Opamps - Die pictures

[Andy B]

In regards to https://www.eevblog.com/forum/projects/opamps-die-pictures/msg5749525/#msg5749525  I noticed that the characteristics I gave for the ST LM358 were restricted to Ta = 25 degrees where the characteristics for the National Semiconductor LM358 were for the full specified temperature range for this IC. Turns out the characteristics for both these ICs were exactly the same for TA = 25 degrees so I edited these out in my original message so newcomers don't get incorrect info.

Cheers
Andy

[Andy B]

In regards to https://www.eevblog.com/forum/projects/opamps-die-pictures/msg5749917/#msg5749917 concerning the National Semiconductor LM358 I'm wondering why the NPN transistor Q16 is relatively big and has a strange emitter outline (like a "s" turned on its side)?

It's on the bottom left corner here.



Here's a better up close picture of Q16 taken from the right side of the opamp so it's connections approach from the other side.



It only sinks a constant 50 uA but I note that this transistor is between the output (out) and the negative supply rail (Vee) so can have almost the maximum supply voltage (minus about 1.5V) across it.

Can anybody enlighten me about this?

Thanks
Andy

[Noopy]

I´m sure there is an explanation for this geometry. Unfotunately I can´t tell what they thought while drawing this transistor. 

[AnalogTodd]

I´m sure there is an explanation for this geometry. Unfotunately I can´t tell what they thought while drawing this transistor. 

May just have been that they needed a bigger device and there really wasn't real estate available on the chip. Looks to be limited to how big you could get it going further up, so just shove in what and where you can. Wouldn't be the first time I've seen a chip designer do something like that in order to squeeze a device in.

[Noopy]

I agree with you but why the big area? There is not a lot of current...

[David Hess]

It is part of a current mirror so the 50 microamp output current is scaled by the transistor's emitter area, and that sets its size.

[Noopy]

Oh, that makes sense! 

[Andy B]

It is part of a current mirror so the 50 microamp output current is scaled by the transistor's emitter area, and that sets its size.

Thanks!

That explains it for me too. Just learnt something new I wasn't aware of!!

Cheers
Andy

[Andy B]

Here's an additional annotated LM358 die shot of a Texas Instruments LM358. It's actually the one I started working on (and got stuck) before I found Noopy's one.



The original version is below.



Credit: Mikhail Svarichevsky (Zeptobars)
Article title: Ti LM358 - dual general-purpose opamp : weekend die-shot
Article link: https://zeptobars.com/en/read/TI-LM358-dual-general-purpose-opamp

HD version 4833 x 3579px link:  https://s.zeptobars.com/Ti-LM358-HD.jpg

Another die shot I got of the same die but from a different IC (but with same markings) is below. It's easy to trace lines on this one. It's provenance is unknown other than the Youtuber mentioned it's a LM358.



Source: Youtube video link below.
Source Title: Various Opamp Die Designs Used in Fake Parts

Note I didn't really want to include the whole YouTube in this post but the forum software does this automatically for YouTube links.



From the markings "LM358H 08" it appears that this is maybe a National Semiconductor die.

LM358H is National Semiconductors metal can version of the LM358.

The "08" part is National Semiconductors die date marking where "0"=2000 and "8" is the weeks of year coding where 8 = week 48 to 51.

https://web.archive.org/web/20241008014507/http://www.chipdocs.com/manufacturers/NSC.html

Notably Texas Instruments (TI) didn't manufacture the metal can version from what I can see before 2000 and only introduced this with the introduction of the National Semiconductor (NS) equivalents in 2000 which mostly had the "N" suffix to distinguish them from TI versions. Notably the "N" suffix was left off the metal cans probably because of this. Judging from this die it seems the TI LMx58-N series were equivalent to the National Semiconductor LMx58 series because they used NS dies. Note TI took over NS in 2011 but still keeps the two lines.

https://www.alldatasheet.com/datasheet-pdf/view/545599/TI/LM158H.html

The only thing in my annotated die shot I'm not 100% about is the J1 Junction Field Effect Transistor in the schematic for the common bias. I think it's above the "LM358H" marking on the die as the connections are right. However there is no obvious visual junction between the two pads.

