A strange behavior

Hallo,
I have made a small microcontroller based on microblaze.
I have connected a differential 16 bit adc to my system.
The adc takes input from a opamp for testing.

The system works well only if I measure voltage between chipselect of adc 
and gnd. In this way it shows every hex number in range: 0 to 7FFF (it is 
two complementer).

If not, I can see only 4-5 numbers of the acquisition range: FB, 1F1F, 3E3E, 
7C7C.

Which trouble could produce a so strange behavior?

Many Thanks
Marco 

Look closely at the datasheet for the A/D converter.  How do they say you
should acquire a reading and how are you trying to acquire it?


"Marco" <marcotoschi@nospam.it> wrote in message
news:df7380$h6e$2@news.ngi.it...
> Hallo,
> I have made a small microcontroller based on microblaze.
> I have connected a differential 16 bit adc to my system.
> The adc takes input from a opamp for testing.
>
> The system works well only if I measure voltage between chipselect of adc
> and gnd. In this way it shows every hex number in range: 0 to 7FFF (it is
> two complementer).
>
> If not, I can see only 4-5 numbers of the acquisition range: FB, 1F1F,
3E3E,
> 7C7C.
>
> Which trouble could produce a so strange behavior?
>
> Many Thanks
> Marco
>
>

"John_H" <johnhandwork@mail.com> wrote in message 
news:bDHRe.27$tN4.242@news-west.eli.net...
> Look closely at the datasheet for the A/D converter.  How do they say you
> should acquire a reading and how are you trying to acquire it?
>

To start acquiring chipselect must go high  for 1 clock cycle.
At the falling edge the adc starts conversion. (it's falling edge 
sensitive).
After 5 clock cycles the serial data out goes from Hi-Z state to 0.
After 1 clock cycle it starts sending data through serial data out for 16 
clock cycles.
Then it goes into power down state and waits for a chipselect.

I have made a state machine which follows that. Into post-place simulation 
works well.

Marco