how to make reference voltage with use of resistor ladder

Status
Not open for further replies.
M

Monady

Advanced Member level 4
Joined
Dec 1, 2008
Messages
109
Helped
12
Reputation
24
Reaction score
3
Trophy points
1,298
Visit site
Activity points
2,151
reference voltage how to make

Hi dear all friends;
i have used of resistor ladder in order to making reference voltages(similar to ladder in flash ADC). but i have some problem with this ladder. reference nodes are connected to large parasitic capacitors via some switches. Thus when any of SWs, being ON, output of that SW, is not expected value. for example, instead of 350mV, it is 330mV.
i know that if i choose value of each transistor so small(e.g. 1Ω), mentioned problem will be solved, but unfortunately implementing such values are not realizable. i don't know that any substitution method is exist for making this reference voltage.
i would be appreciated for any suggestion.
(Now, my resistor ladder is made of 256 resistors and each resistor is 8Ω, technology is 0.18u)

Thx in advance
 

make less noise reference voltage

how about if adjusting the ladder resistor value?
 

Reference Voltage

Thx for your reply. Excuse me but i can't understand it! adjusting what?
if it was your mean that "we must find an accurate and correct value for resistor"; as i said previously, it seems that decreasing resistor value doesn't work here(due to the restriction of fabrication). correct me if i made a mistake Plz.
 

Re: Reference Voltage

my understanding is you have resistor string for resistor ladder, and from each node on resistor string, you have transistor connected as switch followed by large cap. I am thinking if you can try to increase resistors value in your resistor string.
 

Reference Voltage

dear coffeelox, increasing resistors value, can not solve this problem even intensify it. if we increase resistor value, time constant will be increased consequently, then reaching to steady state condition will need more time. thus i think it's not a solution.
 

Re: Reference Voltage

if the resistor ladder is used for reference voltage, I did not see any timing issue as only DC, no AC on it. and once DC and gnd on both sides of resistor ladder, it's done then.
 

Reference Voltage

you are right, but kick back noise, charge injection and clock feedthrough (of switches) can affect reference voltages thus reference voltage must come back to steady state with appropriate time constant (Behzad Razavi has a paper about this issue)
 

Re: Reference Voltage

usually in adc impementation a voltage regulator coupled to both the begin and to the end of the ladder provides to it the current needed to recovery the intermediated voltage references to the target values in the wanted time (depending on the bits of resolution and the clock period of the whole sistem).
The ladder's resistors value are choosen trading-off between the RC constant time between ladder resistors and parasitic capacitances (those quoted by you) and the current the regulator should provide.

I hope this help you.
CBs.
 

    M

    Monady

    Points: 2
    Helpful Answer Positive Rating
Reference Voltage

dear crystalballs,thx for your useful reply.
i didn't know any thing about voltage regulator coupled to the ladder, i would be appreciated if you give me some references for my better understanding (papers, link of some thesis or ....).

Thx in advance
 

Re: Reference Voltage

at the moment how do the begin and the end of your ladder are connected?
that's, for example: between two analogLib voltage generator?

CBs
 

Reference Voltage

ladder begins with 0.75V and ends with 0.25V. i used of Ideal Voltage source for generating mentioned voltages, although i must change them with real sources in future.
 

Re: Reference Voltage

first of all, if ladder internal nodes don't succeed to reach their steady state voltage (or almost it) supplied by ideal supply voltage source you have no possibilities to reach just mentioned target with any real voltage source (for example a voltage regulator mentioned by me).

Simply, I suppose, your RC constant time is to much for the clock period you are using.

What is the application for this ladder?
Can you use for example over-boosted switches to reduce their dimension?
Does the switches give the greater contribution of the parasitic capacitance?
Have you... I don't know... a comparator, an integrator,... after the switches?

CBs
 

Reference Voltage

yeah i know that RC constant is so much but unfortunately i can not reduce resistor value due to the some restriction of fabrication. this ladder that consists of 256 resistors, is connected to other parts of circuit with too many switches, although only some of these SWs are ON in the same time. parasitic cap is about 0.2pF (of SWs and other circuits(comparator) that followed by SWs). at now size of SWs are minimum (0.22u), thus i can't reduce parasitic cap of SWs more that it. it seems that only solution is working on ladder.
i don't know what is over bossted SW, is it Bootstrapped SW? although i think that, simple NMOS SW has a smallest size in all of SWs.
clock freq. in my work is 1GHz.(Resistor string DAC with 8 bit resolution).
 

Reference Voltage

1GHz is a very high frequency (really have you a comparator that works at such freq? If yes what is the architercture?).
Sorry, I was meaning bootstrapped switch... but: also the resistance of the switch could enter in the calculation of the RC constant time (tau). Have you tried to increase the switch dimension? Maybe the capacitance increases but the resistance could decrease more...

Moreover not always the minimum MOS resistance coincides with the minimum length; you can try to simulate it with a stand alone test.
Input capacitance of a comparator sometimes depends on the maximum offset voltage you can accept... is this the case? Because in this case you can try to adopt some smart techniques like Input/Output offset storage.

let me know...

CBs
 

Reference Voltage

yeah my freq is 1GHz, i used of latched comparator, besides i proposed a new architecture that maybe in future i'll publish it(after confidence about its novelty). i have changed size of SWs and i saw that minimum ratio of SW had a best response. unfortunately offset of different reference voltage nodes are not identical . some of nodes have more, and some have less offset. i think Input capacitance of a comparator is not so large, and my comp has a offset less than 400uV. Maybe only way for getting rid of this dilemma is about working on ladder (or decreasing resolution of ladder). what do you think?
 

Status
Not open for further replies.

Similar threads

Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…