[time-nuts] Re: MHM-A1 maser temperature stabilization

[Neil]

Hi Bob, luckily I'm only pulling about 6 watts of heat from my OCXO. The 
Peltier takes about 25 watts once it is stable.  It would be impractical 
for anything needing to lose more heat.

If I lose the cooling, the system still works, but the Morion heater 
backs off and starts to cycle a little more.  There's just enough 
passive cooling via the copper bar to make sure the OCXO heater doesn't 
turn right off, at least so long as the room temperature is below 30C. I 
keep the Peltier cold-side temperature above the worst-case dew point to 
avoid condensation.

I think I could swap the Peltier for a new one without turning the 
system off, but I don't need long-term stability over weeks, just over a 
few hours. It has two gel cells which are float-charged, so I'm pretty 
much immune to mains power variation or outages, but this ain't a maser.

Amazingly good for an old quartz lump, so long as the vibration 
isolation works properly. It gets weird if the gravity vector doesn't 
point the right way or if a heavy tractor drives past the house and 
rattles the equipment racks.

At that point I get plaster falling out of the cracks in the ceiling.  
Clever new-fangled things like foundations hadn't been invented back 
when this house was built, so I have to compromise a bit.

-- 
Neil
https://youtube.com/MachiningandMicrowaves

On 17/01/2023 20:09, Bob Camp wrote:
> Hi
>
> I’ve looked at Peltier gizmos. If you are pumping 100W of heat they seem to use
> quite a bit of energy to get the job done. A compressor based setup of some sort
> might do a better job. With any sort of cooler, my fear is a loss of cooling. If the
> device is very well insulated, damage could occur.
>
> The 120 L bucket of water is about half the size of my “rough guess” thermal mass.
> I fear that the guess may be way low ….
>
> Bob
>
>> On Jan 17, 2023, at 2:09 PM, Neil Smith via time-nuts<time-nuts at lists.febo.com>  wrote:
>>
>>
>> One thing I've tried to reduce diurnal and seasonal thermal variation is putting the sensitive components in a hyper-insulated enclosure with high thermal mass mounted inside another stabilised oven, also with plenty of thermal mass, but then connecting a well insulated copper bar from the inner enclosure to a stabilised Peltier cooling plate mounted outside. That gives me a very stable thermal flow with a large temperature gradient so the inside oven thermal control loop can stay in a nice linear regime and doesn't have to deal with large changes in required heating power.
>> It's overkill for my application, which is a only a 10MHz Morion MV-89a OCXO.
>>
>> I also have a 120 litre water tank with a circulating pump that can keep a cooling plate at a very stable temperature over periods of an hour or so, but its main use-case is for cooling high powered RF amplifiers.
>>
>> Neil
>> https://youtube.com/MachiningandMicrowaves
>>
>>> On 17 Jan 2023, at 18:05, Bob Camp via time-nuts<time-nuts at lists.febo.com>  wrote:
>>>
>>> Hi
>>>
>>> I may well be missing something obvious in the electronics to heat conversion process
>>> ( it certainly would not be the first time … :). That said:
>>>
>>> I set up a box with foam and the 100W from the device inside heats it up by about
>>> 4 C. You immediately get into questions about “4C where?”.  Just accept the number and
>>> move on for now.
>>>
>>> The room wanders a bit. How much depends on a lot of things. 2C is not a bad guess on
>>> most days. At the wrong time of year it could be 2X that. There is a day to night component
>>> that usually dominates.
>>>
>>> Target is to damp out the 24 hour swing. The R/C should be longer than that. How much
>>> will depend a bit on the temp stability target. 1C/24 hrs is an improvement. The folks who
>>> made the gizmo suggest that 0.1C is a better target. Yes, the “C” in the R/C just moved 10:1
>>> as you changed that target. (There are other issues as well, for now, let’s ignore them)
>>>
>>> As I do this in my usual hand waving fashion, I come up with hundreds of liters of water for
>>> the thermal mass. It just goes up if I move from 1C and get closer to 0.1C.
>>>
>>> This assumes that everything else is zero mass. Things do “follow” the room temp with an
>>> lag of a couple hours. Even to get to 1C, significant thermal mass needs to be added.  ( or
>>> lots more insulation, stick with mass for now ).
>>>
>>> Am I missing something or is the “hundreds of liters” guess more or less in the right range?
>>> There are some practical implications to playing with groups of ten  “jerry cans” full of water.
>>> (even as a “try it and see" experiment).
>>>
>>> No, this isn’t an attempt to come up with a full up answer. It’s just a question about what
>>> the rough order of magnitude is in this approximate case. There are enough holes in the
>>> data above that any sort of precise answer is data limited.
>>>
>>> Bob
>>>
>>>> On Jan 17, 2023, at 4:07 AM, Poul-Henning Kamp via time-nuts<time-nuts at lists.febo.com>  wrote:
>>>>
>>>> --------
>>>> Skip Withrow via time-nuts writes:
>>>>
>>>>> One issue that I have been up against is the temperature variations of the
>>>>> lab location.  The diurnal excursions of the oven heaters were clearly
>>>>> visible.  My solution was to build an environmental box (1.5" foam
>>>>> insulation board) around the unit
>>>> Apologies for harping about this again:
>>>>
>>>> Please think in terms of thermal impedance!
>>>>
>>>> Look at it as electronics for a second:
>>>>
>>>> You have put a (very big!) resistor between your AC noise source
>>>> (the heating) and the sensitive kit (the maser).
>>>>
>>>> Because your resistor is so large, you now have an over-temperature
>>>> problem inside your enclosure, which you have tried to mitigate with
>>>> a fan.
>>>>
>>>> What you actually need is a low-pass filter with a cut-off lower than 1/24h.
>>>>
>>>> That means /some/ insulation, but not so much that your maser cannot
>>>> get rid of the heat it produces, and /a lot/ of thermal mass on the
>>>> inside to "short the AC to ground".
>>>>
>>>> This is not magic, and the math is trivial when you already know electronics.
>>>>
>>>> -- 
>>>> Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
>>>> phk at FreeBSD.ORG          | TCP/IP since RFC 956
>>>> FreeBSD committer       | BSD since 4.3-tahoe
>>>> Never attribute to malice what can adequately be explained by incompetence.
>>>> __