Humidity control in DIY hermetic units

[VNUTDENYER]

Hello Metrology

New to eevblog, last few days reading vref posts, and I don't have a question.  Searched and found nothing so here goes.  It seems as if humidity in a sealed, hopefully hermetic, reference could/should be a problem with some devices.  I had planned to build a LM399 ref to avoid some of that but remembered a material used for a similar purpose.  The 4A molecular sieve binds water vapor, some gases and hydrocarbons but is shipped in sealed 2 #, and larger, jugs and does not implode so should be safe.  MSDS should be studied and maybe even a test in the proposed container, without ref, but it is cheap, available, and fast.  Can be dryed out at 250-450 F depending on RH and amount of H2O absorbed. 

[Conrad Hoffman]

I think 4A is a good choice. The real problem is the DIY hermetic enclosure. Aluminum is a problem because it's often porous. Watch out for cast aluminum especially. You want to bake it out, preferably under vacuum, then back-fill it with dry nitrogen.

[VNUTDENYER]

Hi Conrad

Thanks for info, that was my rough plan but figured I could get 4A in small quantities easier than dry nitrogen.  Still looking for enclosure. 

[DaJMasta]

How much humidity are we talking?

If you just leave the near-finished unit in a box with sillica gel for a day or two, the time it takes to seal it up is not going to let in much humidity, you'd end up with maybe 10 or 15% RH at room temperature in your sealed container.  If your container is actually hermetic, so it doesn't let in more water, just assembling in a very dry environment should be sufficient to keep the humidity very low.

[Kleinstein]

Both the molecular sieve and silica gel effectively add capacity to absorb quite some water without much change in RH of the gas. AFAIK the silica gel is effective more at the higher RH levels (e.g. 10-50% RH) while the molecular sieve is more suited to keep it very dry (e.g. 0.1-5% RH). The molecular sieve might even absorb some air and this way add extra pressure variations with temperature which may not be good.

I don't think it makes a big difference if one uses dry nitrogen or not. Even a small quantity of the buffer material can easily absorb that little humidity in the usually small volume in the case to seal.

[DaJMasta]

I guess my implication was to keep the device in a chamber with silica gel before sealing, then to seal it off including no additional desiccant inside the device.  Including a desiccant in the chamber itself may still be a decent idea, but in the case that it's actually beneficial.... your seal is not hermetic and it's only a matter of time anyways.

Not going to work for devices that need less than 10% humidity, sure, but if that's not required, it's pretty easy to manage if you can do the final seal quickly wearing gloves.

[Kleinstein]

I would still include desiccant inside the device. This would effectively make the "volume" larger so that small leaks are less important. It is pretty difficult to get a enclosure really 100% tight. This get's obvious when working with vacuum. So with desiccant inside it takes much longer until significant humidity levels are reached. 1 g of desiccant could kind of replace a few liters and thus extend the time by something like a factor of 1000.

[Conrad Hoffman]

FWIW, the General Radio standard capacitors (the black brick type) included a cylinder of silica gel desiccant inside before they were sealed with tar or resin or whatever that stuff is. No idea how long it was effective.

[coppercone2]

its going to form a humidithermal oscillator I think.

But anyway, be sure to look at youtube to see the difference between ebay molecular sieves and ones from sigma etc. i THINK nilered had a video about it. Or the nurdrage guy.

What might be more stable is actually a salt that forms a hydrate. The salt should react to form a hydrate which won't decay until significant temperature is reached. I don't think molecular sieves have the same stability as a hydrate bond.

and something that reacts with oxygen too.

What kind of bonding occurs in molecular sieves like zeolite and silica??

[SvanGool]

https://www.eevblog.com/forum/metrology/silica-gel-for-stabilizing-humidity/msg1868024/#msg1868024

[3roomlab]

errm wad about the solid state dehumidifiers? rosahl?
https://www.micro-dehumidifier.com/

[VNUTDENYER]

I barely want silica or zeolite inside a "sealed" box but for sure not some kind of salt (seen what it does after saturation)!!

[coppercone2]

if its hermatic you can use a salt but if its not then you will get corrosive goo

[chickenHeadKnob]

Hello Metrology

New to eevblog, last few days reading vref posts, and I don't have a question.  Searched and found nothing so here goes.  It seems as if humidity in a sealed, hopefully hermetic, reference could/should be a problem with some devices.  I had planned to build a LM399 ref to avoid some of that but remembered a material used for a similar purpose.  The 4A molecular sieve binds water vapor, some gases and hydrocarbons but is shipped in sealed 2 #, and larger, jugs and does not implode so should be safe.  MSDS should be studied and maybe even a test in the proposed container, without ref, but it is cheap, available, and fast.  Can be dryed out at 250-450 F depending on RH and amount of H2O absorbed.

