• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Subscribers 46 subscribers
  • Views 1282 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

ADS124S08: Reference Output in Bipolar Mode; Bipolar Supplies for Unipolar Input

Nicholas DiNino
Prodigy 240 points
Part Number: ADS124S08
Other Parts Discussed in Thread: ADS1248,

Customer Question:

1. On the ADS1248 when used in unipolar mode the negative reference output should be tied to AVSS and in bipolar mode states that the negative reference output should be tied to the midpoint between AVDD and AVSS. For the ADS124S08, the datasheet claims that the reference output should be tied to AVSS always.

 I can’t quite wrap my head around how I would use the ADS124S08 reference output in bipolar mode. I want to use the internal reference buffered and amplified to power the strain gauge excite voltage, which is 2.5V referenced to AGND, but if the reference output negative is tied to the negative supply, what is the positive reference output level be with respect to the midpoint of AVDD and AVSS?, In my design AVDD is +2.5V and AVSS is -2.5V. If the reference is also 2.5V and is referenced to -2.5V, is VREFOUT sitting at 0V (AGND)?

 

2. A further question somewhat related is whether there is any harm in using bipolar supplies when measuring a unipolar input? I also want to read a single ended voltage input that has AGND as the reference input. I did this with the ADS1248, but the change in the reference connections has me concerned.

  • 0
    TI__Guru** 113765 points
    Hi Nicholas,

    There are some significant differences between the ADS1248 and ADS124S08 devices. In particular is the internal reference voltage. The ADS1248 has a 2.048V output whereas the ADS124S08 is 2.5V. It is very difficult to establish a reference with no voltage headroom, so basically the reference can be noisy or unstable with AVDD at 2.5V and the reference at 2.5V referred to AGND. For maintaining the proper voltage and stability the REFCOM of the ADS124S08 should be tied to AVSS.

    I agree it is a little difficult to separate the VREFOUT of the ADS124S08 while operating in bipolar supply mode from being AGND, but the reference really operates no differently than it does in unipolar supply mode. AGND in the bipolar supply case is the supply common. It may not necessarily be 1/2 of (AVDD-AVSS). Let's say the supply ranges from +2.55 to -2.45V referenced to AGND (which is realistic but not ideal). However VREFOUT will be 2.5V referenced to AVSS. In this case the VREFOUT referenced to AGND is +0.5V.

    If the strain gauge is supplied by VREFOUT it may be close to AGND, but it is not AGND. So VREFOUT is it's own voltage node, so do not connect the VREFOUT to AGND.

    There are also differences between using the ADS124S08 and the ADS1248 as it applies to single-ended measurements. The ADS1248 had a restriction for the common-mode input range due to the PGA. In this case the input could not be tied directly to AGND in a unipolar supply mode. The ADS124S08 has a similar restriction, but also allows for the PGA to be bypassed. This means that for the ADS124S08 the measurement can be single-ended even with unipolar supplies when using the PGA bypass mode. So the ADS124S08 can make the measurement with PGA enabled when using the bipolar supplies (the same as the ADS1248) or in unipolar supply mode with the PGA bypassed.

    So there is no harm when making single-ended measurements while using the internal reference and bipolar supplies. For the strain gauge measurement, the excitation is still 2.5V, but will be referred to AVSS instead of AGND. Where you may have an issue is with respect to measurement range. There is an input restriction as it relates to measurement with the PGA enabled where one of the end points can be AVDD or AVSS (as the full-scale range is +/-2.5V). The ADS1248 avoided this issue with bipolar supplies as the maximum voltage that could be measured is 2.048V (and full-scale range is +/-2.048V). With the ADS124S08 the input referenced to AGND has a restriction when the PGA is enabled because the measurement range can span the range of the supply. This would require that the PGA to be bypassed. Due to this restriction, it may be easiest just to use a unipolar analog supply with PGA bypassed instead of the bipolar supply which still has the same input restrictions, but also allows the VREFOUT to be +2.5V above AGND similar to the ADS1248.

    It is really a matter of perspective and which nodes are your reference points. This discussion is somewhat confusing, so please respond back with any further questions.

    Best regards,
    Bob B
  • 0 in reply to Bob Benjamin
    Prodigy 235 points

    Bob,

    One thing that is not clear from the documentation is when the PGA must be bypassed. I read in the datasheet that using a unipolar supply configuration (AVSS = GND) that the PGA must be bypassed and the gain is always 1.

    However, in the user manual for the ADS124S08EVM (SBAU272A) on page 23, the EVM for a bridge configuration is shown using a PGA setting of 128 to read the bridge, However, on page 29, the Analog Supply Configuration section states that the EVM has AVSS tied to GND. The schematics confirm the supply is indeed unipolar. The bridge section discussion seems to contradict the restriction on bypassing the PGA in unipolar mode.

