Decapping the LT1236LS8

[branadic]

I was interested in what's inside the LT1236LS8 and what are these extra internal connection for. Here is the answer to the first part. Higher resolution picture is available, but upload is limited to 1MB

[zlymex]

Fantastic. How did you do something like that? Even you have a very good microscope, how did you remove plastic top without damage the chip?
Is the buried zener on the left side?

[krivx]

The internal lands are probably for testing. They all look like they have been probed.

edit: looks like some "ladder" features between lands, so probably used for laser trimming resistors (?)

[TiN]

I can host bigger image, feel free to drop it on my ftp (link in sub).

[branadic]

It's the LT1236 in ceramic LS8 package, so the only thing to do was to slightly mill the lid away. I then took pictures with a microscope (5x enlargement).
Your guess is as good as mine, I would also assume that the buried zener is on the left side in the middle.
There are several damaged bond pads, I guess they come from probing while trimming the reference. There are also some [laser] cutted traces, visible on the high-resolution pictures, that will follow soon.

[branadic]

[Andy Watson]

Awesome work, but,

"Enter password for doc on ftp://ftp.xdevs.com"

[branadic]

Works right now with the help of TiN.

[Andreas]

Hello Branadic

thanks for the pictures.
the output pin (Pin 6) has 2 bond wires: is this a kind of force-sense mechanism (but why not for the Gnd-Pin) or is this a option to switch among the different output voltages for the device?

And why in the hell is the chip marked with 1205 and not with 1236??

With best regards

Andreas

[SeanB]

Probed pads are a trim, using a series of fusible links ( you can see a few have been detonated). clockwise from top you can see 1,  4, 10 have all been blown open to add the diffused resistors into the circuit. The other pads are associated with the bonding pads so they would not have probe damage on them when bonding.

2 bond wires so it can do output sensing without the resistance of the bond wires having an effect, and the other is to reduce bonding resistance. the pads in both cases have a low value aluminium trace connecting them, but the bond wire resistance will be lower than this so it does help. The NC pads seem to be there to add resistors later after encapsulation, probably used to do a final trim for either tempco or output voltage, as the top layers are all associated with the trim pin in some way, adding in resistors so you can do a trim up or down depending on the fuses blown.

Nice photo there, thanks.

[Noopy]

branadic has sent me the LT1236 and I took some more pictures:

https://richis-lab.de/REF05.htm





Some interesting points:




They integrated a option to cut the opamp from the output and give it an own pin. That would make it possible to do voltage sensing outside the package.
To realize that a fusible link connects the two bondpads. One is connected to the push-pull-output and one goes to the opamp (and to a Pull-Down).




There are 13 (!) fusible links to adjust the voltage. With a full 100% adjusting range you can still realize a accuracy 0,012%! 




What´s that directly above the zener? A transistor? Seems to be not connected. Did they plan to connect zener and transistor for a low tc as in the LTFLU?

[iMo]

BTW, there are 3 fuses you may flash by yourself via pins, it seems..
PS: thus you get 3 additional pins with some unknown function..

[daqq]

Thanks for the great pics!

If you are looking for more victims, the LTC6655 also has a ceramic version

[The Soulman]

I could sent you a faulty LT1021dmh-10, pm if you're interested.

[Noopy]

BTW, there are 3 fuses you may flash by yourself via pins, it seems..
PS: thus you get 3 additional pins with some unknown function..

Perhaps even more...

Thanks for the great pics!

Thanks!

If you are looking for more victims, the LTC6655 also has a ceramic version

Epoxy packages are no problem for me:

https://www.richis-lab.de/decap-ofen.htm



 

I could sent you a faulty LT1021dmh-10, pm if you're interested.

Principally I´m interested but as far as I´m informed the LT1021 is exactly a LT1236, isn´t it?

[The Soulman]

I could sent you a faulty LT1021dmh-10, pm if you're interested.

Principally I´m interested but as far as I´m informed the LT1021 is exactly a LT1236, isn´t it?

I don't know, want to find out? 

[Noopy]

I could sent you a faulty LT1021dmh-10, pm if you're interested.

Principally I´m interested but as far as I´m informed the LT1021 is exactly a LT1236, isn´t it?

I don't know, want to find out? 

You are absolutely right, we should look into one! 
...writing PN...

[branadic]

It was Andreas who made this statement:

https://www.eevblog.com/forum/projects/10v-reference-(i-did-it-my-way)-no1/msg606742/#msg606742

https://www.mikrocontroller.net/topic/488107#6111559

and this statement relies on the pictures I made of LT1236LS8 and the pictures of LT1021 presented here:

https://zeptobars.com/en/read/Linear-LT1021-precision-reference-zener

-branadic-

[iMo]

BTW, there are 3 fuses you may flash by yourself via pins, it seems..
PS: thus you get 3 additional pins with some unknown function..

Any idea what function could those pins A,B,C represent when I blow up the 3 fuses (in yellow circles)??

[Noopy]

BTW, there are 3 fuses you may flash by yourself via pins, it seems..
PS: thus you get 3 additional pins with some unknown function..

Any idea what function could those pins A,B,C represent when I blow up the 3 fuses (in yellow circles)??

In my view these pads have all ground potential leading to resistors and then to the opamp. If you cut the fuses you can insert an other potential to the opamp but what happens there I don´t know... 

[magic]

I always assumed that the extra pins are for final, post-packaging trim. Will have a look.

and this statement relies on the pictures I made of LT1236LS8 and the pictures of LT1021 presented here:

Zeptobars already posted OPA627 which he admitted turned out to be fake and OPA2134 which he still didn't admit is fake. Take him with a grain of salt

[branadic]

Zeptobars already posted OPA627 which he admitted turned out to be fake and OPA2134 which he still didn't admit is fake. Take him with a grain of salt

That would mean that someone stripped down LT1236 and repackaged it into TO can to sell it as LT1021... pretty much inconceivable. 

-branadic-

[Andreas]

Look at the datasheets of a LT1236 and LT1021. (they don't lie).
Simplified schematic: identical.
Trimming scheme (with special diode) for the 5V Versions of both references: identical.
The only differences are the initial tolerance and the T.C. of the different grades.
I always thought that the LT1236 is a "enhanced grade" of the LT1021.

And why should one downgrade a LT1236 to a LT1021 to generate a fake device which is sold cheaper.

with best regards

Andreas

[Andreas]

Any idea what function could those pins A,B,C represent when I blow up the 3 fuses (in yellow circles)??

perhaps different output voltages?
The LT1236 is available in 5V and 10V.
The LT1021 is available in 5V, 7V and 10V

with best regards

Andreas

[magic]

That would mean that someone stripped down LT1236 and repackaged it into TO can to sell it as LT1021... pretty much inconceivable. 

Zeptobars didn't say that the chip came in a metal can, was it you or somebody known who sent it in? All bonding and testing pads are missing metalization, which suggests that the die took a bath in acid, and that's how Zeptobars normally dissolves plastic packages.

Andreas' price argument is more convincing You guys are probably right, but I must warn not to trust Zeptobars blindly unless you know where he bought the chips from or who sent them in. Now I also recall that more recently he was "surprised" to find NE5532 inside OP275. I suppose if they used some less familiar die instead, he would have fallen for it.

I had a look at the datasheets. They are indeed identical with only one suspicious difference: the typical output vs temperature curves are totally different