Looking at the HD die shot close up reveals a darken area just below the horizontal centre of the likely component that may be a buried junction but its odd it isn't centred.



It maybe something like the following picture with the region types reversed and with the gate being the the now P region above the substrate connected to Vee. Note National Semiconductor (according to the article by Ken Shirriff from which I sourced this picture) invented the ion implantation technique for this in 1974 but it seems a bit complex for this IC.



Credit: Ken Shirriff
Original Source: https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEiShlZSbpm6az5V4UQ8uNeEXbElJ4n3nZni6bFd_Zl7m6e_ma36thMgJ47SiU-tQVsrxKe2anmRBq_XmKqdrELRB6YtvCKD_QjtgudYiWD1OCfXE7pwQ8VpXYbqN3avpCQ2Tfn27kkcqKNG/w9999/transistor-structure-jfet.png
Source Article title: Silicon die analysis: inside an op amp with interesting "butterfly" transistors
Source Article link:https://www.righto.com/2018/06/silicon-die-analysis-op-amp-with.html
Source Article author: Ken Shirriff

What do you think? Did I get it right or is there more to this?

Cheers
Andy

[Noopy]

Source: Youtube video link below.
Source Title: Various Opamp Die Designs Used in Fake Parts

Note I didn't really wont to include the whole YouTube in this post but the forum software does this automatically for YouTube links.

Well it was interesting for me. He used two pictures from my website but didn´t mention the source. That´s no bigger problem for me. But if someone is trying to earn money it would at least be nice to mention the source somewhere.

https://www.richis-lab.de/DAC07.htm#OP-07
https://www.richis-lab.de/Opamp11.htm

The only thing in my annotated die shot I'm not 100% about is the J1 Junction Field Effect Transistor in the schematic for the common bias. I think it's above the "LM358H" marking on the die as the connections are right. However their is no obvious visual junction between the two pads.

Looking at the HD die shot close up reveals a darken area just below the horizontal centre of the likely component that may be a buried junction but its odd it isn't centred.

[...]

It maybe something like the following picture with the region types reversed and with the gate being the the now P region above the substrate connected to Vee. Note National Semiconductor according to the article by Ken Shirriff from which I sourced this picture invented the ion implantation technique for this in 1974 but it seems a bit complex for this IC.

[...]

What do you think? Did I get it right or is there more to this?

You are on the right track. You can use the collector n doped well and put the p doped base as a gate on top of it. That is a common way to build a J-FET. Here is an explanation:

https://www.eesemi.com/jfet.htm

It has a high tolerance but it was a common way to build such circuits.

Another example:



https://www.richis-lab.de/REF16.htm

[BarsMonster]

Credit: Mikhail Svarichevsky (Zeptobars)
Article title: Ti LM358 - dual general-purpose opamp : weekend die-shot
Article link: https://zeptobars.com/en/read/TI-LM358-dual-general-purpose-opamp

Yeah, looking back I would say it's was a non-genuine part, as Ti chips typically have logo somewhere...

[magic]

There is TI logo, right under the LM358H text. It also looks like typical TI jellybean analog.

Markings on dice have little relation with package markings. I doubt there were any special dice for H packages. In fact, I doubt that anyone ever made H package LM358 at all. H could be a revision or some other internal code, 08 is perhaps year 2008.

The other identical die is of course the exact same TI die, but shot in different light.

J1 is of course an EPI FET as explained by Noopy, exactly like on NS.

[magic]

Source: Youtube video link below.
Source Title: Various Opamp Die Designs Used in Fake Parts

Note I didn't really wont to include the whole YouTube in this post but the forum software does this automatically for YouTube links.

Well it was interesting for me. He used two pictures from my website but didn´t mention the source. That´s no bigger problem for me. But if someone is trying to earn money it would at least be nice to mention the source somewhere.

Well, at least he mentions this thread. He also used a bunch of mine, maybe we should do a DMCA takedown 

Most dice in this video including C665 are LM358 variants, except for those few RC4558 which he suspects to be LM358.

