put component into oil-filled "can".

[cnt]

I want to make a heated zener reference (I know I can buy them, but I want to make it).  My plan is to put a zener, NTC, into a metal "can", fill it with oil, and attach a heater (resistor) to it. 

I can do this by machining something, adding a gasket and screwing it down.  Before I do anything stupid or overly complicated though, I'm wondering if anyone has already done something similar and how they accomplished it.

Any tips tricks links would be appreciated. 

[Gyro]

Check out the Metrology section, you may find something there.

Thoughts that come to mind...

- You want to use medical grade liquid paraffin as the oil (there's certainly a discussion on that in Metrology).
- Your main enemy is probably convection currents and hotspots. Oil filled calibration baths normally include a stirrer. Relative positioning of Zener, thermistor and resistor are probably important.
- You might achieve better results by machining a solid Aluminium (or better, Copper) block with pockets and thermal compound coupling for the zener and thermistor as well, either in the oil can or standalone. ie. Keep the thermistor and zener as closely thermally coupled as possible.

[cnt]

Thanks for the input!

I searched metrology but didn't find much, but I do find searching this forum kinda hard so maybe I missed something.

For my current plan:
I have the zener and NTC both in DO35 packages and I plan to lay them out symmetrically with the "can"(1) itself heated by a to92 to-220 resister in the center, hopefully spreading the heat evenly or at least symmetrically between the two.  IDK what size scales convection matters.  the cavity is probably going to be in the range of 12x6x6 mm.

(1): I can't really fabricate a can so it'll be an alu or brass block with a pocket and a gland around it for a gasket.

[coppercone2]

since its electrically heated and filled with oil, you should put a weak point on it so it does not accidentally pressurize itself with hot oil. It should be designed to pop a seal with over pressure in a safe way. I see the real oil standards are often open or just have a loose lid on the top, but they are basically like crock pots. It seems like something bad could happen if you have a prolonged overload followed by a really heavy transient, but unless its hooked directly into mains I dunno if its a concern.

On the other hand beware of oil leaking and hurting you with an oil slick, since its a home made container.

[RoGeorge]

In theory you can use oil, in practice can be messy, but it all depends at what you want to use the final voltage for.

AFAIK normal Zener diodes are noisy to be used as a metrologic voltage reference (between other disadvantages).  There are quite a few threads about voltage reference and thermal stabilization, e.g. this huge one:
https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/

By the flag you are from US, should be plenty of cheap spare equipment on ebay where you leave.  You can find a metrology reference pile, or an old voltage reference module from a retired lab, or from former military equipment, etc.

For amateur grade reference, maybe you won't get much advantage from oil, a radiator coupling in a small thermally isolated can might do as well as the oil filled can you want to build, but I have no hands on experience with this.  Better set some specs you aim at, then ask for advice in the metrology section.

[mendip_discovery]

From what I know the use of the oil is to prevent rapid changes in temperature caused by the ambient temperature changes. For example with standard resistors, though it also helps with some heat dissipation.

You can heat the oil to warm temp but be wary of thermal expansion if you intend of heating up a considerable amount.

[Vgkid]

I will say, as of the recent decade only resistors have been put in oil, due to the power dissipation. Most voltage(if not all) zener references are in air.

[Conrad Hoffman]

There are all sorts of pluses and minuses but you might be interested in the construction of the old Julie Research standard cell oven, SCO-106. It was for a bank of three standard cells but the technique would work for anything. Air, not oil. Oil is a PITA and you have to stir it. First, they machined a heavy wall aluminum box, probably 1/2" thick walls. It had a heater around it, can't remember if they used wire or flat Kapton heaters. They used a mercury thermoregulator, probably in a drilled hole with close coupling to the heater- you don't want any time lag. You'd use something sane, like a thermistor, in the same way. You control the walls, not the air. They then made a second heavy wall aluminum box to fit the first one, with Styrofoam plates between the two on all sides. Same temperature regulation scheme, but with the outer box just a tad cooler than the inner. The whole assembly went into a wood case (good thermal insulator) with a good amount of Styrofoam between the inner boxes and the wood case. The standard cells had a length of wire nearby, maybe nickel, that formed one leg of a bridge, with a pot on the outside. This had nothing to do with the control but was only used to monitor the temperature of the inside of the inner box and cells so you'd know if anything was amiss or could compensate for variations. I can probably find more info if you need it.

[cnt]

In theory you can use oil, in practice can be messy, but it all depends at what you want to use the final voltage for.

