This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ADS114S08B: 4-20mA Topology Questions

Part Number: ADS114S08B
Other Parts Discussed in Thread: ADS131M08, , ADS114S08, ADS124S08

We are working on a project that will use 4-20mA inputs. I was planning to use ADS114S08BIPBS for this but trying to figure out what the best approach is to managing these single ended inputs. I have a question about which general topology is recommended and also some specific questions about the part that I could not determine from the datasheet. 

Regarding the topologies, I was considering these four options. Could someone comment on which would be best?

1.       125 ohm shunt resistor, PGA bypassed, single ended mode, unipolar supply = 3.3V. AINN = GND.

2.       125 ohm shunt resistor, PGA gain=1, pseudo-differential mode, unipolar supply = 3.3V, AINN = 2.5 (direct from voltage reference).

3.       156.25 ohm shunt resistor, PGA gain=1, pseudo-differential mode, unipolar supply = 3.3V, AINN = 0.25V ( voltage divider from VREFOUT).

4.       156.25 ohm shunt resistor, PGA gain=2, pseudo-differential mode, unipolar supply = 3.3V (does this need to be bipolar?), AINN = 1.375 (voltage divider from VREFOUT).

 I created a table to analyze the effective resolution of each approach. In reality we will add some margin to the range (ie 3-21mA) and use "real" resistor values, but these are ideal numbers. 

Approach AINN Shunt R PGA Gain 4mA 20mA 4-20mA
ADC Range
4-20
Resolution (Bits)
4-20
Underfill Factor
Input Voltage ADC Voltage ADC Counts Input Voltage ADC Voltage ADC Counts
1 0 125 1 0.5 0.5 6553 2.5 2.5 32768 26215 14.68 2.50
2 2.5 125 1 0.5 -2 -26215 2.5 0 0 26215 14.68 2.50
3 0.625 156.25 1 0.625 0 0 3.125 2.5 32768 32768 15.00 2.00
4 1.875 156.25 2 0.625 -2.5 -32768 3.125 2.5 32768 65536 16.00 1.00

 Here are my specific questions about the ADC

·         When the PGA is in bypass mode, does that instruct the ADC to operate in single-ended mode? Does the output now generate 65535 counts when the input is = +VREF or still just 32,767? Or is this just a differential input with one side = GND?

·         I didn’t see how to tell the ADC to use pseudo-differential mode instead of differential mode. Does the output still scale from -VREF to VREF in pseudo differential mode? Do I lose half the range?

·         If I use this differential ADC in single ended or pseudo differential mode, then the common mode input to the ADC varies as the input scales. How does this impact the performance of the ADC? Does it create nonlinearities?

·         I am assuming that in order to calculate the accuracy of the ADCs, I will need to multiply the various sources of error by the underfill factor. In other words, a 0.2% error from the reference will result in about 0.6% if my underfill factor is 3X. Right?

·         If I run this in single ended mode, with PGA bypassed, I have no buffer between the shunt resistor and the ADC. I will have to have an antialias filter there that will add source impedance. I think this is not ideal. How are other customers using this for current loops?

·         There’s a note on page 70 of the datasheet that says the range for pseudo differential inputs is VAINN to VAINN + VIN. Does this mean negative voltages are not allowed? I would think that the arrangement shown on figure 100 is not correct. Can I instead connect 2.5V to AINP instead of AINN and achieve essentially the same results or is it best to have the AINP input always be the “input”.

 

  • Hi bbcuf,

    Welcome to the E2E forum! I'm a little confused by your descriptions and the values used in the table as they don't seem to match.  If I understand correctly you are asking if you should use PGA enabled or disabled as well as how to get the maximum converter resolution.  Both the pseudo-differential and single-ended methods have been successfully used by customers.

    First let's discuss the PGA.  If you need to use gain of greater than 1, then you will need to enable the PGA.  If you measure the input relative to AGND and the AVDD supply is unipolar (AVSS = AGND) then you will need to disable/bypass the PGA or use the pseudo-differential method.  As to the impedance issue, yes the PGA enabled offers a higher input impedance, but the impedance when disabled is also quite high and really will only draw a small amount of current with respect to the input bias currents.  With the PGA disabled the absolute input current is typically 0.5nA and with the PGA enabled is 0.1nA.  The difference will be difficult to detect unless you use a very large input resistor.

