amount of this can be tolerated in a healthy oscillator, but ultimately the crystal‘s effective Q is killed.   So in a circuit that has oscillating problems, padding is one thing you don’t want to do.  Another thing you don’t want to do is add series resistance to the crystal.  This also lower the Q.

You didn’t say that you heard the crystal frequency drift rapidly after the oscillator started.  If this happens, it is the sure sign of the crystal over-dissipating energy and the accompanying temperature rise; the crystal frequency is temperature dependent.  So again I presume that the crystal is not being over-driven and probably wouldn’t be with those old gigantic holders.

The blank’s thickness is what determines the frequency and the wider blank does allows more power dissipation. Since the mass is greater for a “bigger” crystal at the same frequency I could presume that the phase noise of the output signal could be lower.  But this is not a concern here.  That’s not why they were made big back then.

Recently I started to use crystals in my Viking II and had identical symptoms to you.  Several crystals were restored to reliable operation by cleaning, but one crystal still did not work after cleaning the blank and holder.  I had to make a modification to the circuit eventually after checking a number of things.

The circuit may have been marginal design to start with and I researched oscillator circuits in other Johnson transmitters which was a help.  I also wonder about subtle aging of the tube socket, crystal socket, and coil forms which would insidiously de-Q the loop gain of the circuit.

Hopefully this gives you some ideas.  If some simple reasons for the problem such as the tube, low B+, screen voltage or filament voltage is not the problem then you have to start thinking in terms of gain around the oscillator loop.  My simple mod. to the Viking II created a little more feedback and that did the trick.

We need to see the oscillator schematic; camera photo is the easiest way.

[W1UJR]

Thanks Don and Tom, let me tell you I had to take a break from the xtal problem today, just for sanity sake.

First, I like your idea about a light bulb in circuit Don, just to be certain the xtal is not getting blasted.
I should also address your suggestions about cleaning the rock.
I did wipe down the xtal blank with acetone, and inspected it over a work light, no cracks or blemishes noted.

Tom, the unit stays rock solid, does not drift, so overheating does not seem to be a problem.
It may very well be that the something is up with the exciter, I've had it in and out of the cabinet perhaps a total of 5 or 6 times checking things, all seems well. I did replace the oscillator tube, but it was not a new one, so perhaps I should also double-check that.
I'll have to run out to my office, where I keep my tube stock, to see what I have on hand.

Funny thing is that the unit is rock solid when it works, but seems to have no rhyme or reason when the crap-out occurs. It might happen in the first 5 seconds of keying, or it might work 10 minutes, I've tried both into a big dummy load. Perhaps what I really should do is keep the scope on while testing, checking to see what is happening with the waveform, that might be the key. Odd thing is that the grid drive never drops gradually, like the oscillator was having a problem running, it stays solid until it decides to just shut off.

Since I don't have a 160 meter VFO, my plan tomorrow is to run the solid state rig, a Icom 706, into a dummy load at say 1-3 watts, then tap off the dummy load into the input of the xtal sockets. Just to see if it will keep running with a constant signal. If that works, I'll dig deeper into the exciter deck, perhaps follow your lead Tom and do what I need to so it will work.

I took a break from the oscillator problem this evening and have spent most of the evening building up a matching audio and T/R box for the rig, so now, at 2AM its time for bed. I'm leaving the freshly painted chassis in the oven overnight to bake on the paint.

Thanks again for the tips fellows, I hope to have Bessie on the air tomorrow evening, either rockbound or via the Icom silicone slider.



73 Bruce W1UJR

[KA1ZGC]

This may sound a bit too obvious, but have you checked the socket?

My repeater kept suffering random transmitter dropouts that we could never understand, and the only common denominator was that it would happen after I had the cover off to do some work on it.

Turns out the xtal socket just wasn't mating with the pins particularly effectively, and now and again would lose oscillation. My thumping around the chassis would usually shift things just enough for the xtal connection to go intermittent.

Tim eventually found the problem, the socket was tightened up, and they all lived happily ever after.

Hey, it's worth a shot.

--Thom
Keep Away One Zero Grip Crystal

[AB2EZ]

Thom

I must admit that you make a good suggestion. I had a problem with my Viking Ranger in which one of the two crystals that were plugged in would not oscillate. I took the Ranger out of its cabinet, I checked the wires, I measured voltages, I studied the schematic... and I changed the oscillator tube. Finally, I pulled the crystal out of the socket (one more time) and vigorously cleaned the bottom of the holder and the pins with contact cleaner. The problem went away... and has not returned. To this day I do not know if the pins were dirty, or if there was a hairlike conductor between the pins that was causing a short circuit.

Separately, I've been thinking about the question of why a crystal (the holder plus the quartz crystal inside) would be reluctant to oscillate, if the quartz crystal itself weren't cracked or otherwise permanently damaged...

If the quartz crystal were floating free, between the two electrodes, it would respond to the random noise produced by the oscillator... and would begin to vibrate... which would, in turn, start the oscillator... and cause it to reach it's saturation voltage.

