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.

LM27762: Mixed signal PCB design questions

Part Number: LM27762
Other Parts Discussed in Thread: ADS1120

Tool/software:

I'm designing a PCB to power a sensitive analog sensor and read a differential analog signal (+/-10 mV) from the sensor. The input power to the PCB will be a noisy 5V, so I thought I'd go with a LM27762 to provide me with a clean +/- 2.5V power for the sensor (which requires 5V to operate) as well as serve as a voltage reference for the ADC. Cost is not a primary concern for this PCB.

This is my first time designing a mixed signal PCB. I've read the "Grounding in mixed-signal systems demystified" series by TI, and I've read up on this from other sources as well.

Is my decision to go with a bipolar power supply a suboptimal choice? Almost all of the materials online assume that I have a single-ended power supply, and thus instruct me on how to deal with digital and analog current return paths in the ground plane. I think I got the hang of that, but I don't quite understand how that carries over to a bipolar supply. For example, do I still fill the entire ground plane with 0V? In that case, are there guidelines on how I should route the -2.5V and +2.5 rails?

My sensor's cable also has a shield connection. I'm confused about how I should connect that as well.

  • Hi Vladimir,

    Thank you for using E2E,

    My collogue Varun will respond to your quarry tomorrow. I hope that should be fine.

    Regards,

    Moheddin.

  • Hi Vladimir,

    Is my understanding correct that you plan to connect the -2.5V to the GND pin of your sensor, so that sensor is referenced to -2.5V? It should be possible to do this, however the easiest and lowest risk option is to use a unipolar supply. Could you first check if the noise on the 5V rail can be cleaned with low pass filtering. Do you know frequency range of the noise on the 5V rail?

    The LM27762 uses an LDO internally to step down the 5V to the +ve output. So this should help clean up the +2.5V output. For the negative output, an internal charge pump first converts the +5V to -5V, before an LDO converts the -5V to the-2.5V. The switching nature of the charge pump will add some extra noise, but the LDO should help attenuate this. 

    Are you planning to use a TI analog sensor? It would be good to ask the sensor team their opinion on the grounding for bipolar input supply to the sensor.

    Best regards,

    Varun

  • Hi Varun,

    Thank you for your reply. Your understanding about my sensor connection is is correct. My sensor is a load cell - it's a resistor bridge, so if I supply +/-2.5V, my signal will be centered around 0V.

    My incoming 5V power will be coming from customer-supplied USB power adapters, so the noise can be decently high and of unknown frequency. I am hesitant to DIY my own low pass filter, any suggestions will help. My interest in the LM27762 was in large part due to its low noise performance while maintaining 5V between AVDD and AVSS.

    For the ADC, I'm planning to use TI's ADS1120. I can post a question in the ADC forum if that's what you'd recommend.

  • Hi Vladimir,

    Since the ADC's sampling frequency is relatively low, the cut off frequency for the low pass filter would also have to be low. If you make an LC low pass filter, you may need quite large values for the L and C and this maybe impractical. Using the active filtering from an LDO can be useful in this case.

    Could you kindly create a post for the ADS1120 to address your questions on the bipolar supply and grounding?

    Best regards,

    Varun