Maximum continuous drain current vs. RDSon Max

[hazuki]

Hi all. I have a question about a datasheet: http://www.infineon.com/dgdl/BSZ067N06LS3_rev2.0.pdf?folderId=db3a30431ddc9372011ebafa04517f8b&fileId=db3a30431ddc9372011ebb0f41537ff6

This part has a maximum Id rating of 20A when it's fully on and Tc = 25C. However, it's RDson can drop very low to under 6mOhm when driven hard. Graph 6 on page 5 shows how RDson varies, and that graph extends up to 80A.

How is the maximum Id rating determined? I have seen other parts that are advertised with more than double this Id rating yet have a similar package and their RdsOn is even larger than this one, for instance, this part: http://www.ti.com/lit/ds/symlink/csd18537nq5a.pdf

What gives? If I am selecting a mosfet for a switching supply, should I opt for a higher advertised Id or a lower Rdson?

Thanks

[Niklas]

Check graph 3, Safe Operating Area, in the datasheet from Infineon. Pulsed drain current capability when the package is kept at about room temperature is significantly higher than constant on (DC). Back to the table on the first page and read the fine print about the 14 A rating.

The datasheet from TI has similar data. ID in the table on the first page and the graph in Figure 10.

What is most important is to check under what circumstances the current rating apply and here is where the fine details matter. Are they mentioning Tj (junction temp), Ta (ambient temp) or Tc (case/package temp)? I would suggest that you look at the "Mounted on PCB" numbers instead and add some safety margin depending on the worst possible operating conditions. How many PCBs, not the outside of the product, are kept at 25'C or below at all times?

There was a discussion about this about a week ago here in the forum. If I remember correctly someone mentioned how the semiconductor manufacturer perform the testing with the MOSFET dipped in some liquid.
 

[rx8pilot]

For a switching supply consider the gate capacitance as it will have an impact on how long it takes for the FET to turn on/off. The max drive Rds-on will have little significance if the FET spends very little time in that condition. A controller with big drive capability and low-ish capacitance on the FET will ensure very little time is spent in the linear region of the FET where you create heat and lose efficiency. At low switching frequencies this may be less of a consideration. Also, make sure the controller has some energy to deliver to the gate as well. While the gate ideally flows no current, its capacitance needs to be filled as fast as possible. Put a 10uf cap in parallel with a .1uf de-coupling cap right at the controller input and keep the gate trace short. That will help ensure that the drive capability of the controller is fully utilized.

I am not an expert on Rds-on vs capacitance, but it seems that the ultra-low Rds-on MOSFETs have higher capacitance on the gate. This is no problem for DC switches, but could be a limitation in a switch mode PS. Who cares that you have Rds-on of 3 mOhms if the rise time of the gate is too slow to ever see it? I have only designed a few SMPS's and it took a few hours of head scratching to get a fast rise-time on the gate. That allowed higher-frequency, a smaller inductor, and better efficiency. Smaller and cooler in the end.

I am not an EE, but this has been my observation.

[T3sl4co1l]

Those single digit mohm parts are almost never worthwhile.  They're intended more for battery management where switching is slow or infrequent.

In a switching converter, you have to balance Rds(on), Qg, Coss and the diodes' Vf, t_rr (if applicable), Cjo and Irrm (if a concern) for optimal results.

Tim

[rx8pilot]

Those single digit mohm parts are almost never worthwhile.  They're intended more for battery management where switching is slow or infrequent.

This is exactly how I use those ultra low mOhm parts - DC routing. Probably switched a few times per day and they stay cool at heavy loads with no effort of heat sinking. I am very thankful for them in that situation.

In a switching converter, you have to balance Rds(on), Qg, Coss and the diodes' Vf, t_rr (if applicable), Cjo and Irrm (if a concern) for optimal results.

This is where I really started to study MOSFET characteristics. For my initial prototypes of both switching supplies, I naively chose the lowest Rds-On device and was disappointed [and a little perplexed] at the low performance of the circuit. I worked my way up to being able to choose a device that considered the range of characteristics that combined offered the performance I was after. Turns out, design and building SMPS is a LOT harder than building a switch. Go figure.

