AnsweredAssumed Answered

AD7685 4 LSB bit are 1111 (or may be 0000)

Question asked by AlexPhil76 on Oct 4, 2015
Latest reply on Oct 7, 2015 by AlexPhil76

Hello all,

 

I'm using the 3 AD7685 converters in chain mode without busy indicator (mode described in datasheet on p.22 )  and I'm having a problem with one of it.

The first converter in chain does not work properly. The 4 LSB bit are constant 1111 (sometimes it may be 0000). Second and third converters work good.

Similar problem asked and not solved in 2012 year.

https://ez.analog.com/message/44405#44405

schem.PNG

 

 

About schematics - input signal (0-4V) is going to 3 filters(signal <Band 0-1кHz>, signal <Band 0-20Hz> and signal <Band 0-1Hz>). After filters signal is coming to 3 AD7685. Referense voltage is 4.096V(ADR444). I begin test my schematic on constant level from 0 to 4V and near 4V (code 0xFA1F) the first converter jump to next code only by 16 LSB (make constant 1111) other converters works ok jumps to next code monotone and produce small noise about 1-2 LSB. If input voltage rise from 0V to 4V effect raised form near 2LSB (dead zone) above 0.5V,1V to 4LSB at 4V (last bits may be 0 or 1).

Capacitors on REF was changed to 10uF (no changes). I found my schematic error.

I replace first ADC (no changes). I think problem in initializing of chain.

Convertion makes every 200us. After 3 16bit reading makes new conversion and wait near 175 us for start clocks and read data from ADC chain.

Width CNV pulse 500ns. Period of SCK 500ns (2 Mhz). Delay between 16bit reading from different ADC 500ns (1 SCK period).

NewFile2.bmp

NewFile1.bmp

On next graph (with persist on) you can see jumps of 2 LSB on second converter (on graph only 3 LSB bits) and all 16 bit of problem ADC (jumps only bit 4)

 

NewFile3.bmp

 

Delay between last clock and new convert

NewFile5.bmp

 

What do you suggest me?

Outcomes