It's interesting that there are supposedly fake LM358 which are faster than the original. I guess if your process is fast enough to support LM358 topology running at a few MHz, you can reduce the compensation cap and save some money. Clever

[Noopy]

Source: Youtube video link below.
Source Title: Various Opamp Die Designs Used in Fake Parts

Note I didn't really wont to include the whole YouTube in this post but the forum software does this automatically for YouTube links.

Well it was interesting for me. He used two pictures from my website but didn´t mention the source. That´s no bigger problem for me. But if someone is trying to earn money it would at least be nice to mention the source somewhere.

Well, at least he mentions this thread.
[...]

"a pretty long discussion thread on eevblog"
Well, a very vague mentioning...
Your name is at least mentioned. 

[Andy B]

The other identical die is of course the exact same TI die, but shot in different light.

I did look at the pin pads before posting and noted the pad lead connection markings looked different so this is what led me to believe they were different dies.  BarsMonster would know for sure.

J1 is of course an EPI FET as explained by Noopy, exactly like on NS.

Was wondering what the horizontal bars were doing which you outlined. From your outlined drawing I can now see the epi-FET. One thing I now notice is that bar extends between the bars under the metal trace between the pads (there's no dip/hill in the metal between the pads) which isolates the pads from each other which creates the channel as you drew.

Big cred for recognising the whole lot is the FET. Will update the diagram to suit.

Cheers
Andy

[Andy B]

The updated annotated TI LM358 with the complete epi-FET is below.



Cheers
Andy

[Andy B]

There is TI logo, right under the LM358H text. It also looks like typical TI jellybean analog.

It is indeed what looks like a very tiny Ti logo (they do use Ti instead of TI on their dies).

Note as an interesting aside  the "M" mark in a round circle is a copyright mark under US law. However it requires the manufacturers name or abbreviation to be included on the die for this to be effective. Therefore without the "Ti" this die wouldn't be copyright protected.


909. Mask work notice

(a) The owner of a mask work provided protection under this chapter may affix notice to the mask work, and to masks and semiconductor chip products embodying the mask work, in such manner and location as to give reasonable notice of such protection. The Register of Copyrights shall prescribe by regulation, as examples, specific methods of affixation and positions of notice for purposes of this section, but these specifications shall not be considered exhaustive. The affixation of such notice is not a condition of protection under this chapter, but shall constitute prima facie evidence of notice of protection.

(b) The notice referred to in subsection (a) shall consist of—

(1) the words “mask work”, the symbol *M*, or the symbol Ⓜ (the letter M in a circle); and

(2) the name of the owner or owners of the mask work or an abbreviation by which the name is recognized or is generally known.

https://www.copyright.gov/title17/92chap9.html

Cheers
Andy

[Andy B]

Source: Youtube video link below.
Source Title: Various Opamp Die Designs Used in Fake Parts

Note I didn't really wont to include the whole YouTube in this post but the forum software does this automatically for YouTube links.

Well it was interesting for me. He used two pictures from my website but didn´t mention the source. That´s no bigger problem for me. But if someone is trying to earn money it would at least be nice to mention the source somewhere.

Well, at least he mentions this thread. He also used a bunch of mine, maybe we should do a DMCA takedown 

Most dice in this video including C665 are LM358 variants, except for those few RC4558 which he suspects to be LM358.

It's interesting that there are supposedly fake LM358 which are faster than the original. I guess if your process is fast enough to support LM358 topology running at a few MHz, you can reduce the compensation cap and save some money. Clever

I don't think with 9k viewers he is making much if anything from YouTube on this video. However he does include a link for Patreon for donations.

Frankly it is rude not to acknowledge contributors whether its creative commons or not.

Giving contributors acknowledgement by both mentioning them and  including links encourages the kind of stuff we do and hence everyone benefits.

Plus the fact it makes you more authoritative by doing so (your comments with acknowledgements have more substance, otherwise they are just pie in the sky).

I'm careful to acknowledge sources of info in my posts (including emails to them to make them aware of such use) and I often find it gets me even more informative comment (this forum thread is the perfect example).