AFAIK normal Zener diodes are noisy to be used as a metrologic voltage reference (between other disadvantages).  There are quite a few threads about voltage reference and thermal stabilization, e.g. this huge one:
https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/

By the flag you are from US, should be plenty of cheap spare equipment on ebay where you leave.  You can find a metrology reference pile, or an old voltage reference module from a retired lab, or from former military equipment, etc.

For amateur grade reference, maybe you won't get much advantage from oil, a radiator coupling in a small thermally isolated can might do as well as the oil filled can you want to build, but I have no hands on experience with this.  Better set some specs you aim at, then ask for advice in the metrology section.

Perhaps you're right.  Part of what I wanted was to make the control loop from heating element to NTC faster and reducing the thermal gradient top to bottom.  But maybe I'm trying too hard.  Building version 1 dry is a lot easier.

As for specifications, I'm just messing around.  I want to see how good I can make a 5$ zener perform as a reference.   

[Doctorandus_P]

Using oil for this seems a bad idea.
It's messy and time consuming to make, and if your oil ever leaks out it becomes a bigger mess.
I also see no advantages for using oil in this application.

why not just do like normal people do?
Use a bit of silicone based glue to glue your parts together.
You can add some extra thermal mass (some aluminium strip, or wrap copper wire around it) and thermal insulation around it to manage your thermal time constant.

[coppercone2]

and if your curious about oil, you can try it out in a beaker on a work bench before doing the chassis/system engineering stuff.

[PartialDischarge]

I did that, package is not leaking
https://www.eevblog.com/forum/metrology/10k-precision-resistor-project/

[EPAIII]

Liquid paraffin? I don't know, but that sounds flammable. If that is used, I would have one or two LARGE fire extinguishers on hand.

Frankly, I would look for an oil that is difficult to ignite along with the proper electrical characteristics. And yes, do have a safety vent. I remember that my mother's old pressure cooker had a rubber plug in the lid that was designed to fail if the pressure got too high.

Check out the Metrology section, you may find something there.

Thoughts that come to mind...

- You want to use medical grade liquid paraffin as the oil (there's certainly a discussion on that in Metrology).
- Your main enemy is probably convection currents and hotspots. Oil filled calibration baths normally include a stirrer. Relative positioning of Zener, thermistor and resistor are probably important.
- You might achieve better results by machining a solid Aluminium (or better, Copper) block with pockets and thermal compound coupling for the zener and thermistor as well, either in the oil can or standalone. ie. Keep the thermistor and zener as closely thermally coupled as possible.

[Gyro]

It's not like the stuff that you put in lamps and planes, it's more viscous and has no vapour at ambient (or oil bath) temperature. Yes it is flammable, but you'd have to try quite hard. Transformer oil is flammable too.

[NiHaoMike]

If you can find a wax that melts right at the temperature you're trying to regulate at, the phase change will help to stabilize the temperature better than plain oil would.

[Kleinstein]

The galss case diodes are allready hermetically sealed. So there is little need for oil. Oil is making things quite messy, so I would avoid it. Normal parafin wax is OK and a bit easier to handle. For the start, I would go without the oil or wax and get a constant temperature oven done to get a feeling on how good it works.

A zener reference has a quite constant power and thermal gradients are thus not so critical. A quitestion is if the oven is just for the zener or also for the resistors that are used with it to privide a constant current.

For the heater I find a transistor more convenient, as the transistor as heater is easy to get linear, a resistor would be linear only with PWM. Analog control mainly works good if the oven part to regulate is fast. This means a relatively tight coupling from the temperature senensor to the heater and a more smaller oven. It helps to have an extra independent temperature sensor right next to the zener, that is not used for the control loop, but just to test the perfomance of the oven. With a reasonable low TC zener the oven does not need to be very accurate - ideallly it is to get the TC trimmed to near zero via the zener current and the oven mainly to fight the 2nd order part.

[jmelson]

I want to make a heated zener reference (I know I can buy them, but I want to make it).  My plan is to put a zener, NTC, into a metal "can", fill it with oil, and attach a heater (resistor) to it. 

I REALLY doubt you can do better than an LM4140-2.5 run off 4.0 V supply.  Use an AD706 op-amp if required.  I have used this in a number of systems where very stable and quiet power supply was needed.
Jon

[Conrad Hoffman]

I have a piece of test equipment with an oil chamber, not for temperature, but to stabilize a plate capacitor from barometric changes. They put a small sheet of neoprene over a hole and attached a cover plate to hold it. The cover plate was drilled in the center and had a pocket machined on the inside. That way the oil could expand and contract with temperature and the neoprene sheet would simple deflect as necessary, keeping the internal pressure near ambient. Still don't recommend oil but if you do use it the trick might come in handy.