    It seems like there's something I missed in the documentation. Could you explain further?

    Thanks!

    - Doug

  • 0 in reply to Doug Engel
    TI__Guru** 113765 points

    Hi Doug,

    On page 8 in section 7.3 of the ADS124S08 datasheet the input range for various settings are given under the ANALOG INPUTS section of the table and specifically the Absolute input voltage.  Note that the specification is relative to AVDD and AVSS.  

    Regarding the bridge application shown in the ADS124S08EVM user's guide the output of the bridge is going to have an output that is centered about mid-excitation supply.  The excitation supply is connected to AVDD and the AVSS is connected to GND,  So you are correct in that the supply is unipolar.  The reference used is REF0 and this reference is also connected to the same supply as the bridge excitation.

    The example uses a 5V supply, and the sensitivity of the bridge is 2mV/V of excitation voltage with the maximum output of the bridge being 10mV centered around a common-mode voltage of 2.5V ((AVDD-AVSS)/2).  The absolute input voltage would be 2.505V for one input and 2.495V  for the other input for a total difference of 10mV at the maximum loading of the bridge.

    From the datasheet in section 7.3, using a PGA gain of 128 allows for an input of the voltage values between (AVSS + 0.15 + 15.5*|Vinmax|) and (AVDD - 0.15 - 15.5*|Vinmax|).  According to note 3 at the bottom of the table, Vinmax is the maximum differential input voltage expected.  We have already calculated this voltage as 10mV.

    Calculating the absolute input voltage range we find a minimum voltage of 305mV and maximum voltage of 4.695V.  As we have already determined the output of the bridge is centered at 2.5V, we are well within the input range of the PGA using gain of 128. 

    As a side note, we can also determine that a single-ended measurement using the PGA is not possible as you cannot directly tie an input to AVSS.  As the bridge illustration is not a single-ended measurement, but rather a differential measurement centered at mid-analog supply, the PGA can be used.

    If you were to try and measure a single leg of the the bridge (S+ to E- for example), then you would need to disable the PGA and use a gain of 1 as E- is tied to AGND which is also connected to AVSS.

    Best regards,

    Bob B

  • 0 in reply to Bob Benjamin
    Prodigy 235 points
    Hi Bob,

    Thanks for the quick response. Now that you've clarified how the EVM functions, I can tweak it for my design.

    My bridge uses a 10V excitation voltage and has 3mV/V sensitivity, so I'm looking at 30mV centered around 5V (the midpoint of the excitation voltage). That's outside the limits. I still need an instrumentation amplifier to shift the reference, so I don't need to use the PGA if I do that.

    If you happen to have any clever tricks that might help me out, feel free to share them.

    Thanks for your detailed explanation.

    - Doug
  • 0 in reply to Doug Engel
    TI__Guru** 113765 points
    Hi Doug,

    Here is one application where using bipolar supplies would be of benefit. If you connected your bridge to +5/-5V excitation (10V total) you could use the ADS124S08 also using bipolar supplies (+/- 2.5V) and read the output of the bridge directly using the PGA along with the internal reference. The measurement won't be ratiometric, but will allow most of the full-scale range to be utilized when using gain of 64 (FSR +/-39mV).

    There is a benefit of using the internal PGA of the ADS124S08 as opposed to an external amplifier. The external amplifier will simply gain up any noise, but the input referred noise actually lowers with gain on the ADS124S08. Review the datasheet noise tables and observe how the noise lowers with gain. The effective resolution lowers as well, but so does the size of an LSB and there is definitely an advantage.

    Best regards,
    Bob B
  • 0 in reply to Bob Benjamin
    Prodigy 235 points

    Hey Bob,

    That sound like a good approach to the issue. I had a thought last night that perhaps I could use the external reference inputs and precision resistor voltage dividers to create a reference that was ratiometric from the 10V excite voltage. I might look at adding this to the design and compare this with the performance I get with the internal reference and see which one performs better. I imagine I could tweak the external reference voltage with different dividers that could allow me to use even more of the available full-scale range.  On the other hand, am I trying to be too clever here? What are your thoughts?

    Many thanks!

    - Doug

  • 0 in reply to Doug Engel
    TI__Guru** 113765 points
    Hi Doug,

    Using a voltage divider for the reference from the 10V excitation is certainly doable. The consideration is the added thermal noise of the resistors and any drift of the resistor values. The measurement will not be a true ratiometric measurement, but should account for most of the noise and drift of the excitation source. I think it would be worth investigating and comparing to the use of the internal reference.

    Best regards,
    Bob B