I did find his video interesting and informative but it would of been great if he included links so I could of gotten more info about the LM358 die shot he used.

Whilst a DMCA is a solution  I think a message to him to clean up his act and acknowledge the contributions both by crediting the contributors and providing a link to his sources (pointing out the above) would be a better solution.

And yes I would like him to link the source of that LM358 die pic.

Cheers
Andy

[Noopy]

I totaly agree with you. I´m sure magic was joking with the DMCA.

I´m happy when my pictures are helping people, entertaining people, when they are just decorative or encouraging people to learn something about electronics / semiconductors, ...
But there should be a small acknowledgement. Especially if you are trying to earn money with them.

I have messaged him and just stated that it´s not nice to not mention the source. The video is online for a year, he doesn´t has to change it. If it´s possible a link in the description would be nice. If that´s not possible it´s ok for me too.

It´s also a friendly hint. If you are trying to earn money with pictures which are not your own and which aren´t free to use you can get really big problems. I´m not a troublemaker but there are people which are looking for a fight all day.

[magic]

He explains the origin of those TI and C665 images in the previous video #1749 - those are his private geniune/fake chips decapped for him by a viewer.

You also get a glance how he works - a bunch of "screenshot xyz.jpg" images from wherever, probably he doesn't even remember the sources himself, which he just shows and talks about.

[magic]

I always added some clever-ass quote to those posts. So lets go:

You will live to see man-made horrors beyond your comprehension.

Nikolai Tesla about LM358

Here's another one.

Quest: find the epi-FET

[Andy B]

Here's another one.

Quest: find the epi-FET

Good die shot. Can see mostly everything but the epi-FET between Vcc and Q104 eludes me. Can't see any epi-FET connection to Q104 collector/base/tank and it's connection to Q105 base. Tried digging for clues but couldn't find anything.

Cheers
Andy

[magic]

Hint: draw cross sections along these lines

[Andy B]

He explains the origin of those TI and C665 images in the previous video #1749 - those are his private geniune/fake chips decapped for him by a viewer.

You also get a glance how he works - a bunch of "screenshot xyz.jpg" images from wherever, probably he doesn't even remember the sources himself, which he just shows and talks about.

Here is a shot from the previous video of the LM358 IC package his Danish friend extracted the die from.



Source: Youtube video link below.
Source Title: #1749 Fake LM358 De-cap

Note again I didn't really want to include the whole YouTube in this post but the forum software does this automatically for YouTube links.



Markings as follows.

IC Markings: Ti 29ACN6M LM358AP
Die markings: LM358H 08

LM358AP Dual, 30-V, 700-kHz, 3-mV offset voltage operational amplifier
   Package: PDIP

29ACN6M =>

   last digit of year = 2 (1992,2002,2012,2022)
        month = 9 = September
   Independent Assembly Number (LLLL) = ACN6
   Assembly Site Code (S) = M = TI Mexico (manufacturing semiconductors from 1987)

https://www.ti.com/general/docs/product/topsidemarking.tsp?part=LM358AP&pkg=P&pin=8&actionPerformed=viewTopsideMarking
https://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/1013/4527.How-to-identify-STC-_2600_-LTC.PDF

This shot means that my conjecture that it was from a metal can is wrong and the "H" on the die code "LM158H" doesn't relate to this.

As to including the LM158 die in a LM358 package the LM158 chip is rated for a broader temperature range so there is no problem with that and it is common practice amongst IC manufacturers.

Cheers
Andy

[Noopy]

Hint: draw cross sections along these lines

Hm... Are you sure? 

As to including the LM158 die in a LM358 package the LM158 chip is rated for a broader temperature range so there is no problem with that and it is common practice amongst IC manufacturers.

Yes, this is quite usual. 



Now the IMSAI Guy added two links to my site in the description:

Episode 1750
there are lots of different designs used in fake chinese opamps,  just not the one marked on the top
some photos are from here:
https://www.richis-lab.de/DAC07.htm#O...
https://www.richis-lab.de/Opamp11.htm
Be a Patron:   / imsaiguy