The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> Bandgap design: PNP vs diode vs diode-connected transistor
https://designers-guide.org/forum/YaBB.pl?num=1276142600

Message started by newic on Jun 9^th, 2010, 9:03pm

Title: Bandgap design: PNP vs diode vs diode-connected transistor
Post by newic on Jun 9^th, 2010, 9:03pm

I would like to know the differences, pros and cons of using
PNP, diode, and diode-connected transistor in the bandgap reference design in CMOS technology.

Most of the designs in papers are using PNP in CMOS technology. However, it is not properly characterized and low beta. Furthermore, it is not in design kits. In the design kits, it has diode. It is possible that I use diode instead of PNP. Since they behavior the same (pn junction).

could I use diode-connected transistor since it is well modeled in library.

Title: Re: Bandgap design: PNP vs diode vs diode-connected transistor
Post by vivkr on Jun 11^th, 2010, 4:48am

PNPs have low beta. So you need to be careful and use a topology where this mismatch doesn't matter, e.g. use them as diodes. So technically, there is no difference between using PNPs and diodes. Of course, if you had fully floating vertical NPNs, that would be a different thing altogether.

I have to disagree with you about the modelling of PNPs. Usually, there are dedicated (poor performance yes, but dedicated) lateral PNPs expressly meant for making things like bandgap references. Standard PN diodes on the other hand are poorly modelled, and primarily for the reverse mode. So using them in bandgaps where the forward mode matters could be troublesome.

Vivek

Title: Re: Bandgap design: PNP vs diode vs diode-connected transistor
Post by newic on Jun 11^th, 2010, 5:04am

In the library, it has only diode model. I could make it as PNP by putting a guard ring. Just worry about the accuracy of doing that.

Are you sure that the diode is not allowed to use in forward mode due to poor modeling. I am not aware of this. Need to check..

thx for reminding me

Title: Re: Bandgap design: PNP vs diode vs diode-connected transistor
Post by newic on Jun 11^th, 2010, 5:06am

What is the alternative since the design kits do not have PNP?

Built one??

how about diode connected transistors..

Title: Re: Bandgap design: PNP vs diode vs diode-connected transistor
Post by vivkr on Jun 14^th, 2010, 1:28am

well! if your design kit has no PNP, then you will have to go with the diode. There's no other choice of course. But you need to be very careful when using the diode. You need to ensure that you will no be triggering any other parasitic diodes in the substrate etc. else you will have a nice latchup.

I am guessing that you will be using a Pdiffusion-Nwell diode, with the Nwell tied to ground. Make sure you place a large number of contacts all around the periphery in the Nwell, and also substrate contacts all around the Nwell to firmly anchor everything to ground. Plus avoid placing other transistors nearby in the layout, especially PMOS.

Most importantly, talk to someone experienced about the whole thing and get the layout reviewed well.

Vivek

Title: Re: Bandgap design: PNP vs diode vs diode-connected transistor
Post by newic on Jun 15^th, 2010, 10:19pm

Thx for your inputs.

I am aware of latchup issue.

Title: Re: Bandgap design: PNP vs diode vs diode-connected transistor
Post by RobG on Jun 16^th, 2010, 1:45pm

There usually isn't a true PN diode available, but if there is a PN diode will be a lot different than a pnp. The diode is "like" a pnp with a beta of 1, therefore you get the full current going across the base resistance. The IR drop across this resistance is a big limiting factor in BG design -- it is hard to control.

That said, the "diode" is probably some sort of parasitic pnp.