[time-nuts] The 5MHz Sweet Spot

[Bob Camp]

HI

If you doubled the diameter of the blank each time you cut the frequency in half, all sorts of nice things might happen. If you start with a 1/2” blank in at 10 MHz that goes to 1” at 5 MHz and 2” at 2.5 MHz. Around 1 MHz you would get to a 5” blank.

Good luck finding high grade quartz bars to cut 5” (or even 1”) blanks out of. You are going to have to go back to the autoclave fixtures at the very least. Since growth is (at best) linear you cost of quartz will scale with the size of the blank. I’d bet it scales a bit more than that if you want to keep the material at a high level of performance. 

Then you need to cut it / lap it / polish it. All of that gear scales with blank size. 

That’s the easy part.

Now you need to build a cold weld package that will accept your 5” blank. Then you need a proper press to seal it. The ones for the little blanks come in around $2-$3M each. First one you make (with all the back and forth) probably costs you 2X that to develop. Figure the cost will scale with the size of the package.

That’s at least straightforward. Its just money.

If you have a 5” blank, your OCXO scales around it. Yes you can do some neat things, but the package is getting bigger. You now need to convince people to buy OCXO’s that are bigger than anything they have seen since the 1960’s. In most cases that OCXO will be 2 to 10 times larger than their entire sub-system. That’s going to be a tough sell.

No customers = no money to pay for all the fun stuff. 

Bob


On Nov 1, 2013, at 10:56 PM, Richard (Rick) Karlquist <richard at karlquist.com> wrote:

> On 11/1/2013 7:12 PM, Tom Knox wrote:
>> A while ago I mentioned 5MHz oscillators were used in most metrology applications compared to the more commonly available 10MHz because 5MHz was a sweet spot for quartz. At the time I didn't know why. I finally had a chance to ask the person I learned this from why. The main reason is simply physical size. The larger crystal lattice allows many manufacturing advantages that allow for a higher Q. He also explained I was wrong in an earlier statement, metal/quartz migration on quartz oscillator was not a major problem even after decades, but could become more of a factor if driven hard. That does not mean the deposition and lead bonding has no negative effect. The BVA solves this by capacitive coupling the quartz rather then direct metal deposition.
>> 
>> Thomas Knox
>> 
> 
> A lot of issues conflated together here.
> 
> 1.  There is a theoretical QF product for quartz.  Being at 5 MHz basically doubles your Q, all other things being equal.
> 
> 2. Having a higher Q reduces the contribution of the sustaining
> amplifier, but only within the 3 dB bandwidth.  With the Q being
> in the millions, this is only a few Hz.
> 
> 3.  In general, the sustaining amplifier is not a player in
> a well designed quartz oscillator in the first place.
> 
> 4.  Q probably has a negative correlation with flicker noise,
> meaning higher Q is associated with lower flicker noise.
> However, the correlation is not strong.  There is no theory
> that says that Q puts a bound on flicker noise.
> 
> 5.  So that leaves us with the larger physical size.  Perhaps
> it allows higher Q, but again it is unclear how this is connected
> with flicker noise.
> 
> 6.  You didn't mention the theory that more total atoms of quartz
> provides averaging flicker noise over a large population.
> 
> 7.  You didn't mention the notion that larger physical size permits
> higher drive level.  Since the Q is also large, perhaps it doesn't.
> Also, a higher drive level is probably only going to help with
> far out noise.
> 
> 8.  Many, or maybe most, 5 MHz resonators are made with undersized
> blanks which are enabled by energy trapping.  So we don't have a
> simple scaling of all 3 dimensions.  What is the effect of this
> "cheating"?
> 
> If someone can shed additional light on this, please jump in
> and educate us.
> 
> Rick Karlquist N6RK
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.