    For the maximum possible resolution, using the pseudo-differential method allows for the largest possible number of output codes.  The single-ended method only uses the positive range of codes and limits the resolution to a maximum number of 15 bits.  The reason being is that the ADC is always measuring AINP relative to AINN.  If the voltage input on AINP is greater than AINN then the output code is positive.  If AINP is less voltage than AINN then the output code is negative.  So when the internal reference is used and also connected to AINN, a 5V input on AINP is positive full-scale and a 0V input is negative full-scale when using a gain of 1.  The measurement range is +/- 2.5V.  In this case where AINP could be 5V, then the AVDD supply would also need to be 5V.

    So you need to make sure that whatever shunt resistor value you use that the input to the ADC is within both the analog supply range and the input range for the gain being applied.  So it would appear that case 1 would work as described as the shunt is connected to AGND and the voltage across the shunt is 0.5V to 2.5V.

    Case 2 is similar to 1 except that AINN input is elevated so there is no distinct advantage in resolution, however the PGA can be enabled in this use case.  The downside using the pseudo-differential method is the stability and noise at the AINN input.  Any noise or drift of the input will be a part of the conversion result.

    Case 3 extends the resolution as the input range is now 2.5V.  However the measurement will be similar to case 2 in that any noise and drift on the AINN input will be a part of the conversion result.  If you use a resistor divider to create the 0.625V input for AINN you will also add the noise and drift of the resistor network.

    Case 4 does allow for the maximum resolution.  The only drawback to this method is the stability of the 1.875V used for AINN.  This again is similar to cases 2 and 3 where any drift of the resistors used to create the AINN input voltage will directly impact the conversion result.

    As you stated, the values used are ideal.  When determining the final expected result you will need to consider tolerances of the resistors used to ensure that the input range for the ADC along with the reference range of voltage are not exceeded.

    It is difficult to determine which method will work out the best for you in your system.  Sometimes the simplest approach is the best even though it may result in a lower overall resolution in your system.

    As to your underfill questions, I'm not quite sure what you are referring to here.  If you are concerned about the overall resolution, there is no error directly related to the ADC itself other than what is already specified in the datasheet with respect to INL, Gain and Reference error.  The other consideration is error due to noise as is shown in the datasheet.  Total error of your system requires an RSS analysis of all the potential error including the mismatch and drift of resistors from the expected values.  This would include the shunt as well as any voltage dividers you may use for the AINN input.  When executing the RSS analysis all terms must be in the same units.

    Best regards,

    Bob B

  • Thanks for the VERY detailed response. Your analysis of each scenario is very good. You're right that the reason I want to avoid cases 2,3,4 is due to sensitivity to errors on the second input. Especially case 3 and 4 where there are resistor dividers involved. It seems like case 1 with PGA bypassed and single ended input signal is the best approach but I have to look closely at the impact of the filter resistors. Let me follow up with a couple more detailed questions:

    1.I just want to confirm that if I give this a single ended input with PGA bypassed, the output range is 0-0x7FFF, not 0-0xFFFF. Sounds like that is the case, but I was optimistic about the note on page 58 that says setting pga enable to 00 "enables single ended measurements with unipolar supply". 

    2.When I give this single ended or pseudo differential inputs, the common mode input voltage to the ADC scales with the single ended input voltage. The question is how this impacts the ADC performance? My guess is that the ADC performance is characterized in differential mode with a fixed common mode input voltage. Or maybe a more general question is whether the performance specifications (offset error, gain error, etc, INL) are valid for single ended mode or do they degrade somewhat?  

    Taking a step back, maybe there are better parts available. I had been looking at ADS131M08 but wrote it off because it is differential only, but I don't see a difference between driving a single ended signal into ADS114S08B and driving one into ADS131M08. Would that part be more suitable?

  • Hi bbcuf,

    For question 1, the single-ended measurement fixes the AINN input to 0 or AGND.  With a unipolar supply, AINP can not go below AGND so only 1/2 half of the input range is available.  The maximum or positive full-scale is 0x7FFF.

    For question 2 you are correct in that the common-mode does change with respect to the input voltage.  Generally the input range is described as a minimum and maximum with all voltages in between as being valid.  The ADC is characterized with all common-mode voltage range covered within the specified input range.  The common-mode restrictions only apply to when the PGA is enabled.  There is no common-mode restriction for when the PGA is disabled and bypassed.

    There is also the ADS124S08 device that is a 24-bit version of the ADS114S08.  These devices are pin-to-pin compatible so if you need additional resolution you can easily move to the 24-bit version. 

    As for the ADS131M08, this device is primarily used to measure current in metrology applications, such as power meters (ac signals as opposed to dc).  It has much lower input impedance and would most likely need a buffer.  The reference voltage is limited to 1.2V and is a little more complicated to use.

    Best regards,

    Bob B