Zener accuracy over temperature range

[Simon]

I'm looking to use a zener as a voltage reference as I need to work down to -46C and most voltage res only go down to less than -40 (farnell's parametric search is crap and does not include temperature ranges) so I'm thinking a zener as it's just a ref for a SMPS, but looking at zener datasheets there does not seem to be any accuracy versus temperature information, does this mean they will stay accurate over the range or is it just misleading ?

[tautech]

You will need a Temp compensated reference zener.
Check this thread for discussion and sample datasheets:
https://www.eevblog.com/forum/buysellwanted/wtb-9-1-v-temp-comp-reference-zener/msg459407/#msg459407

[David Hess]

You are looking at the wrong zener diodes or data sheets.  Not all zener diodes are characterized for temperature coefficient.

[Andreas]

Hello,

for  integrated cirquits you would have to look for hermetically sealed devices like LM136 LM185 (TI) or AD586TQ (AD).
If looking for Zeners: the T.C. is voltage dependant. Lowest T.C. (around +/-2mV/K) is around 5.6V for uncompensated Zeners and 6.2V (down to +/-10ppm/K) for compensated Zeners like 1N829A. But you will have to use a constant current source to reach tight zener voltage specifications.

Some manufacturers give T.C. specifications (at a certain zener current).
http://www.onsemi.com/pub/Collateral/BZX84C2V4LT1-D.PDF

With best regards

Andreas

[poorchava]

Wouldn't a garden variety TL431 be more convenient? It's not expensive at all.

[T3sl4co1l]

Zeners don't stop zenering, but the voltage varies... you didn't mention how tight your requirement is.

Tim

[Andreas]

Wouldn't a garden variety TL431 be more convenient? It's not expensive at all.

Do you know one that is specced to operate (not stored) below -40 degrees?

With best regards

Andreas

[Simon]

As I can't find much in the way of voltage regs that do under -40 I'm thinking of making my own as i only need a few tens of mA but as the circuit is potted I'm nervous about linear regs as finding something adequate for non potted was hard as I'm dropping from as much as 30V down to 5V. I'm running a MCU so voltage is not too critical but I don't want to drop under 4.5V

[theatrus]

I'm curious where you are finding pretty much any parts specified at below -40.

[Simon]

Well actually most SMD passives and descrete semiconductors will do down to -55 and even logic gates and basic chips like comparators will do -55

[Andreas]

As I can't find much in the way of voltage regs that do under -40 I'm thinking of making my own as i only need a few tens of mA but as the circuit is potted I'm nervous about linear regs as finding something adequate for non potted was hard as I'm dropping from as much as 30V down to 5V. I'm running a MCU so voltage is not too critical but I don't want to drop under 4.5V

What about good old LM723 in cerdip or TO-100 package

[Simon]

Might be worth looking at

[poorchava]

Look at automotive grade parts.

[Simon]

Well so long as they are under -46C

[Dave Turner]

If you're not too worried about power consumption (clearly there is some heat loss already dropping from 30V) how about using some power for some temperature control.

[SeanB]

I used a 723 to make a voltage reference, and the only rail was a nice 60VAC rail, so I used that.  A first stage of a 33V zener and then the 723 in a TO100 can with a small push on heatsink and it works still. Was plenty stable enough for the intended use, as a meter offset to give an extended range.

[Fsck]

Well so long as they are under -46C

auto should be -55 to 125C or somewhere thereabout, or industrial/extended operating range/whateverothernamemeaningthesamething.

[Simon]

Well Dave with the same surname as me the problem is that this is potted so I'm also concerned about what happens at 50C ambient and it is hard to find a regulator that will cope with the power dissipation without heatsinking even in free air. The product needs to withstand very severe shocks so large items are out and this is why it is potted. I have considered heating for low temperatures as I can't get an MCU that will do -46 but stops at -40 but as it's potted so hard to heat consistently I'd rather heat just the MCU and make sure the rest will work down to -46.

[eneuro]

Wouldn't a garden variety TL431 be more convenient? It's not expensive at all.

Do you know one that is specced to operate (not stored) below -40 degrees?

I used to use in many projects TL431I  (Industrial rated -40*C..+85*C operating temperature) but when you look into it's datasheet eg. this one http://pdf.datasheetcatalog.com/datasheet/philips/TL431I.pdf there are many figures with as low as -50*C, so who knows maybe it could survive in such extreme conditions? 
Maybe it will not keep perfect its parameters, but shouldn't be better than any not trimmed Zener diode without vref temp compensated?

[Simon]

Unfortunately i can't rely on random batches being capable of exceeding the manufacturers specs. For the application this is going into I need it to be specified for what I want. the only concession i want to have to make is for the micro controllers because i simply can't get anything below -40 unless it's expensive and complicated and very hard to get hold of so easier to heat +10C to keep them cosy if needed.

[johansen]

is an REF5050 (or one of the other voltages) too expensive?
the datasheet shows a fairly predictable -3ppm/c drop from 0 to -50C

[lewis]

Simon, you always seem to get these weird design jobs with very tight constraints...

The best way is definitely to heat the product if you need to go to -46, you'll really struggle to get components rated for that temperature (you'll need -50 to build in margin) in realistic quantity, at realistic prices and with realistic lead-times.

(Not only that, but you'll really need to take into account the coefficient of thermal expansion of the potting compound if you need the product to work over a very wide temperature range. I make potted stuff that is spec'd from -40 to 70C, and it ain't easy to get right.)

In fact, it's *easier* to heat because it's potted, you can maintain the resin relatively isothermal because the resin is an excellent thermal conductor, rather than relying on heating air in a normal enclosure. The simplest way to do it would be to use some low value resistors dotted around the perimeter of the PCB. When the temperature gets below, say, -10, use a spare pin on your MCU to PWM the resistors (via a mosfet) to begin heating the resin. As the temperature drops, increase the duty cycle to maintain a relatively constant temperature within 10-20 degrees or so. Use any old temperature sensor you like, it doesn't need to be particularly accurate. MCP9700 series springs to mind.

If you have a voltage regulator in there dissipating enough power that you need to worry about it at 50C ambient, then you probably have enough heating as it is anyway.

[Simon]

Yes i always get asked the impossible, I already suggested heating the whole enclosure but my boss was not happy with the idea.

The thing with heating is that I don't want to start the MCU's at -46 as then they may behave strangely and I still don't get heated, so I'd use a thermistor and comparator to kick in various resistors around the board. Does potting compound transfer heat that well ? I'm using scotchcast N8.

linear voltage regs are a pain, you will actually struggle to find any that do more than 0.5W even in bigger packages, they seem to end up needing heatsinks, so I'll either use multiple regulators as I can divide the circuit up or a switchmode even if i make it myself

[Andreas]

Hello,

I hope that you don´t need any crystal for the oscillator.
They also have sometimes unusual behaviour below -20 deg C.
Especially the higher frequencies.

With best regards

Andreas

[tautech]

Does potting compound transfer heat that well ? I'm using scotchcast N8.

The datasheet will have details of thermal permeability.
If it doesn't, use something else.