There is a basic error voltage ref volt nutters make in wanting to keep everything dry as possible. A dry(er) box is not as good as a RH stable box! You want to avoid your ref varying with the seasonal changes of RH in your locale, that means the optimum RH inside is equal to the seasonal average of your lab. And you want to attenuate, if not eliminate any interchange of gases. As Kleinstein points out complete hermeticity is more of an aspiration than realizable goal so if the inside is bone dry at the start then it can only go up in RH and the epoxies inside can only swell, ergo drift in one direction. Keeping some silica  gel inside at the seasonal average RH can act as moisture capacitor, but you specifically don't want to put in dried out desiccant at the start. Let everything that goes inside acclimate to your mean RH before sealing.

Note that any ovenized box will try to forcibly expel internal atmosphere then suck external  atmosphere everytime  it undergoes a heat/cool thermal cycle. And you would be surprised at how much gets by most sealing attempts.

Somewhere in the humungous LTZ1000 thread there is a NIST pdf about seasonal variation showing up in the vrefs, and one site showed virtually no variation. That is, the lab there had essentially unchanging RH. Joe Gellar also reported seeing very little change  and corresponding effect in his lab. He lives somewhere near the Atlantic coast. I live on Vancouver Island about 500 meters from the water and my house and lab is remarkably stable. It really doesn't go beyond a maximum 55%RH and minimum of 49% , which is ideal for human livability  and wood  dimensional stability. Vref's probably don't care as long as it is non-condensing and never changing.

[Kleinstein]

The absolute RH also has an effect, especially on surface leakage. For low leakage one may want a low RH, like 30% RH, but definitely below 50%, even if the outside tends to be higher humidity.

Especially the buffering salts have the tendency to keep the RH about constant at a level specific to the salt, even if the temperature changes. Many desiccants have a specific RH range where they work and do buffer a lot. If it's much dryer or more humid they have little effect as they are saturated in one way or the other. So the desiccant is more like a rechargeable battery not so much like a capacitor that is linear over a large range.

One does not want to get a salt all the way to a saturated solution - the molecular sieves are less critical here.

The active dehumidifiers are only practical if active 24/7 or at least close too. When inactive there can be quite some humidity going back in. So it might still need some desiccant inside for smoothing.

[SvanGool]

@Kleinstein: do you have any documentation/information documenting the stated 30% and/or 50% RH values or are they more based on your own experience?

[Kleinstein]

Surface leakage is known to be highly effected by humidity. I have an old NBS publication on this - looking up where it comes from I just got surprised how old it is. So if you don't mind an 100 year old publication (NBS - Bulletin Vol 11 No3 from 1915) 

http://dx.doi.org/10.6028/bulletin.256

has quite some information on surface leakage.

[Vgkid]

Surface leakage is known to be highly effected by humidity. I have an old NBS publication on this - looking up where it comes from I just got surprised how old it is. So if you don't mind an 100 year old publication (NBS - Bulletin Vol 11 No3 from 1915) 

http://dx.doi.org/10.6028/bulletin.256

has quite some information on surface leakage.

That is actually a really interesting read, thanks for posting it.

[BravoV]

How about seal it inside a glass ampule ? This had been done for decades for liquid medicine ? 

https://youtu.be/GEmINGq8HDQ

https://youtu.be/7hRPs64NOf8

[Kleinstein]

Handling glass is reasonably easy, but getting wires through and still have it sealed is not that simple.

[David Hess]

I have had problems in the past with precision electronics in sealed enclosures suffering from outgassing of other materials so I am more inclined at this point to use a baffle like that found in hard drives to allow barometric pressure equalization and air flow or if the enclosure is truly hermetic, apply a vacuum which has the added benefit of removing errors from convection.

[VNUTDENYER]

A4 sieve likes hydrocarbons as much as water vapor. 

[beanflying]

errm wad about the solid state dehumidifiers? rosahl?
https://www.micro-dehumidifier.com/

Nuts and bolts teardown by BigClive from last year. Interesting device for more than just Metrology.

https://youtu.be/Vabq-s62IVM

[Vtile]

Pfff.... Hermetic seal is nothing more than 3" copper water pipe with both ends soldered and opened TO3 cans as feed through.  Like a big Fluke resistor cans..

Glass is a good material also, its thermal resistance is so high that you only need like 1" of material from melted to something you can hold in hand, but then you would need to get also this untreated covar and I suppose some form of acid treatment before feed through joinery.

[LaserSteve]

Solderable Russian and American hermetic seal pin assemblies with the glass to metal done for you are all over Ebay.

example:  https://www.ebay.com/itm/ABS40WPP-5940-00-811-5060-Feed-Through-Terminal-Lot-of-50/263792708668?hash=item3d6b454c3c:g:fUYAAOSwNMVaY4bI:rk:19:pf:0

Steve