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.

Circuitry required to convert single ended signal to differential signal for 16-bit ADC

Part Number: TMS320F28377D
Other Parts Discussed in Thread: THS4531, TINA-TI, THS4531A, THS4551, OPA835, OPA836, LM27762

Hi All,


I am using TMS320F28377D for my inverter application. I need to use 16-bit ADC to get precise result. I have single ended input signal but 16-bit ADC only used in differential mode. So, my first requirement is to convert single ended signal into differential signal. Kindly suggest circuitry for the same.

I have to maintain constant 1.65 common mode voltage (Because of ADC VREFHI = 3.3V and VREFLO = 0V). For example, My single ended input is 2.2V then my differential signal output should be Vin+ = 2.2V and Vin- = 1.1V.

Vcm = (Vin+ + Vin-)/2 = (2.2 + 1.1)/2 = 1.65V.

The maximum differential input voltage should grater than 3.3V. I have refered THS4531 but its differential input voltage is only 1V. Which IC i should prefer?

Thanks and regards,

Maulik Timbadiya

  • Hello Maulik,
    1. What power supplies are you going to use for the ADC?

    2. Your input signal you mentioned was 2.2Vpp; is that a 0-2.2V signal or is that on some other common-mode dc level?

    3. What is the speed (sampling rate) of your ADC(1.1MSPS??)

    The differential input signal spec you are seeing is an ABS max rating. The way an FDA works is that the two always track each other (like the virtual GND concept in a standard op-amp). Once you answer questions 1 and 2, I will mock up a TINA-TI circuit and send it over to you.

    -Samir
  • Dear Samir,

    Thank you.

    I am using TMS320F28377D controller.

    1. In which ADC is powered by +3.30V DSP supply and i have applied +3.30V on VREFHI pin and 0.0V on VREFLO pin.

    2. My input signal varies from 0 to 3.3V. Which is single ended sensor output. I want to convert it into differential foam. I need to maintain 1.65V (=3.3/2 as per DSP datasheet) common mode voltage.
     For example,

    2.1 If my sensor output is 2.2V then my differential output should be V+ = 2.2 and V- = 1.1 so, my common mode will maintain by = (2.2+1.1)/2 = 1.65.

    2.2 If my sensor output is 0V then my differential output should be V+ = 0 and V- = 3.3 so, my common mode will maintain by = (0+3.3)/2 = 1.65.

    2.3 If my sensor output is 1V then my differential output should be V+ = 1 and V- = 2.3 so, my common mode will maintain by = (1+2.3)/2 = 1.65.

    3. Yes it is 1.1MSPS.

    Regards,

    Maulik Timbadiya

  • Maulik,
    To summarizeyour setup

    1. Single-Ended Input Signal : 0 to 3.3V and thus input common-mode is 1.65. Supply voltage is 3.3V.

    2. You are setting VREFHI = 3.3V and VREFLO = 0V, thus the input common-mode to the ADC is also 1.65V.

    3. Each single-ended input of the ADC can take a (VREFHI - VREFLO) = 3.3Vpp signal, thus the differential input to the ADC is 2* 3.3Vpp = 6.6Vpp. This implies that the THS4531A (this has better precision than the THS4531) should be in a gain of 2. Note however that the THS4531A output is guaranteed to only swing 0.2V from each supply rail, hence each output of the THS4531A will swing (3.3V-0.2V) = 3.1V and (0V+0.2V) = 0.2V. This makes the differential output of the THS4531A = 2*(3.3-0.2)V = 5.8V and thus its gain configuration should be 5.8V/3.3V = 1.76V/V.

    4. Based on the above conditions I can suggest a circuit for you, however I need to know a few things:

    a. What kind of sensor is driving this ADC ( I assume that the signal is coming directly from the sensor and is not previously buffered). The reason I ask this is that a lot of sensors do not have low output impedance so in many cases I have seen the signals being buffered by a regular op-amp before driving the ADC.

    b. I would suggest using the DGK package of the THS4531A. This is pin compatible with the faster and less noisy THS4551. In case we are limited by the THS4531A linearity and noise then we can easily upgrade to the THS4551.

    c. Is the signal from the sensor a pulse signal or more sine-like? In case it is a pulse signal then the required amplifier bandwidth could be determined by the rise and fall time of the pulse, rather than the frequency of the pulse. Please clarify this as well.

    -Samir
  • Hi samir,


    Thank you.

    I think there is misunderstanding on following statement, "thus the differential input to the ADC is 2* 3.3Vpp = 6.6Vpp". I want to convert single ended signal to differential signal. Both signal input to ADC is 3.3 Vpp. Please find attached waveform that i want to apply on ADC. I want to apply VoutP on ADCINA0 and VoutN on ADCINA1. My buffered output is VoutP(from sensor through opamp) only but by using THS4531 i want to generate VoutN like signal. And those VoutP and VoutN (which is differential) will drive my 2 pin of ADC.

    4.a There is a Hall Effect sensor and signal is buffered by opamp first and then it will drive ADC.

       c The signal from sensor is sine-like.

    Regards,

    Maulik

  • Maulik,

     I am not quite sure what the misunderstanding is,

    Your ADC requires a differential input, that differential input is 6Vpp. isThis is also what you have shown in your graph, i.e two single-ended signals of 3.3Vpp each, riding on a 1.65V common mode.

    The THS4531A will take a single ended 3.3Vpp signal and convert it into a differential output that feeds the ADC. The THS4531A however, if running on 3.3V supplies cannot swing all the way to the rails...which is why I said the gain configuration will be 1.76V/V and not 2V/V. If you configure it in a gain of 2, the signal will get clipped.