[splin]

I'd suggest you download the Fairchild Switching Loss Calculator Tool spreadsheet here (you'll need to register though):

https://www.fairchildsemi.com/design/design-tools/switching-loss-calculation-tool/

It's a bit long in the tooth, only calculates for a buck converter (synchronous or not) and only has older Fairchild mosfets and drivers in its tables. However you can add your own easily on the 'MOSFETDatabase' 'ControllerDriver' tabs. There's some help in finding the values you need from the datasheets - especially the various gate charge elements. There's a document explaining the calculations which is well worth studying - look for AN-6005

The good thing is that all the calculations are explained in the document and editable in the spreadsheet so you can customise it if needed. It allows you to select the number of paralleled switching FETS and synchronous rectifier FETs. It works out the efficiency for a range of output currents and breaks down most of the losses (but not inductor core losses or capacitor ESR losses) such as diode conduction and reverse recovery losses, FET switching and conduction losses etc.

And for your purpose, having told it the thermal resistance from junction to ambient of the FETs (ie. Rth(junction to case) from the FET datasheet + Rth of heatsink), it iteratively calculates the FET die temperatures - which it needs to do as RDSon varies with temperature.

Its a good learning tool to play with and see just how drasticly the efficiency varies depending on FET type and numbers, switching frequency, driver etc. and often not how you'd expect. Eg. using a lower RDSon switching FET reduces the conduction loss when it's on but invariably the switching loss will increase as its input capacitance will be greater and thus slower to switch. Lowering the switching frequency will reduce switching losses so the lower RDSon FET may become advantageous.

The main problem I had was knowing what value to use for 'Schottky capacitance' (if you use one) which you can specify on the 'Output' tab before running the calculation from the 'EfficiencySummary' tab where you specify most other parameters. The problem is that the reverse biased capacitance varies with voltage so what value do you use from the shottky diode datasheet?

IRF has a nifty online tool for buck convertors where you specify the input and output voltages, current, switching frequency etc. It calculates the losses using a wide range of different FETs, and number of paralleled FETs, and presents a list of the combinations which you can order by efficiency or cost. You can then examine the individual losses for a particular solution. It's rather limited though (restricted input voltage range) and obviously only has IRF FETs. It does make an allowance for inductor and capacitor losses and shows you what efficiency could be achieved (but real world would almost certainly be a bit worse).

Other manufacturers have online tools but they have the same problem that you are restricted to their devices (naturally) but more importantly can't modify it for your specific circuit/design 'features'.

[suicidaleggroll]

Those single digit mohm parts are almost never worthwhile.  They're intended more for battery management where switching is slow or infrequent.

That's a bit of a self-contradiction...they are very worthwhile when they fit the application, and not worthwhile when they don't, just like any other part.  For all we know DC routing is the OP's intended application and one of those parts would be perfectly suited.

[T3sl4co1l]

Those single digit mohm parts are almost never worthwhile.  They're intended more for battery management where switching is slow or infrequent.

That's a bit of a self-contradiction...they are very worthwhile when they fit the application, and not worthwhile when they don't, just like any other part.  For all we know DC routing is the OP's intended application and one of those parts would be perfectly suited.

To be perfectly pedantic, that's true;

If I am selecting a mosfet for a switching supply

is hypothetical, not a direct statement.

Moral of the story, understand all the parameters they provide in the datasheet, and select a device which satisfies your requirements the best.

Tim

[johansen]

Moral of the story, understand all the parameters they provide in the datasheet, and select a device which satisfies your requirements the best.
Tim

moral of the story.. read everything implied by the post, and select an interpretation that best suits your own ego.

yes those sub milliohm mosfets work great for my 500hz chopper...

[hazuki]

Thank you all for the replies. I am now understanding the tradeoffs between RDSon and gate charge. Looks like I have much more to read up on.

Thank you

[splin]

Those sub-milliohm FETs tend to particularly useful for synchronous rectifiers where the predominant loss is conduction loss.