The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Mixed-Signal Design >> A question on charge injection
https://designers-guide.org/forum/YaBB.pl?num=1220776598

Message started by nobody on Sep 7^th, 2008, 1:36am

Title: A question on charge injection
Post by nobody on Sep 7^th, 2008, 1:36am

Hello, all

I want to design a differential output comparator which implements auto zero method to cancel out the offset. The comparator did cancel out the offset. The charge injection is supposed to go away with a fully differential preamp but still there. I was wondering if any method can remove that injection. The comparator is for SAR adc. Thanks

Title: Re: A question on charge injection
Post by sheldon on Sep 7^th, 2008, 5:29am

Nobody,

Have you considerd adding an additional switched capacitor
stage? If the clock of the second stage is delayed relative to
the first stage, then the first stage charge injection looks like
offset voltage to the second stage and is canceled (or at least
reduced).

Best Regards,

Sheldon

Title: Re: A question on charge injection
Post by nobody on Sep 7^th, 2008, 6:10am

Hello, Sheldon

Thanks for your help. Like you said, adding another SC stage is helpful to reduce the charge injection which goes out to the output terminal of the 1st SC. I should tell my problem specifically. I care about the effect of charge injection at the input or so-called residual offset. When simulating a preamp with an ideal clock(Gswitch in HSPICE)for auto zeroing, there is no charge injection. However, replace an ideal clock with real mos switches, an error is found out.
Regards,

nobody

Title: Re: A question on charge injection
Post by sheldon on Sep 7^th, 2008, 7:29am

Nobody,

How about adding another pre-amplifier stage to increase the
gain between the charge injection and the input? It should not
be an issue unless the offset of the first stage saturates the second
stage. One other question, for most designs the latch is differential
and by default the charge injection is canceled [first order]. How
balanced is the latched stage in your design? Are you using any
other technique to help reduce the charge injection, for example,
using a grounded back-plate sampler?

Best Regards,

Sheldon

Title: Re: A question on charge injection
Post by thechopper on Sep 7^th, 2008, 5:51pm

Hi nobody,

Sheldon is right. If you care about residual offset effect, then you are mainly interested on the charge injection when both inputs are balanced. Therefore if your circuit is designed also in a balanced way you should not see much effect of the charge injection on your input.
Nevertheless, apart from trying out different ways of sampling, you might want to also try sizing your switches differently to reduce the injection effect. For example, increasing their RON should help reduce the injection since less charge is accumulated in the channel prior to turning the switch off. Obviously you can only increase such RON up to the point that it does not affect the settling time requirements for your holding caps that perform the auto-zeroing operation.

Regards
Tosei

Title: Re: A question on charge injection
Post by Tlaloc on Sep 8^th, 2008, 8:04am

One additional point I would like to make on top of the previous posts is that with a fully differential architecture, the charge injection is seen as a common mode hop to the level that the switches and caps match. If you are looking at the common mode or one of the two physical nets, this is visible, but the differential hop should be the charge injection (and offset) divided by the open loop gain of the comparator.

Title: Re: A question on charge injection
Post by thechopper on Sep 8^th, 2008, 8:40pm

Tlaloc brought a good point. However if there is offset in the amplifier (as will be the case in reality) that means the amplifier will not be perfectly balanced (though the topology might be). Consequently a small portion of the CM charge injection noise might show up as a differential signal.
In addition since the amplifier will be unbalance due to the input offset, the CMRR will be finite and addition diff signal will be observed as a consequence of the CM hops at fs translated into diff signal.

Nobody: have you tried checking the charge injection effect as a function of the input referred offset to be canceled. If the effect is noticeable then you might be dealing with something like I just described. In this case you might want to improve the CMRR features of your amplifier.

Regards
Tosei