3 ms·
Reverse-engineering the TL431
- joshu 12y agoI keep wanting to write him and ask if there is a way to fund his future writing (funds to buy things to disassemble?) Everything he turns out is fantastic.
- kens 12y agoThanks for the kind words joshu. I'm reluctant to take money because then I feel obligated to write an article reasonably promptly about something I'm not as interested in, and it replaces the fun with pressure. (I hope that makes sense.) But let me know if there's something specific you're interested in. And if you want to donate, my daughter is raising money for wells in Africa: http://righto.com/change http://righto.com/change
- joshu 12y agoI understand. Do you have any projects you want to do but need parts/hardware/etc? I liked the systematic analysis of power supply article, but doing that sort of thing seems pricey. I brought your daughter's wells fundraiser to an even 100%.
- kens 12y agoThank you very much. That's very generous of you.
- userbinator 12y agoGreat explanation. I still continue to be amazed at how tiny the circuitry in ICs is -- the TL431's die is only 1mm^2, and is on an older larger-size process, but already contains over a dozen individual parts. The same area in a modern nm-scale process would fit a few orders of magnitude more transistors. You can find plenty of examples of CMOS digital ICs reverse-engineered but analog/bipolar stuff tends to be rarer; here's a bipolar digital IC (RTL logic), in which one of the transistor's connections has suffered catastrophic failure: http://www.eevblog.com/forum/reviews/small-teardown-an-old-tek-ic/ http://www.eevblog.com/forum/reviews/small-teardown-an-old-t... I wonder if most/all of TL431 produced have the same layout, and not just ones from TI. There's a noticeable absence of any copyright date/manufacturer logo on the die.
- kens 12y agoI'm very curious about the same thing, if all the TL431 have the same layout. The two different TL431s in my article have the same layout, but don't match the schematics in the datasheets. I don't know if they changed the design in the past 26 years, or if there are different designs out there. There's a small "TI" on the die, sideways above and to the right of the TLR431A text, which I assume is for Texas Instruments, but the lack of copyright is interesting. (I think chips are supposed to have a (M) for the mask.) Since I ordered the parts on eBay from China, I don't know who really manufactured them.
- sigterm 12y agoDoes anyone know why the P and N regions conveniently show up as different colors on the photo? Is this due to different oxide thickness?
- etimberg 12y agoInteresting question! I don't know the actual reason, but if I had to wager a guess I'd say that it's due to the different types of doping of the silicon. Some searching did indicate that is what gives some types of gemstones their colour (rubies for instance receive their red colour from chromium impurities distributed throughout the lattice structure http://en.wikipedia.org/wiki/Dopant#Examples http://en.wikipedia.org/wiki/Dopant#Examples). It seems plausible, but there is probably a more accurate answer out there.
- marcosdumay 12y agoOne makes the entire dye with a kind of dopant, creates a oxide layer above it, and remove the oxide where it should be doped again. Thus, I'd guess the difference in color is due to the different thickness of the oxide layer.
- analog31 12y agoThickness makes sense due to the same phenomenon as Newton's rings.