I am a bit confused about Figures 3 and 4 in the datasheet of the DMN3010LSS MOSFET transistor: Why do they show how the Rds(on) varies as a function of the ambient temperature TA?

I thought that the junction temperature Tj was important for the Rds(on).

Figures 3&4

  • \$\begingroup\$ With changes in ambient, junction changes too \$\endgroup\$ – PlasmaHH Mar 16 '17 at 15:08
  • 1
    \$\begingroup\$ I added the link. My guess is that TA is a more practical choice as the MOSFET often can only survive certain conditions a pulsed mode. For example: Fig 4, TA = 150 C, Id = 20 A then Rds = 16 mohm. That means P = 6.4 W ! With 50 C/W thermal resistance the junction would be more than 450 C, violating TJmax (150C). So this can only be done in a pulsed mode. When the pulse starts the junction is also at TA and the conditions apply. So for these extremes, the junction must not heat up (much). \$\endgroup\$ – Bimpelrekkie Mar 16 '17 at 15:14
  • \$\begingroup\$ You must remember that when a MOSFET is hard-on it is just a resistor. This is why you can put MOSFETs in parallel. \$\endgroup\$ – skvery Mar 16 '17 at 15:53
  • \$\begingroup\$ If it read junction temp, and you were using many shunt FETs to minimize temp rise, would you be likely to ignore the effects of ambient temp? I think it forces one to compute self heating of parallel FET temp rise on each other's ambient so as not to make thermal calc errors. Rja applies in all cases, but Ta is the near field ambient. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Mar 16 '17 at 16:10

It's rather strange, indeed.

As usual, the datasheets states that \$T_{j,max}=150\ ºC\$ so operating it at \$T_A=150\ ºC\$ would require \$R_{\theta,JA}\approx 0\$.

Is that possible? In my opinion, NO. Why? because even with a big heatsink and forced air convection so that \$R_{\theta,CA}\approx 0\$, there will still be a a temperature gradient between junction and case, of typically \$R_{\theta,JC}\approx 2-3\ ºC/W\$ for that kind of case.

From the datasheet, we can see that the power dissipation would be \$P_D \approx 20^2\cdot 0.016 = 6.4\ W\$, so even with an optimistical \$R_{\theta,JC} = 2\ ºC/W\$ we would end up with \$T_j \approx 150\ + 2 \cdot 6.4 = 162.8\ ºC > T_{j,max}\$, which isn't a feasible operating condition.

Thus, if the figure isn't representative of static operating conditions, then it should be, as FakeMoustache says, a pulsed operation mode. Take a look at the transient thermal response of the device:

Transient thermal response

But then, it would take BOTH a very low duty cycle AND a very low pulse duration to achieve \$R_{\theta,JA}\approx 0\$ even with a heatsinked device with forced convection!

So we are left with only one conclusion: the curve in figure 3 is rather absurd from a practical point of view. You should regard it as a theoretical limit that just tells you what will be the worst-case \$R_{ds, ON}\$ that you must expect.

Anyway, I would agree that it is a rather misleading way to disclose performance information of a device!


This device is spec'd at TA = 25°C.

When a device is spec'd at TJ it is more difficult for the engineer to calculate TJ from TA or Tcase.

The manufacturer is trying to make your job easier.

I see this more and more in recent time. Some devices will now specify a specific point on the case to epoxy the thermocouple when measuring temperature.

I prefer a device to be spec'd at TA or Tcase

The datasheet for this device seems to have some conflict on the high temp side. The way I look at that is that 100°C+ is not a good temperature zone to be operating in anyway. With Thermal Resistance, Junction to Ambient of 50°C/W, I would be looking for a different part. I find thermal resistance one of the most important device characteristics when selecting a device.

  • 1
    \$\begingroup\$ I totally agree with what you are saying regarding the \$T_J\$ and that it is much more difficult to calculate in comparison to \$T_{case}\$. Specifications regarding \$T_{case}\$ definitely make our lives easier. On the contrary, the ambient temperature is vaguely defined, I would say. And definitely not suitable for accurate calculations. \$\endgroup\$ – nickagian Mar 17 '17 at 15:06
  • \$\begingroup\$ I'll take ambient over junction any day. Definitely case is the way to go especially with a metallic pad on the case for the temp measurements. I always found 25°C to be the unrealistic spec. I see more and more at 85°C these days. \$\endgroup\$ – Misunderstood Mar 17 '17 at 15:12
  • \$\begingroup\$ @Misunderstood "The manufacturer is trying to make your job easier". Mmh... OK, but is this actually the case of the figures in the datasheet pointed by the OP? The numbers don't add up. \$\endgroup\$ – Enric Blanco Mar 17 '17 at 16:43
  • \$\begingroup\$ @EnricBlanco I did acknowledge this particular datasheet had issues when I said "The datasheet for this device seems to have some conflict on the high temp side." But the case / ambient vs junction is an industry wide issue. Its in the press and changes are happening. The OP did not ask about errors in the datasheet. The OP was asking about Ta vs. Tj being used to spec the device, i.e. "Why do they show how the Rds(on) varies as a function of the ambient temperature TA?" \$\endgroup\$ – Misunderstood Mar 17 '17 at 17:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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