Bsrgvty

The official JEDEC standard to number pins on a TO-92 package is the following:

But in many datasheets from Texas Instruments, this order seem reversed. For example, this is a screenshot from the datasheet for the LM185 :

Link to the full datasheet :
http://www.ti.com/lit/ds/symlink/lm185-1.2-n.pdf (page 1)

Texas Instruments's official TO-92 documentation doesn't help since there is an ambiguity as if the drawing is from the top or the bottom : http://www.ti.com/lit/ml/msot002d/msot002d.pdf

In all the datasheets where the pinout is reversed, they used a view from the top or the bottom. In some datasheets the pinout is correct, but they used an isometric view which eliminate all ambiguity. This could be the source of the confusion, but I just don't see how I could misinterpret "Bottom view" or "Top view" for a TO-92 package.

Is Texas Instruments wrong or I don't understand what a bottom view is?

To prove that this is not an isolated mistake in a single datasheet, here are few links to some Texas Instruments datasheets where this happen :

http://www.ti.com/lit/ds/symlink/tl431.pdf (page 4)

http://www.ti.com/lit/ds/symlink/lm285-2.5.pdf (page 3)

http://www.ti.com/lit/ds/symlink/lm4040-n.pdf (page 4)

The LM385 was originally produced by National Semiconductor. The earliest reference I can find is from their 1980 Linear Databook. It shows the same 'wrong' pinout. Other manufacturers whose datasheets have the same pinout include Motorola and Telcom Semiconductor.

So it seems the 'wrong' pinout originated at National Semiconductor, and has been copied by second-source manufacturers. Texas Instruments acquired National Semiconductor on September 23, 2011. This would have given them the opportunity to use National's datasheets for parts they were second-sourcing.

I don't see any ambiguity in any of the datasheets, and the only one that disagrees with the conventionally accepted numbering scheme is the first one. The others show the pins in solid lines (not dashed) so they are clearly bottom views.

By the way, TI was not the originator of the LM385 chips, rather they bought National Semiconductor. You can find the same numbering scheme on NS datasheets, for example here. The part is rather old, the same pin numbering is shown on page 2-47 of NS's 1982 databook (the oldest one I happen to have on hand), it probably dates back to some time in the 1970s. The databook refers to NS Package number Z03D, and you can find a datasheet for that package here

As you can see, it's consistent with your LM385 datasheet, but inconsistent with more modern numbering.

A relevant question might be when (or even if) numbering for the TO-92 package was standardized. I don't recall it being standardized during those years, instead we would use E-B-C or similar letters for the pin designations.

For what it's worth, alternate sources of the LM385 et al may disagree on what number to use for a given pin, but they certainly agree on the pinout, however you number them. The part is somewhere in the range of mature to obsolescent in life cycle so it's not much to worry about provided you follow one data sheet and don't mix datasheets.

You could hunt down the JEDEC TO-92 standard yourself. I've included the link, but you need to register (for free).

Yes, TI disagrees with JEDEC. But as long as your board works, does it matter?