Cooling an SMA

[Opticalworm]

Hi all,
I'm currently designing a zener regulator using (0BZG05C8V2 http://www.farnell.com/datasheets/69919.pdf) and i'm hoping to dissipate around 3 watt. Does any one know of a practical way to cool down an SMA (DO-214AC)?

[armandas]

If you check your datasheet, you will see that 3W dissipation is an absolute maximum for this diode. Plus, with a junction-lead thermal resistance of 30 K/W you will have a hard time finding practical way to cool it. Why don't you try and find a part in a larger package?

[Zad]

Put it on the base of an NPN transistor and use it as a shunt?



http://www.circuitstoday.com/zener-controlled-transistor-voltage-regulators

[mikeselectricstuff]

The only way you'd get anywhere near 3W without it unsoldering itself would be on an aluminium clad board, or with a lot of forced air. If you really need to dissipate that much, spread it over several series devices, or use a heatsinkable package.
 

[alm]

Funny how they specify a junction-ambient thermal resistance of < 30 K/W for dissipating 3W, but the thermal resistance between junction and lead is already 30 K/W, so you wouldn't be able to meet the spec even if you soldered it to a huge heat sink. Guess they use forced air to achieve that.

I agree with the others: use some other device (eg. transistor, or regulator) to dissipate the power, or get a zener in a package with a lower thermal resistance.

[Afrotechmods]

3 watts into an SMA package is really, really high. I would strongly recommend you switch to an SMC diode, or try out Zad's idea. Either way, you still have 3 watts to dissipate, so you'll need 2-3 square inches of copper plane to act as a heatsink.

[Opticalworm]

Thank you all.

Sorry Zad I'm afraid that my design has to use Zener diodes has mean to limit voltage, this is due to the nature Intrinsically safe equipment which sees simple components as infallible, but thank you for your suggestion.

Thank mikeselectricstuff, I will use several Zeners in parallel to share that power and hopefully reduce the power the zenner sees.

[Zad]

Sounds to me like a VERY narrow failure mode analysis!

[mikeselectricstuff]

Thank you all.

Sorry Zad I'm afraid that my design has to use Zener diodes has mean to limit voltage, this is due to the nature Intrinsically safe equipment which sees simple components as infallible, but thank you for your suggestion.

Thank mikeselectricstuff, I will use several Zeners in parallel to share that power and hopefully reduce the power the zenner sees.

I've done some IS stuff in the past - increasing rating by paralleling isn't very practical, as you can't guarantee that the load will be spread evenly between parallel zeners - as well as tolerance issues there is the possibility of one or more failing open, increasing dissipation in the others.
They have to be lower-voltage ones connected in series. To get the required fault coverage you may need 2 or 3 seperate parallel strings. See if you can reduce the worst-case power going in, or split it into seperate zones, as this will make things a lot easier all round. 3W is a lot of power in the IS world.
Using a device with a higher inherent power rating is likely to save you a lot of grief.

[Opticalworm]

I've done some IS stuff in the past - increasing rating by paralleling isn't very practical, as you can't guarantee that the load will be spread evenly between parallel zeners - as well as tolerance issues there is the possibility of one or more failing open, increasing dissipation in the others.
They have to be lower-voltage ones connected in series. To get the required fault coverage you may need 2 or 3 seperate parallel strings. See if you can reduce the worst-case power going in, or split it into seperate zones, as this will make things a lot easier all round. 3W is a lot of power in the IS world.
Using a device with a higher inherent power rating is likely to save you a lot of grief.

Thanks mikeselectricstuff IS is harder than what I originally though and on top of that I now realise that I was going on about it wrong. I was trying to use the Zener as a mean to regulate the system voltage to (Say 30v @400mA+Iz).

I Now know that I should really be using the zener (or TVS) to short circuit the supply when it goes above the normal voltage operation . I'm a little fuzzy as too how I would go about picking and matching the correct TVS diode to a Fuse. But I've post the question into a seperate Post https://www.eevblog.com/forum/index.php?topic=2545.0
 

[Frant]

Thank you all.

Sorry Zad I'm afraid that my design has to use Zener diodes has mean to limit voltage, this is due to the nature Intrinsically safe equipment which sees simple components as infallible, but thank you for your suggestion.

Thank mikeselectricstuff, I will use several Zeners in parallel to share that power and hopefully reduce the power the zenner sees.

I've done some IS stuff in the past - increasing rating by paralleling isn't very practical, as you can't guarantee that the load will be spread evenly between parallel zeners - as well as tolerance issues there is the possibility of one or more failing open, increasing dissipation in the others.
They have to be lower-voltage ones connected in series. To get the required fault coverage you may need 2 or 3 seperate parallel strings. See if you can reduce the worst-case power going in, or split it into seperate zones, as this will make things a lot easier all round. 3W is a lot of power in the IS world.
Using a device with a higher inherent power rating is likely to save you a lot of grief.

You are quite right, but the power rating is not always the most reliable parameter, especially for diodes. It all depends on the way it is measured and presented in the datasheet. Doing an IS project myself, I ran into a similar trouble with Zener diodes. Common types of diodes in SMD packages (e.g. SMA) are poorly designed in terms of power dissipation. Thus their real RthJA is very high and the manufacturers tend to publish "optimistic" thermal data obtained under unrealistic or not clearly defined conditions. Fortunately, there are some newer diode and Zener diode types that can dissipate more power than the older ones of the similar size.

[cyberfish]

If you want to parallel diodes/transistors, you'll also need to make sure they have positive temperature coefficient - resistance goes up when temperature goes up.

If resistance goes down when temperature goes up, if one diode takes slightly more load and heats up, the resistance will decrease, and it will take even more current, which causes it to heat up more, and take even more current...

Eventually it will take all the current, and pop. Then it goes to the next one, etc.

[Frant]

If you want to parallel diodes/transistors, ...

"Current sharing" by paralleling components of individual rating lesser than the required rating is not possible in circuits upon which the intrinsic safety (IS) depends. If an infallible shunt safety assembly is to be built, at least two parallel paths of diodes or Zener diodes must exist. Each of the used diodes or Zener diodes must be rated to carry at least 1.5 times the total current and to dissipate at least 1.5 times the power in normal operation. Faults must also be taken into account according to the IS standard.