The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Mixed-Signal Design >> Bootstrapped switch body tie
https://designers-guide.org/forum/YaBB.pl?num=1385711196

Message started by aaron_do on Nov 28^th, 2013, 11:46pm

Title: Bootstrapped switch body tie
Post by aaron_do on Nov 28^th, 2013, 11:46pm

Hi all,

I was attending a talk by a distinguished professor a few days ago, and he was talking about bootstrapping for a switch. The question that came up was how do you tie the body terminal of your switch? The professor's answer was the best place to tie it is to ground.

Now I didn't think of it at the time, but now I'm wondering if there's going to be a reliability issue. The gate of a bootstrapped switch could be swinging as high as 2VDD, and if you tie the body to ground, you're going to have 2VDD across the oxide. To me that's a big no no. Does anybody have any comments on this (whether there is a reliability problem, or the best way to tie the switch body)?

thanks,
Aaron

Title: Re: Bootstrapped switch body tie
Post by rfmagic on Nov 30^th, 2013, 11:56am

I would say that the risk for potential breakdown is higher with the bulk connected to ground but there is some relaxation here. The voltage at the gate with the bootstrapping can reach 2VDD but this is peak-to-peak voltage. The rms will be much lower and from my experience it will not degrade reliability too much.

In my personal opinion connecting the bulk to ground is not a good suggestion as you will break the bootstrapping mechanism. The Vt modulation will degrade the linearity performance of the switch. IF you connect the bulk to the source you have all the nodes "floating', maximizing the efficiency of the bootstrapping.

Title: Re: Bootstrapped switch body tie
Post by aaron_do on Dec 1^st, 2013, 5:32pm

Hi rfmagic,

thanks for the reply. For the reliability part, my understanding is that it's not correct to look at the rms voltage. The reason is the time-to-failure of the device reduces exponentially with the gate voltage. i.e. having a duty cycle of 50% simply means your device will last 2x longer than the same device with a DC bias of 2VDD (as opposed to the same time as a DC bias of √2VDD.. That said, I have seen papers where their experimental data shows that if your signal is an RF signal, the time-to-failure is much longer (memory effects maybe?). My switching speed is on the order of a couple of hundred MHz.

I agree with you about the Vt modulation. However, if you connect the body to the source/drain, I believe there is a possibility of the junction diode turning on when the switch is supposed to be OFF. I have found a solution for my particular implementation of the bootstrapped switch, but it isn't pretty. My main question is really how come there is no reliability issue. Maybe its because the professor's work is too academic?

Aaron

Title: Re: Bootstrapped switch body tie
Post by boe on Dec 2^nd, 2013, 2:11am

Hi Aaron,
if Vgate>VDD, the switch is ON, so gate stress (in settled condition) relates to the channel voltage, not the bulk voltage.
- B O E

Title: Re: Bootstrapped switch body tie
Post by aaron_do on Dec 2^nd, 2013, 4:52pm

Hi boe,

sounds logical ;) thanks

Aaron

Title: Re: Bootstrapped switch body tie
Post by RobG on Jan 12^th, 2014, 8:13pm

When boot strapped switching was new there was a lot of talk about reliability, but it must not be a problem given its ubiquitous use nowadays.

Excluding active circuits, the only way I know to tie the bulk to a non-ground place is to place two switches in series and tie the bulk to the center point. But then each switch has to be 2x wider to get the same overall resistance... I'm willing to bet you would get better performance by simply doubling the width of a single switch and leaving the bulk tied to ground.

Title: Re: Bootstrapped switch body tie
Post by aaron_do on Jan 13^th, 2014, 7:05am

Thanks for the reply. I guess you're right about the reliability. Its just that I was thinking of other ways to tie the body, and I was wondering if the way the professor tied the body was the most common way.

Aaron

Title: Re: Bootstrapped switch body tie
Post by loose-electron on Jan 13^th, 2014, 6:33pm

If it is NMOS the bulk is already substrate to ground. Only PMOS will give you the freedom to float the bulk.

But thats a vanilla Nwell Psubstrate CMOS process.

Title: Re: Bootstrapped switch body tie
Post by aaron_do on Jan 13^th, 2014, 9:20pm

I have access to deep nwell. Wouldn't that be common for ADC designs?

Aaron

edit: just realized I didn't mention anything about ADCs in this post...

Title: Re: Bootstrapped switch body tie
Post by loose-electron on Jan 15^th, 2014, 4:19pm

aaron_do wrote on Jan 13^th, 2014, 9:20pm:

I have access to deep nwell. Wouldn't that be common for ADC designs?

Aaron

edit: just realized I didn't mention anything about ADCs in this post...

Depends on the process, its all connected together one way or another.

Even SOI has some coupling.