Power MOSFET Footprint Run Amok :-(

Anyone want some MOSFET entertainment on Sunday evening? Got time and some good IPA?

I was casually researching some large sized MOSFETs and smacked into some disinformation which makes me want to laugh or cry…not sure which to do first. Why limit disinformation to American/Russian/British politics? If I can report this without typos to add to the confusion I will be amazed. I started out sorting some 40V to 60V FETs on the Digikey website.

I am looking at the NTMTS0D6N04C from ON Semi. My first reaction is “That is pretty good. It looks like ON has gotten into making FETs in 5 x 6 mm footprint similar to Infineon and Vishay.” But wait. This is not 5 x 6 mm. It is roughly 8 x 8 mm. OK it is a big FET. But that is not all; they put the gate connection on pin 1 instead of pin 4 as on the Vishay and Infineon 5x6 mm packages.

OK. I wonder if any other manufacturer puts FETs in the same package and pinout?

According to the sorting table at Digikey, for this FET the package is an 8-PowerTDFN and the supplier device package is 8-DFNW. According to the ON datasheet, the package is DFNW8, and Case 507AP. On the datasheet front page, it is also called “Power 88”.

That is all simple enough, right? (sarcastic) But I will try sorting the table at Digikey. I try filtering to the 8-PowerTDFN package/case. (Use the “more filters” button.)

Hey! I find the Infineon BSC007N04LS56ATMA so I open up the datasheet. Which says that the package is their TDSON-8 FL. And that is a 5 x 6 mm package with the gate connected to pin 4. No Problem! (sarcastic again.)

So let’s try the “DFNW8” designation which is listed on the ON Datasheet. I do not see that under either “Package/Case” or “Supplier Device Package” filters but I do find “8-PowerWDFN” under “package/case” so let’s try that. Arbitrarily picking out a couple from the list:

Here is another 5x6 mm package; an ST STLD200N4F6AG and an ON Semi NTTFS5C454NL which is in a 3x3 mm package.

I think I could try to follow this trail further, but I do not want to end up somewhere such as either Alice in Wonderland or North Korea. But if you THINK that you do not understand then you do “get it” and in fact you do understand… Just be VERY CAREFUL in selecting and using these MOSFETs!

Confusion begetting confusion…

1 Like

Well MOSFETs never really came in anything that could be considered a single standardized package. Even with TO-92 you had all possible pin assignments available. There is a reason why there is more than one generic symbol for FETs in the KiCad library.

Also the TDSON-8 FL name is a manufacturer naming. You need to check its JEDEC name and see if that name agrees with the other manufacturer’s statement to know from the name if two packages are equal.

3 Likes

Thanks, Rene. This is interesting. While I have used JEDEC designated components such as 2N3904 and 1N4148 for my entire career, I have mostly forgotten about JEDEC with regards to anything less than 30 years old. In response to your post, I did a quick and non-rigorous check.

I searched a couple of IC datasheets for the term JEDEC and found JEDEC package designations for the first two IC data sheets I looked at. These were TI and LTC-ADI ICs which came up in “recently opened files” for my .pdf viewer. These JEDEC designations were MO-153 and MO-220.

Then I checked a BSC prefix MOSFET from Infineon and a TSOP6 MOSFET from Vishay. Neither datasheet indicated a JEDEC package designation. I thought that Nexperia may be different so looked at a datasheet for a PSMN026-80YS MOSFET in “SOT669” (as shown in the datasheet) package. The Nexperia datasheet includes the table which I now uploaded as an image:

This includes the JEDEC MO-235 designation which ought to be useful. But when I check the Digikey listings for this part, they do not even mention that. If I use it as a search term it comes up with nothing. “SOT669” also gets me nowhere. But insert a dash to get “SOT-669” and that scores a bunch of hits including the Nexperia MOSFET.

To any of you who may feel marginalized by the engineering “powers that be”, this all looks like a conspiracy to confuse us or at least make our jobs more difficult. I am mostly kidding but those are certainly the effect. It certainly makes on shake one’s head. We are no better than politicians in this way. :frowning:

2 Likes

I highly doubt it is a conspiracy by design. Sure the manufacturers are aware that there are “positive” side effects but my guess is that modern technology simply moves forward faster than the standardization organizations can work.
So manufacturers make up their own names before there is a standardized name for that thing. And if there is finally a standard declared later, than it is added as a footnote instead of replacing ones own name.

