I am trying to select and ADC with single ended inputs which can input positive and negative voltages (around +500mV to -500mV). Due to the nature of the system I am selecting it for, the voltages cannot be DC biased and also cannot be amplified (in other words the exact DC level is the information). I was wondering if it would be possible to set the reference voltages to REF+ = 500mV and REF- = -500mV, while the power supply is 2.5V to -2.5 V (the ADC allows for true bipolar operation). For instance the ADC I am looking at is an Analog Devices AD7176-2. I noticed in the data sheet that the limits for the reference is that their total voltage difference can be 1V, but the tests they have are only for +5 V to Ground supply, not bipolar operation. Hopefully someone has experience with operating an ADC in this mode. Thanks
-
\$\begingroup\$ Working with 24-bit ADCs isn't easy. A question that comes to mind is, Why not just use the device's internal reference? With fully-differential inputs, you don't need to keep the inputs strictly within the VSS-to-REFOUT range. \$\endgroup\$– The PhotonCommented Mar 24, 2013 at 16:29
-
\$\begingroup\$ Well, could you elaborate? DC voltage is the output of my system and it can range from positive to negative. Also the internal reference would be higher, I think it is 2.5V so your voltage step size would be 2.5V/(2^24). If I could just make it within the limits of my output then I would only have 1V/(2^24). Basically I am looking for a way to resolve the exact DC level (positive and negative), very small values, with high precision. (World's best ADC haha) \$\endgroup\$– mkitchenCommented Mar 24, 2013 at 16:38
-
\$\begingroup\$ My concern is, it will be very difficult to generate your -0.5 to +0.5 V reference with 24-bit accuracy...maybe you will lose as much precision from errors in your external reference as you do from having 2.5 V instead of 1.0 V reference. At the same time you add expensive high-precision external parts for the 1.0 V reference, and also increase the risk of a design error due to added complexity. \$\endgroup\$– The PhotonCommented Mar 24, 2013 at 16:42
-
\$\begingroup\$ I was concerned about that too, but that's where these components come in: analog.com/static/imported-files/data_sheets/ADR130.pdf \$\endgroup\$– mkitchenCommented Mar 24, 2013 at 16:46
-
\$\begingroup\$ Not bad. AD7176-2's reference has better initial accuracy and tempco, but ADR130 does win on long term stability. \$\endgroup\$– The PhotonCommented Mar 24, 2013 at 16:53
1 Answer
The datasheet of the AD7176-2 is pretty careful to always refer to specific pin names, like AVSS or AVDD, rather than just generically talking about "ground".
For example, when they talk about the input voltage range (p. 4), under the line "Absolute AIN voltage limits", they give the min limit as "AVSS - 0.050" and the max as "AVDD1 + 0.05". Similarly when they give the "Absolute Reference Input Voltage Limits" (p. 5), they give the same limits. As long as you obey these limits, the chip should work as advertised.
In particular, if your AVSS is -2.5 V, then REF- of -0.5 V is within the limits.
That said, if you can get the attention of an Analog applications engineer (like if you are doing this for a commercial project or at a university), it wouldn't hurt to show them your final design before sending it out for fab.
-
\$\begingroup\$ Thanks for the answer. I will probably be getting the evaluation kit for this chip just to simplify my own design (for university). I was speaking to one of their engineers and they didn't seem sure about the reliability of operation of the ADC with references such as this. But then again, the engineer didn't know that this chip in particular would even operate in bipolar. I need to speak with them again. \$\endgroup\$– mkitchenCommented Mar 24, 2013 at 16:30
-
1\$\begingroup\$ See if you can talk to a factory apps engineer who specifically supports this product line. I find Analog's product range is too broad for their field apps guys to be able to know every detail that you might need to ask them about. \$\endgroup\$ Commented Mar 24, 2013 at 16:33