Building your own voltage reference - the JVR

[EC8010]

Perfectly happy to make an oven - I've previously immersed components in oil. (Top tip: Use silicone sealant intended for engine assembly or oil seeps through tight threads.) I have plenty of J111, so I'll bin some for VTO to find a reasonable pair for the cascode. Been wanting to make a Peltier device do something useful for some time. Previous experience is that they need a huge heatsink to be effective.

Surprised that humidity should be a problem with epoxy packages, but if you've measured it... And absolutely mystified by your used rather than new desiccant recommendation! Please explain.

I thought I remembered something about JFET references in AoE 3rd Ed. So I looked it up (p680). This is where the IC manufacturers have a serious advantage. AoE describes a two FET reference where the voltage difference between two FETs of identical geometry but slightly different doping is measured by an op-amp, thereby reducing the tempco.

[dietert1]

Last automn i put dry desiccant into a glas with rubber seal, together with two NOMC resistor arrays (plastic package to test) and a meter. Until february the meter showed 0 %RH (out of range). Now it reads 2 %RH. To get equilibrium near 50 % one may have to wait many years. If you use "invalid" desiccant instead, the humidity difference between inside and outside will be less. With a difference of 10% instead of 50 % the expected drift within a year would be a fraction of 1 %RH.

Regards, Dieter

[Kleinstein]

The peltier elements need a large heat sink only for cooling. For heating they are perfectly OK with a relatively small heat sink, though a reasonable large one still helps.
For the relatively small and low power JFET based reference a simple oven based on heating alone is fine, especially if a more moereate temperature like 40-50 C is used.
With the linear TC reasonable trimmed the oven does not have to be that accurate or stable. Against the square part even a rather crude oven is quite effective.

[iMo]

Is that The "Special Reserve Kaluga Huso Hybrid Caviar" or the "Beluga Hybrid Caviar" in that jar?

[dietert1]

No caviar here. The desiccant (orange when fresh, green when invalid) turned black when drying in the oven. The theory with "invalid" desiccant is to have a buffer that absorbs and yields humidity.

Regards, Dieter

[EC8010]

Used desiccant obvious once stated. Now all I have to do is to find some. I'd rather gently heat a small oven using a TO220 resistor. In fact, I'll just make it a bit of 6mm thick aluminium with some 4.8mm holes for the JFETs, epoxied in place and an NTC buried in a little hole with more epoxy. With only a little mass, not much power will be needed. I've just measured rather a lot of J111 for VTO at 20nA. Looks like -8.2V to -8.4V is the most popular bin, and I saw quite a few -8.25V devices. Tomorrow, I'll look for matches within that bin. Because the VTO is so high, they won't have a particularly high mu, but that can be surmounted by making a triple cascode. Probably need 24V rather than 12V to power it.

[Kleinstein]

For a cascode I would consider lower threeshold devices on top. As an extra bonus one also looses less voltage. The upper one may be better outside, so the variable power from a variable votlage does not cause to much variable heat.  For a voltage reference the cascode part would be to get a higher voltage, like references in series. The drain voltage is from an amplified voltage (e.g. 1.25 x or 1.5 x).

For the oven it is not only the size that matters, but also the power of the parts in the oven. With high power there, the isolation can not be that good as one would otherwise like. For the oven design it really helps if the sensor reacts fast to the heater, possibly even faster then the actual reference part.  To avoid stress to the transistor it may not be such a good idea to glue it to the aluminum block.

[magic]

FYI, a potential approach to combine opamp scaling with high impedance constant current output.

You would replace M1 with the JFET reference and derive V_B from its source voltage.

[EC8010]

Point taken about outer device and its variable heating. I had noted that the drain voltage came from amplified reference in the same way that many linear regulators power their reference from their output.

Yes, I'm aware that the sensor needs low thermal mass and good coupling so that it accurately reflects what is going on. Also about size and power. Ultimately, there's a thermal mass whose temperature you want to change by heating, so the lower the mass, the less power needed. At the same time, there are losses to the environment, and although it would seem sensible to minimise them, they are actually needed to help in stabilising temperature, because if you apply too much heat, you want the system to cool quickly (ideally with the same time constant as for heating). Subconscious wondered about mechanical stress to the JFETs and gluing in block. I'll think about that one some more.

Later edit: Conscious has thought about it and decided that gluing JFETs into a block held at constant temperature is fine. What wouldn't be fine would be gluing them into a block that changes its temperature and has a different coefficient of expansion to the epoxy containing the silicon die, thereby applying variable stress to the silicon as temperature changes.

[EC8010]

The J112 is not that bad, mainly limited with the plastic case that can make it susceptible to humidity effects. So far my favorite is the 2N4391, as one of the cheaper ones in a metal case and relatively well available. The current (1-1.5 mA) and voltage (around 5-8 V depending on the batch) are reasonable.

Kleinstein, could you expand on your epoxy humidity comment, please? What are you suggesting, surface leakage or bulk leakage?

[Kleinstein]

The weak point with a plastic case is an effect of humidity. Epoxy and many other plastics slowly take up some humidity from the air and this leads to swelling and thus a change in the mechanical stress to the part inside. The mechanical stress can effect semiconductor parts and resistors, e.g. with a change in the voltage or current. The humidity effect is usually slow, like a few days to weeks to stabilize.
Heating the parts makes the effect faster and with a higher temperature compared to the environment the rel. humidity is reduced (like half for every 10 K). For a part used intemittendly in a more humid climate this means some humidity drift after turn on.

[EC8010]

Thanks for that. I'm aware that plastics absorb water (it's why you can't make accurately dimensioned nylon parts), and the asymmetric shape of a TO92 case will apply changing stress with different absorption. Interesting. I was thinking in terms of direct electrical leakage rather than a mechanical effect. And that has an interesting knock-on...

I have previously thermally bonded TO92 devices by drilling a 4.8mm hole in 5mm thick aluminium clamped to a piece of sacrificial aluminium so that when separated, and the device is inserted, the TO92 flat face very slightly protrudes from the face. I then clamp that piece of aluminium to another and achieve a good thermal bond all round the device. But from what you've said, a better solution for long-term stability is to simply drill a complete 4.8mm hole and fill the gap with epoxy. It might not be exactly the same kind of epoxy as used by the semiconductor manufacturer, but it's a much closer humidity absorption match than aluminium. If you look at a TO92 device, the legs (and presumably die) are on the diameter of the perimeter, so returning the epoxy to a cylinder would reduce mechanical stress from absorbed humidity.

Thanks again.