2
\$\begingroup\$

Sometimes the spec sheet for a MOSFET or a BJT has a Safe Operating Area graph and sometimes it doesn't. When it doesn't, is it safe to assume that the transistor can handle the maximum current at the maximum voltage with continuous DC (depending on heat sink requirements, of course), or do I need to find the SOA curve somewhere else?

Don

\$\endgroup\$
4
  • 1
    \$\begingroup\$ Reaching the maximum current for any period of time whatsoever, at any voltage, requires unrealistic heatsinking for any MOSFET. \$\endgroup\$
    – Hearth
    Apr 19 at 15:18
  • \$\begingroup\$ The ratings for blocking voltage and conduction current are mutually exclusive logically. The FET cant both conduct and block other than during the switching instant. Generally: calculate the dissipation and check the thermal impedance for a realistic DC figure. This is mostly dictated by the package, not so much by the specs of the semiconductor junction \$\endgroup\$
    – tobalt
    Apr 19 at 16:14
  • \$\begingroup\$ Why would you need your MOSFET to operate at its maximum voltage while continuously drawing the maximum current? What kind of application do you have in mind? \$\endgroup\$
    – kaosad
    Apr 19 at 18:19
  • \$\begingroup\$ @eromlignot can you provide some examples? \$\endgroup\$
    – Voltage Spike
    Apr 19 at 21:44

3 Answers 3

4
\$\begingroup\$

If it's important in your application (such as switching an inductive load at substantial current), I would say you need to find a different part; one that has a manufacturer who provides the specifications and limits that are important to you.

It is most definitely not safe to assume there are only thermal limits.

If you must use such a part, it's safer to use a part that has (say for a BJT) Ic(max) a great deal (like 5:1 or 10:1) higher than your operating current.

\$\endgroup\$
2
  • \$\begingroup\$ Well, I'm often disappointed in the realistic constraints of an SOA, particularly for a MOSFET. The spec in the Digikey parametric search may give impressive currents like 5 amps for a small SMT component, only to find, in the SOA, that this is valid only at 0.1 volts, or only if it is a 1 us pulse with sufficient cooling time between pulses, which is useless for a DC application. I wish there were a parameter that would allow me to narrow down transistors without having to open dozens and dozens of spec sheets, only to find that the SOA has ridiculous constraints. \$\endgroup\$
    – eromlignod
    Apr 19 at 15:27
  • 1
    \$\begingroup\$ Well, 0.1V at 5A is 0.5W which will fry an SOT23 tout de suite quite predictably. The part of SOA you need to be concerned about (in that it doesn't show up elsewhere) is not the DC limits which are set by worst-case Rds(on) and thermal considerations but the pulse limits that are more restrictive than the thermal limits. And yes, it's true that a '5A' SOT23 MOSFET might only be really usable at 1A or less because you want it to work more than once under realistic worst-case conditions. Let alone the ridiculous hundreds-of-amperes numbers on larger parts. \$\endgroup\$ Apr 19 at 15:36
2
\$\begingroup\$

You seem to overestimate the problem! For the transistor SSM3J338R that you mention in the comment there should be no problem at all with the SOA, as we can see from the datasheet, in Fig. 7.14. You do however say that you need 15V and this transistor is rated for 12V, but to see that you wouldn't need the SOA curve, it's in the max. ratings table.

To be specific, if we would use this transistor to switch a current of 700mA (the value you mention) then:

  1. when it is switched on it will have a Vds of Rdson*0.7A so that's about 0.2V (using Fig. 7.5 of the datasheet), and thus it will dissipate around 0.14W. Acc. to Fig. 7.13, that's OK up to more than 100 degrees ambient temperature!
  2. When it is switched off it will see the supply voltage (assuming you will not switch off inductive loads without a protecting diode). Since the transistor has Vmax=12V I would only use it for supply voltage up to 10V, but again, that is just a max. rating problem, the SOA curve is still not needed to see this.
  3. During the switching you could for a short time have the combination of Vds up to the supply voltage plus the current of 700mA, and now we need the SOA curve of Fig. 7.14. Since the switching can be done within 1us, we can use the highest limit in that graph, which even allows 10A of current, and you only have 0.7A, so you see there is absolutely no problem here with the SOA curve.

This all assumes you use it as a switch. If you really need a transistor to handle 15V and 0.7A continuously for a long time, then it will dissipate 10W so you need one with a much larger package and with a heat sink. But again, the SOA curve is not needed to see that! That fact is immediately clear from the 1W power limit in the max. ratings table. Availability of the SOA curve simply does not seem to be your real problem...

\$\endgroup\$
2
  • \$\begingroup\$ I just picked that transistor at random because it had a conservative SOA. But I did learn something very important from you. I have been using the supply voltage on the SOA curve rather than the Vds/Vce drop...duh. This makes a big difference! Thanks. \$\endgroup\$
    – eromlignod
    Apr 19 at 18:47
  • \$\begingroup\$ Yes, I thought that the use case you have would be switching, which does indeed make a difference. But why do you say this transistor has a conservative SOA? For short time (100us) Fig. 7.14 allows 10 V by 10A, so that's 100W! And for longer time you just go gradually back to the DC values of the max. rating table. That is about the best SOA curve you can possibly imagine! For bipolar transistors you'll find much worse cases... \$\endgroup\$ Apr 19 at 20:10
1
\$\begingroup\$

Safe Operating Area is also sometimes called derating in datasheets, if you don't find any graphs or statement in the datasheet that specifies this area then you should apply by yourself some margin over the maximum operating conditions to ensure safety and durability of the component.

Here are some safe margins that you can apply if you have no indications in the datasheet :

For BJT transistors, do not exceed 30% of the maximum specified power, and do not exceed 70% of VCE, ICE and VBEinv maximum specs.

For MOSFET, do not exceed 70% of VDS and 75% of junction temperature. You can apply 100% of the maximum spec for VGS. In the case of switching power supplies, derating of these 3 values can be brought to 90% (mandatory for VGS).

Hope this helps.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ Well, it seems most of the transistors that I'm interested in require extreme deratings for DC. For example, looking at a Toshiba SSM3J338R, it touts a 12V, 6A maximum in the Digikey search. But when I look at the SOA, even if I derated my current by 70%, down to 1.8A, I'm only allowed a Vds of about 0.7 V. This is ridiculous. I need 700 mA at 15V continuous DC in as small a package as I can get. Must I really open hundreds of spec sheets at random? If I derate 90%, will I miss out on transistors that would have worked and will I still run the risk of blowing it up anyway? \$\endgroup\$
    – eromlignod
    Apr 19 at 15:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.