AD7176-2 reference and AD8475  gain

Hi,

I my application I need at least 22 bits to measured 1.5V to 4V input voltage, with Output common mode voltage of ~2.55V that source from Fully-Differential Isolation Amplifier (current shunt). This amplifier is located on one custom card (card 1). For that application I thought to use the AD7176-2 cascading to AD8475 (on another card, card 2, located ~70cm from card 1) component but in order to do so, I need you to answer the following questions:  

  1. According to Figure 47 in the AD7176-2 device datasheet, you used 5V external reference (ADR445). Why is that? Is it because the signal that comes from the AD8475 component is between 4.5V to 0.5V (In+/ In-)?
  2. According to Figure 55, In the AD8475 device datasheet, it seems that the gain is 0.2 and not 0.4. I understand it connected somehow to that the differential gain is double the single ended gain… what will be the output voltage in my application and 0.4 gain (1.5V to 4V instead +-10V as showed in Figure 55)
  3. Can you recommend another Precision, Selectable Gain, Fully Differential Funnel Amplifier but with gain 1? (I only need the drives switched capacitor and Σ-Δ ADCs capability with precision to ~22bits).
  4. If the answer to question 3 is "no" (don’t have recommended component), can you recommend another Precision, Fully Differential Amplifier but with gain 2 (I'll amplifier the signal right after the current shunt amplifier (Card 1) just before the AD8475 (Card 2) in that case I might reduce the noises that will probably added to the 70cm between cards)

Thanks a lot,

Idan

http://www.analog.com/en/products/analog-to-digital-converters/ad7176-2.html#product-overview

http://www.analog.com/en/products/amplifiers/adc-drivers/fully-differential-amplifiers/ad8475.html

 

  • 0
    •  Analog Employees 
    on Jul 23, 2017 7:27 PM

    Hi Idan,

    Thank you for interest in ADI products. To answer your questions:

    1. 5V reference was used to maximize the inputs of AD7176-2, which is + VREF ( in this case, the ADC could accept a Differential voltage of + 5V). 5V reference ensures that the attenuated signal from the sensor ( +12.5V) is acceptable by the ADC inputs without the need of further attenuation.

    2. The attenuation factors, 0.4x and 0.8x, is the absolute attenuation value when we measure the output differentially. So for each single-ended outputs, the gain is always half. In figure 55, VOUT_dm = 0.4 ( Vin) = 8 Vpp_dm or 4Vpp_se. For the input of 1.5 V to 4.5 V at gain = 0.4, you would expect an output of 0.6 V to 1.8 V or Vop = (0.3 V to 0.9 V) + Vocm, and Von = (0.3 V to 0.9 V) + Vocm.

    3. Given your input conditions, a driver might be unnecessary. Just to be sure, could you elaborate your sensor output? Is it 1.5 V to 4.5 V (3 Vpp) per single-ended output or 3 Vpp differential out? A clear picture would help.

    Regards,
    Jino

  • Hi,

    First, thank you for your response!

    The input stage that function as current sense driver is "AMC1200" which is a Fully-Differential Isolation Amplifier with common mode voltage of 1.29V (when powered at 3.3V) or 2.55V (when powered at 5V) and with output swing as depicted in figure 14 (and Figure 15 but I think the 5V will be more suitable). With Voh=3.83V and Vol=1.27V and with Vcm=2.55V as output levels I believe that the differential level is 2.56V.

    Can I use this driver (AMC1200) to drive the switched capacitor and Σ-Δ ADCs from ~70cm long between the cards? (As I explain in the previous correspondence, the AMC1200 is located in Card 1 (Load card) and the AD7176-2 ADC will implement on Card 2 (FPGA card)). If you don't sure about that, can you think of any reason why the AMC1200 couldn't drive a precision, fully differential input stage amplifier with gain 1 (such as the AD8656 in Figure 48). With a buffer stage I will gain the ability to driver to Σ-Δ ADCs (AD7176-2), to put analog LPF/ antialiasing without damage the ADC input sampling capacitor and even serve as input signal protection.

    To make my application more robust (amplify the signal over the 70cm between cards, twisted pair and shielded) I thought it will be better to implement a precision, fully differential input stage amplifier with gain 2 (RG = RF = 1kΩ, on Card 1, from the "AMC1200") and then use the AD8475 with gain 0.4/0.8. Can you recommend such an ADC chain? Can you recommend a Low Noise, Precision CMOS amplifier for card 1? (Similar to AD8656 but with output swing of 0V-10V, with selectable gain- even with precision resistors).

