What do date codes indicate precisely?

[fmzambon]

Hello everyone,

a few weeks ago I ordered a few 2N6422 BJTs from DigiKey. They were the last ones available, offered through their "marketplace" program directly from the manufacturer, Solid State.
Since they were unavailable everywhere else (unless one wants to order at least 100 from Microchip at $27 each ), I figured that they must be some new old stock gathering dust in a corner of the manufacturer's warehouse.

I placed the order for the available stock on May 2nd and the package arrived today (courtesy of the Italian customs ).
Well, take a look at the date codes:



(yes, one is upside down, all the electrons are gonna fall out )

2318, the same week when I placed the order!

They were listed as available before I placed my order, and now they are not any longer. So I highly doubt that the folks at Solid State fabricated a batch of 2N6422s dies just for me. But at least the date code printing must have happened after I placed my order.

Perhaps the manufacturer had bare dies in storage and packaged them into TO-66 packages when I placed the order? In that case the date code would represent the time that the silicon was packaged, not when the silicon was fabricated.

So, my question is: what do date codes indicate exactly? I was under the impression that they indicated when the actual silicon was produced, but perhaps they only indicate when the last step in the manufacturing process is carried out?

Just a note in case someone from Solid State happens to read this: I'm not complaining, at all. I'm just curious

Best regards,

Andrea

[ataradov]

For MCUs date codes indicate the date devices were tested for the final release. It may not even be the date they were packaged. This is done because the same die may be used in many device variants, so the final marking are not generally known until the last minute.

But there is usually a lot code identifier, which is unique to the wafer. Sometimes lot code includes the date code as part of it.

In this case "2318LF" may be the full lot code. But there is no way to tell the die manufacturing date from this.

[TimFox]

Could "2318" indicate the 23rd week of 2018?
Apparently, different manufacturers use "WWYY" or "YYWW" for the four-digit code (W = week, Y = year).

[tooki]

Could "2318" indicate the 23rd week of 2018?
Apparently, different manufacturers use "WWYY" or "YYWW" for the four-digit code (W = week, Y = year).

This is exactly what I was thinking.

Or it could be something even less obvious, like the 318th day of 2022.

Without a date code decoder from the manufacturer there’s no way to know for sure.

[TimFox]

Could "2318" indicate the 23rd week of 2018?
Apparently, different manufacturers use "WWYY" or "YYWW" for the four-digit code (W = week, Y = year).

This is exactly what I was thinking.

Or it could be something even less obvious, like the 318th day of 2022.

Without a date code decoder from the manufacturer there’s no way to know for sure.

The date codes with which I am familiar use week and year.  It becomes obvious with numbers >52 or >23, but it is usually stated on the packaging page of a good data sheet.
National (now TI) had a range of date codes, from 1 to 4 digits http://www.chipdocs.com/manufacturers/NSC.html  but usually used YYWW

[fmzambon]

The datasheet doesn't seem to indicate the date code format, unless I'm missing something.

I assumed the YYWW format because I've seen it used many many times, especially on older parts. But mostly because it would be one HECK of a coincidence that I placed the order precisely on the 18th week of 2023 

I just remembered that in 2018 I ordered a few Solid State 2N2920A dual NPN BJTs. I dug one out and, sure enough, the date code looks to be in YYWW format (see picture).
I understand that it is not conclusive proof since each product could use a different date code format, but it seems quite strong evidence to me...

[TimFox]

I believe that SolidStateInc of New Jersey is really a distributor, rather than manufacturer.
That 2N6422 datasheet appears to be from Motorola (not sure).
Different original manufacturers used different code formats.

[fmzambon]

That 2N6422 datasheet appears to be from Motorola (not sure).

Indeed it looks similar to this Motorola datasheet, though shortened and with less data.
Though, to be fair, several 2N6422 datasheets seem to be copy-pasted from each other.

[srb1954]

That 2N6422 datasheet appears to be from Motorola (not sure).

Indeed it looks similar to this Motorola datasheet, though shortened and with less data.
Though, to be fair, several 2N6422 datasheets seem to be copy-pasted from each other.

It certainly looks like a Motorola datasheet. Maybe Motorola sold off the production rights to someone and the datasheet went with the sale.

Or they just copied the Motorola format. I have always regarded the Motorola discrete device datasheets as being the best in the business. They are easy to use and generally contain more far better laid out and more detailed information than many other manufacturers. Having the more detailed information gives me more confidence that the manufacturer has properly characterised and tested the device and that I can rely in it performing to specs.

I thoroughly dislike having to use devices where the datasheet is so abbreviated that it is only 2 pages long     And one of those pages is just a drawing of a standard transistor package!

[tooki]

One other thing I just thought of: weeks are counted differently depending on which week numbering system is in use.
While I would expect that most use ISO, that’s no guarantee. :/

The end result is that for a given day, the week number can vary ±1.

[tooki]

That 2N6422 datasheet appears to be from Motorola (not sure).

Indeed it looks similar to this Motorola datasheet, though shortened and with less data.
Though, to be fair, several 2N6422 datasheets seem to be copy-pasted from each other.

It certainly looks like a Motorola datasheet. Maybe Motorola sold off the production rights to someone and the datasheet went with the sale.

I guess that depends on whether the design was sold off before or after Motorola’s discrete semiconductor division was spun off to become OnSemi.

[TimFox]

One other thing I just thought of: weeks are counted differently depending on which week numbering system is in use.
While I would expect that most use ISO, that’s no guarantee. :/

The end result is that for a given day, the week number can vary ±1.

Sufficient resolution for any normal purpose.

[fmzambon]

That 2N6422 datasheet appears to be from Motorola (not sure).

Indeed it looks similar to this Motorola datasheet, though shortened and with less data.
Though, to be fair, several 2N6422 datasheets seem to be copy-pasted from each other.

It certainly looks like a Motorola datasheet. Maybe Motorola sold off the production rights to someone and the datasheet went with the sale.

I guess that depends on whether the design was sold off before or after Motorola’s discrete semiconductor division was spun off to become OnSemi.

Previously I linked the 1987 Motorola data book, and the 2N6422 was there. I found the 1995 Motorola Bipolar Power Transistors Data book, and it doesn't mention the 2N6422. There is only mention of the related 2N6421 in the "deletions" section, so I presume that the 2N6422 was dropped some time between 1987 and 1995. If so, it makes sense that Motorola sold off the design to try to make some extra money off of it.

According to Wikipedia, ON Semiconductors was created in 1999. So yeah, the 2N6422 was dropped by Motorola long before ON semi was created, and so it makes sense that the only datasheet to copy from was the old Motorola one.

Back to the original date code question, I understand that the date code may not be in YYWW format or that the week number may be off by one. But I really can't get over the coincidence of ordering a part marked 2318 precisely on the 18th week of 2023.

Andrea