zener cunundrum,

[Simon]

I need to regulate 20-30V down to 15V in an ambient of up to 83C but also down to -46   

So first huge problem is the 83C because the zener de-rates badly, i could use more than one in parallel but then i now need so much more current to make them all work and higher wattage zeners have such bade tempco's that it's pointless and they need even more minimum ballast current.

Any ideas ? I only need 12mA and it only needs to be stable +/-2V, I can make the ballast resistor so that at 20V input my output drops to 12V because if i try and get the correct resistor for 15V @ 20V input when i am at 30V input that is 3x V and 9X W 

I could use a voltage regulator but again can't get any that do -46 only -40 and I have no idea what will happen at  -46, again it don't need to stay accurate, just not go haywire.

[rob77]

hermetically sealed box with heater and temperature control... and all of the sudden you need parts for 0-85 celsius

[rob77]

btw... where do you get a temperature span of -46 upto +83 ? just for the sake of curiosity

[Simon]

Well unfortunately that is the temp range I was given, I now have found a zener that is 2W and says that the "knee current" is 1mA i assume this is the current required to get any form of reaction from the zener and sounds more like it. I do have heating built in but unfortunately this zener is providing power to the comparator and the temp sensor that switch on the heater which is just 10W of resistors buried along side the MCU which at the moment is the only part that won't go under -40C

[Andy Watson]

I could use a voltage regulator but again can't get any that do -46 only -40 and I have no idea what will happen at  -46, again it don't need to stay accurate, just not go haywire.

Your google foo is failing! RS list 14 devices that go down to -55, here's one:
http://uk.rs-online.com/web/p/linear-voltage-regulators/8060956/
Edit: That part may be a bit short on input voltage, try this one instead - RS 688-9329

[Simon]

I've been using digikey, RS aren't reliable enough to have all of what I want and i need to order all from one supplier to avoid extra carriage charges, don't get me started on Farnell, they don't even put temperatures in their parametric search and what they do put in is invariably wrong as i told their rep on the phone the other day. Problem with RS is they may have it today but not tomorrow and then i have to wait 2 months, I'm having to make design decisions based on the availability of parts at my temp range and if the availability is going to change every other day I may as well stop designing.

[Kremmen]

I need to regulate 20-30V down to 15V in an ambient of up to 83C but also down to -46   

So first huge problem is the 83C because the zener de-rates badly, i could use more than one in parallel but then i now need so much more current to make them all work and higher wattage zeners have such bade tempco's that it's pointless and they need even more minimum ballast current.

Any ideas ? I only need 12mA and it only needs to be stable +/-2V, I can make the ballast resistor so that at 20V input my output drops to 12V because if i try and get the correct resistor for 15V @ 20V input when i am at 30V input that is 3x V and 9X W 

I could use a voltage regulator but again can't get any that do -46 only -40 and I have no idea what will happen at  -46, again it don't need to stay accurate, just not go haywire.

If you have such relaxed accuracy requirements, why not do something like this:
-firstly, do a 1 transistor constant current source to drive the zener into a reasonable operating point.
-Next, use the zener voltage to create a 1 transistor "regulator" to off-load the zener from the output current. You know the trivial circuit i am thinking of - a resistor (here the CC source), zener and NPN transistor.
Surely you can find a combo of common components to survive that temp range.

[Simon]

Yes your right i know the one you mean, the load will be pretty constant so it's just the variation of supply voltage that causes a problem, I think i have a zener that will do it now

[Simon]

Ah yes i remember, now, I can't use a regulator because I need up to 60V input as the TVS's can clamp at up to 48.5V so normal regulators usually can't cope with that.

[AndyC_772]

if the availability is going to change every other day I may as well stop designing.

Step away from the computer, hang up your lab coat, and go home.

Component availability is *always* changing. Choosing parts that have a good track record of availability is a skill, and a design criterion along with things like cost and functionality. But you can never, ever guarantee it, and part of an engineer's job is to make sure other people in the organisation understand this as well.

[Simon]

What i mean is RS is fast demonstrating a track record of not caring if stuff is in stock, unfortunately parts that work down to under -46 are already hard to come by so it's not like you find a drop in replacement just like that. Being rarer I'd expect these parts to be in enough demand to warrant keeping on the shelf. With digikey they not only tell you how many are in stock but how many more they can call on from the factory, their information is much more detailed.