    Use the circuit in Figure 95 of the THS4531A datasheet, however make Rf = 2kOhm and Rg = 1.15kOhm. This will give you a gain of 1.76V/V.

    According to table 5-49, the sampling capacitor is 16.5pF. The external charge reservoir capacitor should be around 50*16.5pF = 800 pF (C5 in the attached TINA ckt). You can use a 50 ohm series resistance (R5 and R6)to drive this. Also keep a place holder for 100pF common-mode capacitors (C4 and C6 in TINA ckt).

    Note however that you are going to need a low impedance source to drive the VCM input on the left hand side of the R4 resistor. You can use the OPA835 for this.THS4531A.TSC

    -Samir

  • Dear Samir,

    Thank you for help.

    I have attached TINA-TI oscilloscope result herewith and VM1&2 is varying from 1.65 to 3.3V. I am expecting same waveform for 0 to 3.3V as same as my previously attached graph.

    When VM1 = 3.3V at that time VM2 = 1.65 and V(CM) = (3.3+1.65)/2 = 2.475. I am expecting VM2 = 0V when VM1 = 3.3V.

    Please read this thread once: e2e.ti.com/.../2100577

    Regards,

    Maulik Timbadiya

  • Hello Maulik, For some reason the THS4531A TINA model is struggling to get to 200mV of the positive rail. I changed the gain a little bit by increasing the 1.15kOhm to 1.3kOhm. I also changed the output loading by increasing the resistors and reducing the caps. It works well now.

    I usually run a transient simulation vs an oscilloscope reading. See attached file for circuit and results. The Vcomm output will give you the actual measured output common mode which is now 1.65V as expected.5707.THS4531A.TSC

    -Samir

  • Dear Samir,

    Thank you for your valuable prompt response. Now common mode is fine.

    Why VF1 lower lower voltage level is 383mV and upper is 2.92mV? (Instead of 0V and 3.3V)

    On practical implementation, can i use resistor divider to generate 1.65VCM?? What u will recommend?

    Regards,

    Maulik

  • The amplifier is not a rail to rail opamp. Look at the "Output" datasheet specification for example on page 4 of the datasheet for 2.7V supplies. It states that the amplifier output can only swing 200mV from the supply rails.

    You cannot use a resistive divider since it isn't a low output impedance source. You will need a voltage resistive divider that is then buffered by an opamp such as the OPA836 to create such a source.

    -Samir
  • Thank Samir,

    Now simulation is running fine. I had slightly modified your simulation. I had modified gain to 2V/V, Vs+ = 4V and Vs- = -1V.

    Now i am going to check it practically. i will let you inform after test.

    Thanks again.

    Warm Regards,

    Maulik

  • Maulik,

        With 4V and -1V supplies you have plenty of headroom. You can easily use the full input range of the ADC. I should mention, that with 4V supplies and 3.3V on the ADC, there is a chance that if you have a spurious large signal into the amplifier then its output could go close to 3.8V...in which case the ADC my get damaged.

    There are a couple of things you can do:

    1. You already have some built in protection with the 50-ohm resistors on the outputs of the amplifiers. This will help to limit the current a little bit. For example with 2mA of output current you will get a 100mV drop across the resistors. Increasing those resistors to 100-200ohms will further increase the level of protection at the cost of settling time.

    2. You could drop the power supplies to around 3.7V....this way, the amplifier can swing to around 3.5V (actually, worst case scenario, it can swing closer to 3.64V), however 200mV-400mV overvoltage shouldn't damage the ADC. You also have the additional protection from the 50-ohm series resistors from before over here as well.

    -Samir

  • Dear Samir,

    I have some confusion regarding to "ABSOLUTE MAXIMUM VOLTAGE AND OPERATING VOLTAGE". I found on datasheet pg no.2, supply voltage Vs- to Vs+ = 5.5V.

    Can i apply +3.8V to Vs+ and -1.5 to Vs-? This will make my total supply voltage 5.3V. So, Can i use it for continuous operation?

    I have selected LM27762 IC to supply THS4531. (Do u have any comment on it?)

    Thanks and Regards,

    Maulik Timbadiya

  • Maulik,
    You should never exceed the ABS MAX RATING. The ABS MAX and MAX OPERATING VOLTAGE can be the same, however sometimes we make the MAX OPERATING VOLATAGE about 10% less than the ABS MAX, since the LDOs and other supplies being used to power the device have their own variation.

    You are okay running on 3.8 and -1.5V supplies.

    I am by no means an expert on power products, however briefly looking at the LM27662, I cannot see any issues using it in your application.

    -Samir
  • Thank You Samir,

    Now I am using THS4531 IC for producing differential output for 16 bit mode ADC. I want to apply input from LMH 4463 which is of 1.69v as I am applying this on Pin no. 8 which is Vin of THS 4531 I am getting 0.62v and 0.63v as differential output and also the output of LMH 4463 becomes 0v (near about) but after connecting. When I am applying 0v or not giving any signal to Vin I am getting 0v and 3.39v but when I am applying any other input I am not getting the required output. And same happens when I apply ac input of 1.65v from function generator on Vin I am still getting 0v DC, the frequency of 1.65v is 50hz.What can be the issue?

    Regards,

    Ami and Mona

  • Dear Samir,

    Can you help in recent post by ami and mona?

    Regards,
    Maulik Timbadiya
  • Hi,

    Have you tried to simulate your circuit in TINA-TI to make sure it works? If you are trying to debug your circuit on the bench, then would it be possible for you to attach your schematic, and briefly explain the input and output voltage levels you are providing/expecting?

    Best Regards,

    Rohit