[time-nuts] Making a HP 10811 better

Bob Camp lists at rtty.us
Tue Mar 30 23:46:14 UTC 2010


On Mar 30, 2010, at 5:06 PM, Magnus Danielson wrote:

> Bob Camp wrote:
>> Hi
>> 
>> There are a couple of problems you will run into.
>> 
>> The first is that if you do a very good job with the dewar flask, the "dead
>> heat" (power of the oscillator, regulator etc) can indeed raise the
>> temperature of the OCXO beyond the oven control temperature.
>>  
> I.e. there needs to be a leak into the surrounding, otherwise the heaters will turn totally off without achieving the chill-down effect, so it will almost behave like a bang-bang system with quick ramps in temperature and then long effects of unregulated cooling down.
> 
> So, the regulator needs the chill-load to create a heating balance.
What you do is heat the plug, and keep the energy on the "inside" of the flask very low. The basic idea is that everything past the plug is iso-thermal. 
>> The second is that the gain of the control loop is in part determined by the
>> thermal resistance. Increase the thermal resistance by a very large amount
>> and the control loop gain goes way up. The control loop gain is probably
>> pretty high already, so increasing it by a large amount is likely to make it
>> unstable.
>>  
> In moderate levels it could be good thought... no?
If it was optimum before the change you will make things worse.
>> The next thing is that control loop gain and effective thermal gain to the
>> crystal are not the same thing. Even if it is stable, a large increase in
>> control loop gain probably will decrease the effective thermal gain. The
>> reason is a bit complex. The simple answer is that the thermistor is not
>> mounted on the crystal blank, thus it "sees" something different than the
>> crystal.
>>  
> Do elaborate. I am curious. I am aware of thermal gradients and issues with geometry.
This gets a little long winded .... sorry about that.

First consider the junctions of the heater transistors relative to the heated "shell" of the oven. As ambient goes down, more heat is needed and the junctions get hotter relative to the shell. They are going backwards (ambient goes down and they go up).  Somewhere between the outside world and the heater junctions, there's a thermistor. It drives the control amplifier. If you have an enormous amount of DC gain the thermistor does not change temperature. Stuff in the area of the heaters gets "to hot" stuff closer to the outside world than the thermistor gets "to cold". 

A simple way to look at the 10811 is that the crystal is a bit towards the "hot" side relative to the thermistor and you play with heat balance to peg it there. If you had a different geometry you could get things closer. In this case it's fixed already. The other thing you are doing with balance on the 10811 is dropping the gradients across the crystal. The net result is that you balance the offset of the crystal against the finite gain of the controller. By doing the balance "just right" you can get very high thermal gain to the crystal.

The high effective gain can be achieved by moving the thermistor with fixed gain and heaters. You can fix the thermistor and gain and move the heaters to the job done. Change any one of them and you are likely to mess things up. Not all combinations of any two can be "fixed" by varying the third, but you see some pretty amazing combinations. 

----------------

If you dig into Rick Karlquist's 1997 FCS paper there's some good stuff on just why effective thermal gain only gets you so far with something built like a 10811.

>> With the large increase in gain, since it's a simple controller, you will
>> need to re-set the control point. There's no handy integrator in there to
>> crank the offset out for you. 
>> Simply put, you design a dewar flask / vacuum bottle OCXO in a very
>> different way than one that's conventionally heated. Miliwatts of
>> dissipation do mater in a vacuum design.   
> OSA 8600 and 8607 has one side of the flask still open. There is a plug there, but it isn't near as good isolator as the flask. It needs to leak overflow heat that way. Consider that on these most of the electronics sits on the side of the flask, so only the actual oscillator core and heating transistors is inside the flask.
If you put much more in the flask, you would never get to a rational upper end temperature.
> 
> The 10811 has more stuff inside it.
>> Reducing drafts is a good thing. Moderating ambient variation hour to hour
>> is a good thing. Burying an OCXO in the back yard works, dunking it in a big
>> barrel of water also works. Both require you to remember the waterproof bag
>> before you take the final step :)....
> Oh, we always toss our oscillators into the water when powered, didn't you know? ;) That's why we throw a towel at it.
Hop out of the pool and grab a towel, what could be more normal?
> 
> Reducing *quick* changes in ambient temperature or breeze, is what comes cheap and does not significantly changes the equation.
> 
> Cheers,
> Magnus
> 
> _______________________________________________
> 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.
> 




More information about the time-nuts mailing list