About LTC1864

Hi,

We will look into this. I will contact the product owner and get back to you.

okay, Thank you so much for looking into this issue.

i have created resistor banks from 10 ohm to 1Giga , i get stable ADC counts only for 10 ohm to 100K resistors above that range fluctuations in ADC count start , from 500k on wards till 1 giga ohm range of resistor i am getting fluctuations in ADC count like the count is varying from 100 to 1000 multiple for every second.I have not separated the analog and DC ground plane so my be that's because dc noise is adding in Analog signal.I need solution for how i can make LTC1864 circuit more stable for high RS(source resistance) values like from 5k to 1G resistances when any power supply is given from battery or from SMPS or from Adapter.

in Datasheet in Analog inputs secion - its mentioned that

"Because of the capacitive redistribution A/D conversion
techniques used, the analog inputs of the LTC1864/LTC1865
have capacitive switching input current spikes. These current
spikes settle quickly and do not cause a problem if
source resistances are less than 200Ω or high speed op
amps are used (e.g., the LT®1211, LT1469, LT1807, LT1810,
LT1630, LT1226 or LT1215). But if large source resistances
are used, or if slow settling op amps drive the inputs, take
care to ensure the transients caused by the current spikes
settle completely before the conversion begins."

I do not understand how to make sure that transient caused by current spikes settle quickly before conversion begins

should i use high value capacitor at analog input i tried 10uf tantalum at In+ analog pin it did not worked and also added 1uf tantalum at vref and vcc it also did not worked when i connected 12v 1A adapter. my input range is 0-5 v full scale.

then i tried giving power supply with SMPS the counts were stable when i connected 10uf tantalum at analog pin in place of C7 in above schematic, and 1uf tantalum in place of c8 and removed R12.

my circuit works for12v 1A  SMPS  but same circuit gives fluctuations when Adapter supply is provided. i need solution for when i give supply by Adapter  adc count varies.

Please help me in this issue i ave been trying a lot to get stable ADC counts.

The high value resistors will generate noise on their own. The resistor noise formula is E = √(4⋅R⋅k⋅T⋅ΔF) where R is the resistance, k is Boltzman's constant, T is temperature in degrees K and ΔF is the BW. You can limit the bandwidth with a larger filter capacitor.

The op amp you are using has a BW of only 10kHz so it cannot respond in a timely manner to the transients at the ADC input. 

The LTC1864 sample time (tsmpl) starts after CONV goes low and the conversion is completed. This is the time that the analog input has to settle. You can increase this time by reducing the sample rate. The RC at the ADC input forms an RC time constant and the glitch generated will need to settle to less than 1 LSB. The number of time constants required to settle will depend on the filter capacitor size. A larger capacitor will generate a smaller glitch. The op amp used will need to be able to drive the capacitor without oscillating. 

The high value resistors will generate noise on their own. The resistor noise formula is E = √(4⋅R⋅k⋅T⋅ΔF) where R is the resistance, k is Boltzman's constant, T is temperature in degrees K and ΔF is the BW. You can limit the bandwidth with a larger filter capacitor.

The op amp you are using has a BW of only 10kHz so it cannot respond in a timely manner to the transients at the ADC input. 

The LTC1864 sample time (tsmpl) starts after CONV goes low and the conversion is completed. This is the time that the analog input has to settle. You can increase this time by reducing the sample rate. The RC at the ADC input forms an RC time constant and the glitch generated will need to settle to less than 1 LSB. The number of time constants required to settle will depend on the filter capacitor size. A larger capacitor will generate a smaller glitch. The op amp used will need to be able to drive the capacitor without oscillating. 

so i should calculate RC time constant which should be < 1LSB ,right?

This ckt works for SMPS but lots of glitches introduce when adapter is connected. Actually i was trying to build the robust ckt which will work for any power supply whether from SMPS or Adapter or battery.

i will try this solution of RC time constant. Thank you so much.

Calculate how big the glitch will be based on the relative size difference of the sample capacitor and the filter capacitor. Then calculate how many RC time constants it will take for the glitch to settle to under 1 LSB.

I would suggest using a linear regulator as the power source for the ADC and using the power supply to power the linear regulator. This should minimize any noise effects from the external power supply as long as there is a good ground plane and adequate bypassing.

I am  giving Adapter supply to LM2575 which will give regulated 12v and i am giving that to heater and to 7805 to get regulated 5v and this regulated 5V i am feeding to ADC LTC1864.

I will now design a SMPS which will give Pure DC 12v 1A, and try on calculating RC time constant to remove or minimize the glitches and will replace 1uF tantalum at Vref by 10uF.

correct me if i am wrong,

LSB = full scale range / adc code, full scale range = 0-5 v and adc code 2^16 = 65535

LSB= 5/65535 = 7.629 above this range i have to calculate RC time constant.

right?

According to current schematic R= 2.2kohm and c= 10uf,35v so timeconst will be -