9
\$\begingroup\$

I have a general question regarding MOSFET Selection. I am trying to select a MOSFET for DC use. I am looking to replace a 5A 24V Relay with a N Type MOSFET.

The MOSFET would be driven from a micro so I would need a logic level gate. The micro is 5v Logic.

I am going to be mass producing these so cost is my main driver.

Most of the MOSFETs that I have come across do not have a DC area called out in the SOA curves. For instance the one I was potentially looking at was the IRLR3105PBF.

Datasheet here

Here are the params I looked at:

VDSS Max = 55V which is >> than my 24Vdc Bus so that is fine.

Power Calc - 5A*5A*0.37mOhm = .925W (High but I think a DPAK can handle that)

enter image description here

FIGURE 1 & 2 - VGS @ 5V -> VDS = 0.3V @ 25C (but the graph 20uS Pulse I want this to be DC?) VGS @ 5V -> VDS = 0.5V @ 175C (again I want this to be DC?)

SOA Curve

FIGURE 8 - Looking at VDS - 0.5V (Worst case) it only shows 1V. 1V can go up to 20A way more than I need for a 10mSec pulse. (I am actually confused on this should I just assume that I will have VDS of 1V looking at this?)

But then comes my main question I want DC where do I look for that?

Is this just a bad choice? (I get the feeling it is because no where in the data sheet does it talk about DC) What should I look for when searching Digikey?

TLDR How should I select FETs for DC use?

\$\endgroup\$
  • \$\begingroup\$ Note that you could use a MOSFET driver (chip, or DIY) between the uC and the power MOSFET. The requirement that your MOSFET switches with 5V (or 3.3V?) at the gate limits your options considerably. \$\endgroup\$ – Wouter van Ooijen Sep 4 '13 at 18:42
  • \$\begingroup\$ Agreed 100% I was just trying to keep the BOM cost as low as possible. @WoutervanOoijen \$\endgroup\$ – EE_PCB Sep 4 '13 at 19:50
  • \$\begingroup\$ Heavy-duty MOSFETS are not cheap. You seem to have 24V available. A small FET or transistor + a couple of resistors can lift your 5V to 10V, which might widen your options to cheaper MOSFETs, which might more than compensate for the extra components. Or not, but you won't know if you don't try that alternative. Systems design: evaluating the alternatives! \$\endgroup\$ – Wouter van Ooijen Sep 4 '13 at 21:17
2
\$\begingroup\$

If you need DC operation, you should really use a MOSFET that has a DC rating in its Safe Operating Area.

MOSFETs that don't have the DC curve may suffer from thermal runaway when used in DC applications and are intended or specified for switching applications only. Internal, local hotspots may occur and the MOSFETs may fail ("Spirito Effect").

The reason is a falling gate-to-source threshold voltage for a rising temperature, usually at low gate-to-source voltages. The details of this issue are usually not specified in the data sheets, so the only indicator is often the SOA diagram that has or doesn't have a DC curve. Fig. 3 in your MOSFET's data sheet looks like the point of thermal VGS crossover is a bit below 4 V. In my opinion, you are on the risky side when you use this particular MOSFET with a driver that can supply 5 V only. For a worst-case scenario, consider your supply to be on the low end (4.5 V), and allow some voltage drop for the driving stage. Sooner than you might like, you end up somewhere around 3.5 V.

Note that the absolute maximum ratings (25 or 18 A at 25 or 100 °C, respectively) are specified at a gate-to-source voltage of 10 V, when your MOSFET is fully on . They do not apply at lower gate-to-source voltages.

More about the background here: https://electronics.stackexchange.com/a/36625/930

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Is there a way to search for that? I looked at 5 or 6 different data sheets and they all had pulses for both the typical output characteristics and for the SOA curves? \$\endgroup\$ – EE_PCB Sep 4 '13 at 14:49
  • \$\begingroup\$ @EE_PCB Not that I am aware of a way to find it - neither in parametric search tables nor on the front pages of data sheets. \$\endgroup\$ – zebonaut Sep 4 '13 at 14:52
1
\$\begingroup\$

Have a look at products from Solid State Optronics. http://www.ssousa.com/home.asp The ones we're using (SDM4101, SDM4102) have an optoisolator built in but they're only 3.4A. I'm about to start testing a configuration with 2 in parallel for greater curreent capacity. The thermal characteristics of Mosfets means resistance increases with temp so if one starts to draw more current it will heat up, increase resistance and more current will flow through its twin. Or so the theory goes!

| improve this answer | |
\$\endgroup\$
0
\$\begingroup\$

They mention the maximum drain current to be 18A continuous at 100 degrees C. If your original relay never saw more than 5A continuous, you will be fine.

To answer your question: look at the continuous rating. It is at the top of the first page and also listed as one of the first electrical characteristics as an absolute maximum. Later on, it is in the source-drain characteristics table on the end of page 2.

It is important to do what you did and evaluate the power dissipation (RDSon * I^2) This looks like a reasonable FET. In a DPAK I imagine you will be soldering it to a PCB for heatsinking.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ That shows a 18A Cont with a VGS @ 10V. I will only have a VGS of 5V. Does that still apply? How would i derate this? Thats why i thought that it may not apply? @warren hill \$\endgroup\$ – EE_PCB Sep 4 '13 at 14:54
  • \$\begingroup\$ With VGS of 5 volts, you will be above the threshold. Judging by the pulsed graphs, the device will be sufficiently on to conduct 5 amperes. What isn't specified is the drain-source resistance. I recommend you purchase or sample some devices and experiment with them to determine their merit. \$\endgroup\$ – HL-SDK Sep 4 '13 at 15:21
0
\$\begingroup\$

The figures on the Absolute Maximum section cover DC continuous operation. The SOA curves show you can exceed these ratings for short durations but you can have 18 amp continuous provided you keep the case below 100C.

Just estimate the power from I^2 Rds_on. But remember Rds_on increases with temperature I usually allow for a 50% increase in Rds_on.

| improve this answer | |
\$\endgroup\$
-1
\$\begingroup\$

Lower your Rds at worst case drive voltage of say 4.5 for reliability and there are many similar cost FETs with 16A capability or smaller devices that are cheaper.

http://www.diodes.com/datasheets/DMN6040SK3.pdf. 50mΩ, 4.5V gate 16A. $0.20 @1 reel

| improve this answer | |
\$\endgroup\$

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.