However, the quartz crystal is not floating free between the two electrodes. It is pressed between the two electrodes within the crystal holder. As a result, there would be some amount of static friction that must be overcome to get the quartz crystal to start vibrating. Once it starts vibrating, the amplitude of the vibration would build up, and it would keep vibrating until something caused it to stop. For example, if the quartz crystal heated up enough (due to a combination of many factors, including mechanical losses associated with small amounts of friction within the crystal holder), it might do something like the following: melt some debris inside the crystal holder... which would flow into the space between the crystal and the electrodes. Also, if the quartz crystal were thinner (or thicker) than the holder was designed to be used with, then if the quartz crystal is vibrating with too large an amplitude, it might become wedged in a position between the electrodes that causes it to abruptly stop vibrating. Likewise, if the quartz crystal is vibrating, it might dislodge some very tiny piece of debris that would become wedged between the crystal and the electrodes.
 

This might explain Don's comments about bringing a crystal back to life by opening it up and carefully cleaning the quartz crystal surfaces and the electrodes inside the holder.

By the way... the "activity" of a crystal is, as I understand the term's use in our applications, simply a measure of how much initial electrical noise must be applied to the crystal to kick start it into vibration. If the quartz crystal flexes more in response to a given applied voltage (applied to the electrodes), then it will be able to overcome the initial static friction with less of a "kick". If the surfaces of the quartz crystal are dirty, then it will take more of a "kick" to get it started... and probably result in more heating of the crystal once it starts vibrating.

Best regards
Stu

[WA1GFZ]

clean your crystals in good quality vodka.....That is how the old timers made sure they were clean.  Check the resistor and cap values in the circuit as well as the tube emission. over loaded crystals tend to start hard while underloaded will be off frequency. Old rocks usually age down in frequency. fc

[WU2D]

Ah Crudmugeric Crystiction

I have also moved the blanks from larger down to FT-243 and it worked fine.

Have you tried a swap putting the blank in another large holder, just to see if it settles down?

How about trying another oscillator tube?

Mike WU2D

[Ian VK3KRI]

By the way... the "activity" of a crystal is, as I understand the term's use in our applications, simply a measure of how much initial electrical noise must be applied to the crystal to kick start it into vibration. If the quartz crystal flexes more in response to a given applied voltage (applied to the electrodes), then it will be able to overcome the initial static friction with less of a "kick". If the surfaces of the quartz crystal are dirty, then it will take more of a "kick" to get it started... and probably result in more heating of the crystal once it starts vibrating.

I always figured the 'activity' was more to do with the  Q of the crystal, caused by how well it was ground in the first place and how much 'gunk' is on the surface damping it.   A question that arises out of this is how far does a HF xtal actually move?  I've never given it much thought but I'm guessing given the size of FT243 blanks, and the frequency , it can't be much, maybe  only a few atoms diameters?
 
                          Ian VK4KRI

[AB2EZ]

Ian

I checked some references available on the web. I found that the parameter that describes how much the crystal actually moves is "b_xy" where x and y specify directions relative to the axis of the crystal. It has the units of meters/volt (i.e. strain in meters per meter / field strength in volts per meter), and it is around 10**-11 for quartz. Thus, as you suggest, with a few volts across the crystal, the amplitude of the displacement, if the displacement is in the direction of the electric field, is only around 10**-10 meters ... which is on the order of the size of an atom. If the electric field runs across the thickness of the crystal blank (e.g., the blank is 0.1 mm thick), and the strain occurs in a direction that is perpendicular to the field (e.g. the blank is 20 mm wide) then the displacement is larger by the ratio of (in this example) 20mm / 0.1mm = 200. So for a crystal oscillating in that mode, the displacement would be about 2 x 10**-8 meters... which is still very small.

I found this excellent reference... and I copied the following from it:

From: http://www.ieee-uffc.org/freqcontrol/quartz/vig/vigqrtz.htm

"As the drive level (the current through a crystal) increases, the crystal's amplitude of vibration also increases, and the effects due to the nonlinearities of quartz become more pronounced. Among the many properties that depend on the drive level are: resonance frequency, motional resistance R1, phase-noise, and frequency vs. temperature anomalies (called activity dips), which are discussed in another section of this report. The drive-level dependence of the resonance frequency, called the amplitude-frequency effect, is illustrated in Figure 8 [10]. The frequency change with drive level is proportional to the square of the drive current; the coefficient depends on resonator design [11]. Because of the drive-level dependence of frequency, the highest stability oscillators usually contain some form of automatic level control in order to minimize frequency changes due to oscillator circuitry changes. At high drive levels, the nonlinear effects also result in an increase in the resistance [5]. Crystals can also exhibit anomalously high starting resistance when the crystal surfaces possess such imperfections as scratches and particulate contamination. Under such conditions, the resistance at low drive levels can be high enough for an oscillator to be unable to start when power is applied. The drive level dependence of resistance is illustrated in Figure 9. In addition to the nonlinear effects, a high drive level can also cause a frequency change due to a temperature increase caused by the energy dissipation in the active area of the resonator."

Note the highlighted sentence.

Best regards
Stu

[Tom WA3KLR]

In the basic model of a crystal which is a series resonant RLC circuit, the inductance called the "motional inductance" is usually in the region of Henries. 

The motional capacitance is something like 0.01 picofarad and the the resistance is in the order of 15 - 100 Ohms. 

The Q is the motional inductance's reactance over the series resistance, so yes the term "activity" is another term for Q essentially and ultimately refers to the series resistance.  Typical Q's in a good crystal are 45,000 to 75,000.

Today most crystals made are in the hermetic holders with metal deposited on the sides of the blank and a spot-welded wire coming off of the face.  So this is much more reliable, probably cheaper, takes less space, and allows the final frequency tweak by playing with the plating rather than etching the quartz.  The series resistance contact-dependence problem is gone.