[time-nuts] How long do ovens take to cool to ambient after power is removed?

Tom Van Baak tvb at LeapSecond.com
Thu Oct 2 14:58:35 EDT 2014

The most extreme example of slow ovenized oscillator warm-up I've seen is the vintage hp106. These mid-1960's oscillators were designed as the ultimate, "hp way", pre-atomic, frequency standard -- expected to be powered up, uninterrupted, for years and decades. So there was no hurry in the (perhaps once-in-a-lifetime) initial warm-up. Here's a plot/photo of one I recently tested:


These HP-106 oscillators are among the best I have ever measured: stability and daily drift rates in the very low -13's. Like the SR-71, these were designed by gut and slide rule. And yet achieved extreme performance, even by today's standards.

The amazing thing -- as you know from your enviable career at HP -- is that an instrument produced in 1964 can still work 50 years later in 2014. No blown fuses, no electrolytics, no filaments, no f/w upgrades, no Y2K, no decaying EEPROM, no batteries, not even any IC's. No user s/w, no USB, no drivers, no OS. Not even an on/off switch! Just a 5-pin 24VDC backup or 2-prong AC cord in and a pure 5 MHz BNC out, that's all.

How many of the instruments we use today will still work out-of-the-box in 2064?


----- Original Message ----- 
From: "Richard (Rick) Karlquist" <richard at karlquist.com>
To: "Discussion of precise time and frequency measurement" <time-nuts at febo.com>
Cc: <hmurray at megapathdsl.net>
Sent: Wednesday, October 01, 2014 9:03 PM
Subject: Re: [time-nuts] How long do ovens take to cool to ambient after power is removed?

> On 10/1/2014 1:04 PM, Hal Murray wrote:
>> drkirkby at kirkbymicrowave.co.uk said:
>>> Anyway, later today (tomorrow ??) I will post a plot of frequency vs time.
>>> The question is though, how long is thing thing likely to take too cool?
>> I'd expect an exponential decay so you need to specify how close to ambient
>> you want to get.   I'd guess a ballpark of 10x the warm up rate.
>> You can probably measure it if you have the warmup graph.  Turn it off, wait
>> a while, turn it on, measure the freq, consult warmup graph.
> When I was still with Agilent, I did some experiments with unpowered 
> 10811's.  Both the oven and oscillator were unpowered and I measured
> the temperature by looking at the B mode resonance of the crystal.
> I wanted to get rid of any linear frequency drift.  As a rough
> rule of thumb, 1 hour of cool down is pretty good for most purposes.
> For extreme measurements, I would allow 10 hours.  This reduced
> any exponential tail to below the ability to measure temperature and/or
> below the effects of the ambient.  I had to put a box over it to
> reduce the effects of air currents.  If I did not do that, then 1 hour
> was all I needed.
> Rick Karlquist N6RK

More information about the time-nuts mailing list