REF01 - different generations - die pictures

[Noopy]

Hi all!

I hope it´s ok to start a new topic for the REF01. I found no existing topic.

Up to now I have decapped four REF01-generations and would like to share them with you.


The first generation die was packaged 1976:




The second generation was developed 1977:





The third generation is missing. Probably it´s only one year in which it could have been produced. "My" generation 2 package dates back to 1986 and generation 4 was designed 1987. But there must be a generation 3 because the designs have names with a revision coding: 1800Z, 1800Y, 1800W => The X is missing. Perhaps generation 3 was buggy.


The fourth generation was developed 1987:



There is a die picture in a datasheet that probably shows the generation X. With an overlay you can spot the small changes:




The fifth generation is actually a ADR01 as the REF01 moved to Analog:



Sorry, bad picture...


From 1 to 2 PMI did only small geometry changes at the output and at one double transistor.
From 2 to 3 there were added more resistors and fusable links to adjust the Output voltage and the temperature drift. The startup circuit was changed. The die got bigger. Two spare transistors were added, perhaps for the startup circuit. (Actually I have no generation 3 but with the picture out of the datasheet it seems it got all the changes you can see in generation 4.)
From 3 to 4 the spare transistors were covered by the output metal layer.
From 4 to 5 Analog wanted to earn more money and canceled the REF01 by putting ADR01 in the packages.


Overview over all:
https://www.richis-lab.de/REF00.htm


Greetings,

Richard

[magic]

I wanted to see what's inside ISL21009 (discontinued years ago) so I ordered a few from AliExpress, found this inside. Looks familiar, this must be the 1987 generation REF02. I turned it on before taking it apart (sorry Dave) and the output was 5V.

The only differences from REF01 that I see are output scaling resistors (bottom right corner, above GND pad) and more filled area in capacitors. Apparently REF01 is decompensated to achieve the same AC performance at higher closed loop gain. We can make an educated guess what the 2.5V version looks like.

If somebody is interested, full schematic can be found on Noopy's site, it looks alright. The one in current AD datasheet is bogus, maybe it refers to gen 5?
https://www.richis-lab.de/REF01b.htm

One odd area which catches the eye is Q5 and Q10. Apparently Q5 base diffusion is surrounded by N+ diffusion to catch holes injected into Q5 collector by the emitter of Q10 and keep them from "leaking" into the base. There is also a P diffusion around the whole perimeter to collect some more holes and route them to the substrate before they recombine inside Q10 base aka Q5 collector, thus increasing Q10 beta. Or at least that's my guess.

edit
But wait, the official schematic shows Q10 collector connected to Q5 collector rather than to ground, so I'm not sure what's happening here.

[Andreas]

Hello,
I would also have measured
- Current consumption (ISL21009 is low power)
- Voltage on Pin 3 (usually chip temperature on bandgap devices)
- Voltage on Pin 8 (usually noise reduction so something around the zener voltage on buried zeners)

with best regards

Andreas