Need help deciphering/circuit analyzing this ADC setup

[Mach3Maelstrom]

I have a design that used to use one type of ADC, but for design reasons I must switch to a related ADC of similar pin layout (I already have the bare boards).

The problem: My design worked flawlessly with the old ADC, but I have problems with reading data from the new ADC and I need to analyze why.

Old ADC, ADS7820: **broken link removed**
New ADC, ADS8504: Link to datasheet

Summary of major differences between the two chips:

• ADS7820 uses straight binary logic for its 12-bit output, ADS8504 uses two's complement.
• ADS7820 can only take analog values between 0V - 5V. ADS8504 can take analog values between -10V - 10V.
• My analog signal will ALWAYS be between 0V and 2.5V, so both differences above shouldn't make much of a difference

For ADS7820 (pic found on page 9 of the datasheet):

The purpose of that potentiometer-circuit is to give REF a 2.5V reference, so I installed a 2.5V source there direct to pin 3 (REF) instead. The analog signal would feed into pin 1 directly. This circuit would work flawlessly in my design as it portrays the 12-bits of information to my CPLD.


For ADS8504 (pic found on page 14 of the datasheet):

From what I understand, the datasheet mandates that I install a voltage divider (Offset) to pin 1. It still wants 2.5V on REF, so I want to still use my 2.5V reference to provide that accurately.

Two things I can't understand are:
1) Why is this voltage divider needed (the 200R and 33.2K divider)? The datasheet states it provides overvoltage protection but, as I stated above, I never go over 2.5V.
2) What voltage are they suggesting the net Offset (between the left pot & the 33.2K resistor) to be? This isn't clear to me in the datasheet how to ascertain this value. Hooking this directly into the 2.5V reference won't work as per my bench experiments.

Any help would be greatly appreciated. Thanks in advance!

 

[iop95]

By offset you may adjust "zero" for ADC and by gain may adjust full scale limit. Check **broken link removed**

 

[Mach3Maelstrom]

iop95 - Thanks for your response. That article is terrific, it elegantly explained the purpose of offset and gain to me.

It was brought to my attention that there's another difference between the two ADCs that I didn't realize: The LSB is 1.22mV for the ADS7820 and 4.88mV for the ADS8504. After changing the software to account for this difference and after building in the offset with the voltage reference, I essentially have the end signal I want.

There's just two more issues left to address:

• For the first 0.5 seconds, my ADC reads a zero. Then it finally displays the data. I think this may be the offset error.
• The data appears choppy/noisier than the ADS7820 circuit. I'm uncertain how to clean this since the analog signal (Vin, pin 1) is clean when I hook it up to an oscilloscope.

 

[iop95]

0.5s at power-on? This may be needed time for "setup and checking for good conversions" at power-up. From datasheet 8504 seem to be easy noiser than 7820 and faster sampling capability may conduct to noiser output. But to eliminate external noise source, check by connecting a battery with as short cable posible to convert and see if there are difference. Also check for ground loops and for noise caming from power supply.

 

[Mach3Maelstrom]

Thanks, I'm not familiar with identifying if one ADC will be noiser than another - how do you determine that?

Also, to ensure I understand you correctly, you're proposing connecting a battery to Vin (pin 1) or as an alternative power source to Vdig & Vana?

 

[iop95]

I propose to test if there are noise output when connect a battery (which have theoretichaly no noise) at Vin instead of actual voltage source. Noise may cam also from power supply, PCB design or ground loops.