[amyk]

-46 to 83 temp range, 60V input, 15V output@12mA, Digikey, that was easy...

LT3010MPMS8E#PBF
LT3013MPFE#PBF
LM117HVH/NOPB
LM117HVK/883 (probably overkill)

[Simon]

Thank you, the first one looks good although expensive but I'm struggling to find how many watts it will dissipate, I'll have to study the datasheet carefully as i have with everything selected so far and there is often a "catch"

[Kremmen]

Ah yes i remember, now, I can't use a regulator because I need up to 60V input as the TVS's can clamp at up to 48.5V so normal regulators usually can't cope with that.

When i wrote "regulator" i did not mean an actual regulator chip. I meant the 3 component circuit with an NPN as the pass element. You can have as high an input voltage as the transistor Vce max allows - and some allow a lot. See attached for the circuit i had in mind. The zener will be off-loaded to the extent you wish so its power dissipation is a non-issue.

[Simon]

Yes i understood that.

[T3sl4co1l]

Zeners don't have to be run at a minimum current, the voltage is just somewhat lower (e.g., a 1N5242B might be 11V at 1mA, 12V at 10mA, etc.).  This is true of all avalanche devices (Vz > 6V); true zeners are operationally defined by their low voltage drop (under 6V) and much softer knee (a 3.3V device might be only 2V at 1mA).  For lower voltages (or more accuracy, assuming the temp range is good with it), you want a bandgap reference.

Tim

[rob77]

eventually if someone want's a sharper "knee" then a transistor with reverse biased CE and floating base would to the trick (for very cheap) the only major issues are:
1. voltage ranges approx 6V - 13V (at least i didn't ran into a transistor not in that range)
2. failure mode is often an open circuit - compared to the short circuit failure mode of a zener (thanks to the metallurgic bonds of the zener).

[T3sl4co1l]

B-E junctions aren't nearly as good as proper zeners, in any of the measures a zener is good for, from regulation to reliability.  Perhaps noise, if you consider that a good thing.

Tim

[Rufus]

I need to regulate 20-30V down to 15V in an ambient of up to 83C but also down to -46   

Any ideas ? I only need 12mA and it only needs to be stable +/-2V, I can make the ballast resistor so that at 20V input my output drops to 12V because if i try and get the correct resistor for 15V @ 20V input when i am at 30V input that is 3x V and 9X W 

12mA @20v in needs a dropper resistor of about 417R. At 30v in that will draw 36mA.  That's a bit more than 0.5W in the resistor and 540mW in the zener (if you are pulling 0 not 12mA from the supply).

A 1.3W BZX85 glass zener (Fairchild) derates by about 300mW at 83C ambient. 540mW into a 1W rated zener doesn't look like a problem to me and even less if you use 2 7.5v zeners in series which will also give you better temperature stability.

[rob77]

B-E junctions aren't nearly as good as proper zeners, in any of the measures a zener is good for, from regulation to reliability.  Perhaps noise, if you consider that a good thing.

Tim

sharp knee is the benefit of a reverse biased CE and that way i formulated my previous post.
i'm not saying it's better than zener, i'm saying it has a sharper knee than zeners do

[T3sl4co1l]

sharp knee is the benefit of a reverse biased CE and that way i formulated my previous post.
i'm not saying it's better than zener, i'm saying it has a sharper knee than zeners do

I doubt it's actually sharper, or at least, sharper for the same current density.  Amp for amp, you might be right: in other words, the E-B junction looks like a much smaller zener than anything normally sold as a zener, so an E-B at 1mA behaves like a regular zener at 100mA.

I'll have to give this a look this evening, I haven't played with E-B's enough anyway.  Might as well verify that they're as bad as I think they are.

Tim

[Simon]

I took a lot at a couple of regulators on digikey, one was so small that I doubt it has the power handling that was conveniently omitted and plenty said about how it could take sporadic high loads but not continuous in-out levels or consistent high currents, another was derated to about 25% @ 85C so pretty pointless, Therefore I'm afraid that I remain with my SMPS design but I might tighten up that resistor value a bit and let the zener ride loose, it's really only there to help deal with input voltage extremes.