The fact that the standardized names are not that readable might not be much help either.

2 Likes

“Never attribute to malice that which can be adequately explained by stupidity.”
Hanlon’s Razor

2 Likes

One Standard to Rule Them All, One Standard to Find Them, One Standard to Bring Them All, and in the Darkness Bind Them?

Such are the intentions of some misguided souls, though it ain’t gonna happen! :sunglasses:

Thanks all for the good comments. But at risk of dragging this out too far; think of when was the last time you saw something like a contract that was written in plain (English or perhaps other language)? Lawyers intentionally write in a way that it takes a lawyer to read it. Take a little bit of that idea, and a little bit of the idea “Too many cooks (marketing engineers) spoil the dinner” or similar. One marketing engineer at one company might have a good plan. But put them all on Digikey (or Mouser) and you get nonsense.

Full disclosure- That was my 2nd job at one of the said disty’s (I was first on the phones as Phone Tech). Circa 2004, we did not have anyone technical in the web group, so I felt like a kid in the candy store, with being able to add all of the parameters/filters that I had often wished were there.

It quickly became very clear which products lacked uniform standards. This image had a lot more meaning to me…image

Inductors are crazy; their saturation current is taken at many different temperature points with no apparent standard. (I recall suggesting we set up a test jig and offer an in-house test to have some universal test, maybe someday).

Lighting LEDs (>20mA) are another; some had their lumen rating at 25°C, and others at 85°C, but rarely both. Now we just list the test temp as a separate parameter so customers can do their own apples to oranges.

I’m not in that department anymore, but I have plenty of respect for anyone trying to normalize data. (Imagine how much better Amazon would be if they offered normalized filters, that you could trust.)

2 Likes

I really like your illustration. Yes most online selection filters are really bad and I complain about Amazon and Home Depot too! But I expect better from electronic component distributors. However even on ON Semiconductor’s website for MOSFETs I saw them mixing Rds ohms with milliohms. So the disty does not have an easy task.

PNP transistors and P channel mosfets drive me up the wall, with Vceo/Vds sometimes negative, which plays hell with range searches

2 Likes

Your comment reminds me of an old Fairchild datasheet for a 7912 type 3 terminal negative voltage regulator. The minimum output voltage was -12.6 and the maximum was -11.4. If we follow that line of reasoning, then -1000 Volts is a very low voltage so it should be safe to handle. I would not volunteer to test that theory. :slight_smile:

1 Like

1k is not too bad. Try 4k DC. ‘Great stuff’.

I just ran into another good example of “footprint run amok”. May be applicable to MOSFETs but maybe not specifically. I wanted to compare SC88 and SC70-6 packages (for a dual schmitt inverter; TI versus Nexperia) and found this:

https://www.onsemi.com/site/cdroms/micropackages/pdf-docs/onsemi/SOT563%20footprint-prof%23B3380.pdf

Note that the packages indicated as similar are not necessarily identical. In my specific case ( SC88 and SC70-6) there is at least a difference in tolerance. Anything to make our lives more confusing.

I’m saying more than I actually know . . . but . . . it’s an unusual case where the differences among those packages matters. A few incarnations ago I worked for a company that did moderately high-rate production, with high-speed automated equipment. It wasn’t every day, but not uncommon for the production engineers to ask for seemingly inconsequential layout changes - e.g., add or remove 2 or 3 mils to a pad; shift a trace by a few mils; increase a spacing by a mil or two. I don’t recall the exact values they were looking at, but it was something like reducing re-work from 10 units per 20,000 produced, to 8 per 20,000. That’s the kind of environment where you start to pay attention to package tolerances, and how tightly one manufacturer holds his production within the tolerance window versus an alternate manufacturer.

Dale

3 Likes

Sounds like you have “been there and done that.” Yes I did not mean to imply that the tolerances were of no consequence. But what is more my modus operandi is to make a footprint which takes several really different packages, such as SOT-3 or SOT6. In 2000 I designed power supplies for a Taiwan company which did power supplies. The pcb layout was my responsibility, but the “hands on” aspect of that was one of several women who ran (I think it was P-CAD). I do not know who defined the footprints.