Interesting BiCMOS circuits in the Pentium, reverse-engineered
15 comments
·January 21, 2025rossdavidh
FuriouslyAdrift
I remember the Exponential Technologies bipolar PowerPC chip running at 533 MHz back in 1997... so sad Apple killed it
https://web.archive.org/web/19970712065424/http://www.byte.c...
kens
That's an interesting chip. You don't see many bipolar processors. (The Pentium has a few bipolar transistors sprinkled around, but it's mostly CMOS.)
formerly_proven
DEC made a one-off ECL CPU in 1993: https://bitsavers.org/pdf/dec/tech_reports/WRL-93-8.pdf
300 MHz / 115 W. Ten years later, you'd have Pentium 4 at 3000 MHz and 115 W.
kens
Is BiCMOS radical woke? :-) Author here for your Premium questions...
chasil
What areas still use bipolar? Does a switching power supply use substantial bipolar? Does anybody still implement TTL or ECL?
Quoting you below...
"The most unusual circuit is the BiCMOS driver. By adding a few extra processing steps to the regular CMOS manufacturing process, bipolar (NPN and PNP) transistors can be created. The Pentium extensively used BiCMOS circuits since they reduced signal delays by up to 35%. Intel also used BiCMOS for the Pentium Pro, Pentium II, Pentium III, and Xeon processors. However, as chip voltages dropped, the benefit from bipolar transistors dropped too and BiCMOS was eventually abandoned."
I didn't realize that BiCMOS lasted so long. I thought it was only used on the original Pentium, but I really didn't look hard.
Edit: BiCMOS has a wiki.
RicoElectrico
I think the best keyword to seek is BCD (Bipolar, CMOS, DMOS) which is a process pioneered by ST. It is quite alive indeed.
Bipolar has for example lower noise than CMOS when it comes to opamps.
qwezxcrty
Lower noise for lower source impedances, to be slightly more precise.
Ah, memories. I was a manufacturing engineer in a bipolar factory from 1989 to 1992, and BiCMOS was the perennial hope for a future for our acquired skills. When word broke that the Pentium would have one, it seemed significant. On the whole, though, it didn't have the impact that was hoped for; bipolar hangs on (like COBOL) in certain niches, but I think despite the theoretical advantages, the disadvantages of having to think about both bipolar and CMOS transistors in the engineering, was too much of a price to pay.
Lesser known reasons: as bipolar transistors went to polysilicon gates (which have a tiny, ~1-2 Angstrom thick layer of oxide in them) and MOS transistors started to become leakier through their ever-shrinking gate oxides, the distinction between MOS and bipolar transistors became fuzzier. Modern MOS transistors leak less current through their gate oxides than the bipolar transistors did through the polysilicon emitter, but the physics of the two is not as different as it was twenty years ago.