The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> High resolution voltage divider
https://designers-guide.org/forum/YaBB.pl?num=1386562322

Message started by aaron_do on Dec 8^th, 2013, 8:11pm

Title: High resolution voltage divider
Post by aaron_do on Dec 8^th, 2013, 8:11pm

Hi all,

I need to design a high resolution (14-bit accuracy for example) voltage divider (divide by 2 only). The input is a DC signal, and it must be able to drive a switched-capacitor. Any ideas how to go about this?

thanks,
Aaron

Title: Re: High resolution voltage divider
Post by nrk1 on Dec 9^th, 2013, 6:56am

Since you have a switched capacitor, it may be possible to charge only half of it directly from dc input to get the same effect as charging the full cap. with half the voltage.

Title: Re: High resolution voltage divider
Post by Tiger_ITRI on Dec 9^th, 2013, 10:05pm

Hi,aaron_do:
You may search about "ADC driver" for this topic. Fully differential, small input/output signal, wait until sc is settled and sampled, small resistance for low noise or just apply sc circuit, negative feedback ...are the ideas to help to improve linearity. You need to drive a sc circuit , so you can estimate the load's equivalent resistance you drived, if power consumption is not a design issue, use class-A as output stage is fine or class-AB, class-AB have high gain, high driving capibility, low quiescent current but more noise compared to class-A. The SNR for 14-bit design is need very careful, layout mismatch is an issue in this task. By the way, to measurement such high linearity circuit, good pcb layout and sheilding techniques also a pre-consider matter, althogh there are many instrument can measure such SNR, but the good measurement enviroment is still remain to the designer.
Too talkive~ anyway search about "linearity" "ADC driver datasheet", you will find out more.

Title: Re: High resolution voltage divider
Post by aaron_do on Dec 9^th, 2013, 10:36pm

Thanks both of you for the suggestions.

Aaron

Title: Re: High resolution voltage divider
Post by loose-electron on Dec 31^st, 2013, 10:53am

Instead of resistive division why not just use a charge redistribution circuit?

Charge 1C and split it into 2C, or series charging of capacitors and then flip the switch to divide the voltage - there's a couple of different topologies, where you are dependent on charge injection of switches and capacitor matching for accuracy, but it should be do-able.

Title: Re: High resolution voltage divider
Post by aaron_do on Jan 5^th, 2014, 9:08pm

Quote:

Instead of resistive division why not just use a charge redistribution circuit?

I was thinking maybe a capacitor divider C1/(C1+C2) and then switch the positions of the capacitors to get C2/(C1+C2), and somehow average out. My main concern is the accuracy. I'm not certain I can get 12b accurate capacitor matching. Also, this divider is supposed to be part of an ADC, so ideally I want it's calibration to run in the background. Still kinda piecing it together....

thanks,
Aaron

Title: Re: High resolution voltage divider
Post by loose-electron on Jan 13^th, 2014, 6:29pm

The device can be calibrated as part of the start routine or background calibrate.

Dividers can be done that switch or tweak the capacitance.
Resistive systems has that capability as well.

Get clever and find a way to make the divide by two as part of S&H so its double use saves space.

Title: Re: High resolution voltage divider
Post by RobG on Jan 13^th, 2014, 7:17pm

Aaron - you ought to be able to this with two matched caps and two steps:
1) charge one cap to Vdd and the other to 0V
2) connect both caps together.

Assuming the caps are matched, the voltage across the capacitors will be Vdd/2. It isn't going to be hard to get caps to match to 12 bits in a modern process.

Another way that has a larger coolness factor is to alternately connect two caps in series with the supply, then in parallel with eachother. In the steady state there is no charge transfer, therefore the capacitors must be charged to the same voltage in each configuration. That is, the voltage on each will be Vdd/2. This is true regardless of the cap matching. Pretty cool, but the parasitics of the flying cap must be the same on top and bottom or you will get an error. (You accomplish this by making the flying cap out of two caps with their terminals swapped.)

I'd just use the first method. If you really can't match the caps well enough maybe try a little of both with 3 or more steps (caveat - I haven't actually tried this one):
1) discharge both caps;
2) connected them in series with the supply (each will be approximately Vdd/2);
3) connect them in parallel (even closer to Vdd/2).

Each time you repeat 2&3 the voltage will get closer to Vdd/2, again keeping in mind the parasitic on both plates of the flying cap must match.

Title: Re: High resolution voltage divider
Post by aaron_do on Jan 14^th, 2014, 5:41am

Thanks for the suggestions.

Quote:

Another way that has a larger coolness factor is to alternately connect two caps in series with the supply, then in parallel with eachother

I think what you described is similar to what I had in mind except you explained it a bit more clearly.

Quote:

I was thinking maybe a capacitor divider C1/(C1+C2) and then switch the positions of the capacitors to get C2/(C1+C2), and somehow average out

Quote:

Get clever and find a way to make the divide by two as part of S&H so its double use saves space.

yeah that's definitely the plan. Its complicated though...

thanks,
Aaron