    The project ADC volume is ~1000 chains.

     

     

    http://www.ti.com/product/AMC1200/technicaldocuments

    Thanks,

     

    Idan

     

  • 0
    •  Analog Employees 
    on Jul 24, 2017 11:19 PM

    Hi Idan,

    First, we don't support products other than ADI's, so I can't help you with the AM1200 part. Second, what are you trying to sense? Why do you need an Isolation Amplifier? What is your application?

    You could share us your application requirements so that we can provide you the possible solution chain for your application.

    Regards.

  • Hi,

    In my application I need to sense the load current from isolated 300VDC. In order to do that, I used the TI "AMC1200" with floating supply (to the 300V) and with 5V supply (to logic- analog). I had to use this specific module for the current sensing… The rest of the chain (and a lot of component on the custom card) will implement with Analog Devices products.

    Generally speaking, can would you recommend of using a buffer in both ends (right after the "AMC1200" (output stage of Card 1) and right before the Σ-Δ ADCs (input stage of Card 2). (There is ~70cm between cards, twisted pair and shielded). I think so because with a buffer stage I will have the ability to driver to Σ-Δ ADCs (AD7176-2), to put analog LPF/ antialiasing without damage the ADC input sampling capacitor and even serves as input signal protection. For that purpose I thought to use the AD8476 or AD8656 (from both ends). In addition, if you cloud recommend a low noise, precision, fully differential amplifier with selectable gain (even with external resistors…) I might be able amplify the signal from card 1 (from 0V to ~10V), stretch my signal dynamic range to ADC measurement full scale and improve to the SNR (from noise that might picked up during the 70cm between cards). Card 2 will also have the same driver (with different attenuation gain) or even the AD8475 amplifier. This will make my design a lot more robust. What do you think, can you recommend such amplifier? What about the AD8139 from both ends? Can he drive the Σ-Δ ADCs (AD7176-2)? Can he work with VS = 10V, VOCM = 0V with gain ~2? (The input signal is the differential lines with Voh=3.83V and Vol=1.27V and with Vcm=2.55V so I will get Voh=7.66V and Vol=2.54V and with Vcm=5.1V)

    Maybe I should use the AD8476 in (gain 1) both ends? Would it be better (no external resistance…)? 16 bit is fine for my application and the input voltag protection is more important then 20 or more bit resolution.

    Thanks,

    Idan

  • 0
    •  Analog Employees 
    on Jul 25, 2017 10:38 PM

    Hi Idan,

    Thank you for the information.

    Let me deal with the solution after the AMC1200.

    1. We need to choose/decide on how our ADC would work. We must ensure that the voltages will be within the ADC's input voltage specifications. Let's say, we set the REF at 5.0 V, this will allow us to have a Vin_diff= + 5V. So our inputs should be swinging around these limits. We can not go beyond this limits if we wish to extend the capability of the ADC. With that being said, our inputs should be playing around + 5.0 V. Also, the absolute AIN will be limited to AVDD and AVSS. So let's assume AVDD = 5.0 V and AVSS = 0. That will give us an Absolute Input Voltage Range of -0.050 to 5.05 V. This will be the maximum optimized ADC performance.

    AVDD VREF IVR_differential IVR_common mode
    5.0 5.0 + 5.0 V -0.05 to 5.05 V

    2. A fully differential line driver is necessary after the AMC1200 to drive the shielded twisted pair. AD8139 would be too fast for your application. Allow me to suggest you two options, one being ADA4940 and the other is AD8132. See table below for comparison.

    Part # Supply Range Bandwidth Isy Noise Gain Setting Vos
    ADA4940 3.0 to 7.0 V 260 MHz 1.25 mA 3.9 nV/rtHz External Resistors 0.06 mV
    AD8132 2.7 to 11.0 V 350 MHz 10.7 mA 8 nV/rtHz External Resistors 1.0 mV

    If you wish to gain up the outputs of AMC1200, use AD8132 then an attenuator/ funnel amp before the ADC. Just keep in mind that this scheme would mean more noise, noise that would greatly degrade the SNR performance of the signal chain. Each component has its noise contributions, so make sure that this added noise is within your specification. So I would suggest that using only ADA4940 as a driver for the twisted pair will be the most conservative solution for your application.

    Regards,

    Jino