The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl Design >> Analog Design >> why we don't use simple diodes to build BGR https://designers-guide.org/forum/YaBB.pl?num=1294834814 Message started by nus_lin on Jan 12th, 2011, 4:20am |
Title: why we don't use simple diodes to build BGR Post by nus_lin on Jan 12th, 2011, 4:20am dear guys, the bjt used in bandgap reference is actually diodes, so why don't we just use nwell-diodes, or n+p- diodes? |
Title: Re: why we don't use simple diodes to build BGR Post by Alexandar on Jan 12th, 2011, 6:47am BJT model is more accurate so simulation results will look more like the real thing. |
Title: Re: why we don't use simple diodes to build BGR Post by nus_lin on Jan 12th, 2011, 6:07pm that convince me partially. if diode makes no difference to bjt in a reality, the foundry could offer a better model!!! so that can't be the reason. |
Title: Re: why we don't use simple diodes to build BGR Post by nus_lin on Jan 12th, 2011, 6:18pm i figure out a possible explanation. if we look at their voltage drop, you will see that Vbe(bjt)<Vbe(Nwell diode)<Vbe(N+p- diode), under the same current density. if you look at the serial resistance, bjt should be better than both nwell diode and N+p- diode. i think beta helps a lot. R_eq_c=R_b/beta, where R_b is the base resistance, R_eq_C is an equivalent resistor in parallel with the resistance of the collector itself, which should be much smaller than R_b, due to the large area of collector. |
Title: Re: why we don't use simple diodes to build BGR Post by nus_lin on Jan 12th, 2011, 6:19pm here are other explanations: in Wiley Sansens design essentials (slide 165, page 457) : "A bandgap reference voltage uses a bipolar transistor, connected as a diode. Its current-voltage expression is then quite accurately given by the exponential. A real pn-junction may have a coefficient of 1.05–1.1 in front of the kT/q; a bipolar-transistor connected as a diode does not" |
Title: Re: why we don't use simple diodes to build BGR Post by Alexandar on Jan 13th, 2011, 12:00am Uhm the BJT model also includes the substrate current, since it is a three terminal device, while an ordinary diode has 2 terminals. So you can model it more properly. In layout it is basically the same thing. Beta is a number between 1-5 for these kind of BJT. Base resistance is quite unpredictable, and the low beta won't change much. |
Title: Re: why we don't use simple diodes to build BGR Post by nus_lin on Jan 13th, 2011, 12:20am current BGR circuit is evolved from the old technology, where the beta is greater than 20. that makes a huge difference. for technologies below .13node, i think diode may be better than bjt in terms of area saving. |
Title: Re: why we don't use simple diodes to build BGR Post by Alexandar on Jan 13th, 2011, 12:34am CMOS bandgaps exist for quite some time, and they don't rely on a high beta. Concerning the "area saving". The substrate tap is just a small ring around it. By the way, the amplifier part of the BGR also needs to be quite large to get decent matching etc. So any gain from the standpoint of area is very marginal. |
Title: Re: why we don't use simple diodes to build BGR Post by vivkr on Jan 13th, 2011, 1:29am nus_lin wrote on Jan 12th, 2011, 4:20am:
Why should the foundry bother to model one more device (a diode) when it already is modelling a BJT, which is anyway part of the design kit? Since a BJT may be used for other things and not just as a diode, it is also modelled much more extensively as pointed out by the others. Anyway, you never have plain diodes in any process. There are always other parasitic diodes which you need to model anyway. Also, the BJT layout is usually well-defined and fixed. If you decide to use a perfectly good BJT to make a diode, then that's your decision. It is also possible to make excellent bandgaps where the BJT C-B junction is not shorted out. Vivek |
Title: Re: why we don't use simple diodes to build BGR Post by RobG on Jan 13th, 2011, 9:48pm The series resistance is a non-ideal effect that is a limiting factor with a simple diode (whish is really just a pnp in most cases anyway). On the other hand, with a pnp, the IR drop across the resistance is reduced by a factor of β. That said, I just stumbled across a pn diode in an IBM 0.18um process that is designed for bandgaps and is good across several orders of magnitude current -- much better than a substrate pnp. So they are used, and in this case they seem to be a better choice. rg |
Title: Re: why we don't use simple diodes to build BGR Post by nus_lin on Jan 13th, 2011, 11:22pm thanks, alex, and robg, the discussion becomes interesting. i have to confess that i was wrong about the advantage of bjt over diode. actually, a BGR circuit requires a diode that can be forward biased. that leaves us one only choice--p+ diode over nwell, which is nothing but a bjt. actually, i had more to share. the simulated ptat output (based on an ideal resistor) is actually proportional to T-(2E-4)*(T-300)^2, and is nothing to do with the area ratio and the current density (except for a very low current or very high current). i am also curious to find out why |
Title: Re: why we don't use simple diodes to build BGR Post by loose-electron on Jan 22nd, 2011, 12:52pm You can use the diodes, not a problem, but the collector tied to substrate PNP is commonly used if someone has developed a proper model for it. PN junction models are often not properly modeled with spreading resistance and similar. |
Title: Re: why we don't use simple diodes to build BGR Post by Dan Clement on Oct 31st, 2011, 7:50pm Schottky diodes can also be used. |
Title: Re: why we don't use simple diodes to build BGR Post by loose-electron on Nov 1st, 2011, 3:56pm Dan Clement wrote on Oct 31st, 2011, 7:50pm:
Most standard CMOS does not include that. |
The Designer's Guide Community Forum » Powered by YaBB 2.2.2! YaBB © 2000-2008. All Rights Reserved. |