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.

ADS1148 - Separating Digital and Analog supply

Other Parts Discussed in Thread: ADS1148, ADS1178, ADS1248, TPS60403, TPS60402, TPS60401

Hi,

Actually, we are designing the circuit using ADS1148. We need to provide analog as well as digital supply. I am planning to have the connection to both the pins from the same supply, so how do I separate the noise from each other? Do i need to have a inductor for DC supply?

Thanx,

Himanshu.V.Ajmera

  • Hello Himanshu,

    Is this a different design from the previous post where you have the VDD supply, and you generate the bi-polar analog supplies +2.5 AVDD using a voltage regulator and -2.5 AVSS using a charge pump and a negative output linear regulator?  Assuming you are using the configuration as above, the voltage supplies will be separated by the linear regulator.  In order to mitigate the effects of noise in the supplies (noise from the digital supplies and from the DC to DC converter), it is recommended that you place coupling capacitors right next to the VDD, AVDD and AVSS supply pins.  The ADS1148 EVM uses 10uF coupling capacitors next to the pins.  In addition, the voltage regulators have usually a combination of parallel capacitors at their outputs for stability and to mitigate noise.  For example, I have seen designs with 2.2uF and 10uF parallel capacitors on the Output of both the positive voltage regulator and the negative voltage regulator.  This in addition to the 2.2uF capacitors at the inputs of the voltage regulator.

    If this is another design where you are using the same supply for AVDD and  DVDD; the same consideration applies.  Use coupling capacitors in close proximity to the supply pins will help mitigate noise. 

    In addition to the coupling capacitors, board layout and grounding is very critical.  You may consider looking into the design layout on the ADS1148EVM user's guide and the overall flow of the layout design. One common problem include the mixing of digital and analog signals where they can cross each other and increase noise; therefore, in general, you will want the analog signals to come in from one side of the board and the digital signals come in from the other side.

     You will also need to be concerned about the ground planes. If you are using a design with switching supply (DC to DC converter); you may need to isolate in the layout the the switching supply part of the circuit.  The ADS1278EVM  user's guide layout attached includes a switching supply and shows the board layout as an example. 

    Please find below a link to "Design Notes" regarding Grounding and layout discussion.

    http://e2e.ti.com/support/data_converters/precision_data_converters/w/design_notes/1390.aspx

    Please find the ADS1178 EVM Rev D user's guide; pages 43- 51 show the schematic and layout; where the board layout isolates the switching supply.

    http://e2e.ti.com/cfs-file.ashx/__key/CommunityServer-Discussions-Components-Files/73/3365.ADS1178EVM_5F00_sbau129d.pdf

    Best Regards,

     Luis

     

     

     

     

  • Hi Luis,

    Thanx for clearing all the doubts.

    It is the same design as what I had shown you, generating +2.5V through voltage regulator and -2.5V through negative charge pump and voltage regulator. Using a coupling capacitor for DVDD is enough?

    According to the your reply, we are planning to separate the analog and digital signal from each other so that the signal won't cross each other.

    We are thinking of a single ground plane for both the analog and digital ground. And, also to isolate the switching supply in the layout. 

    Thanx,

    Himanshu.V.Ajmera

  • Hi Himanshu,

    Which voltage regulator and charge pump are you using? Specifically, at what frequency do they switch at? Using regulators with switching frequencies near the sampling rate of the oversampled delta sigma modulator can significantly degrade noise performance by aliasing into the sampling band. When finding a regulator, I'd suggest looking for a very low ripple one that switches at frequencies that are not integer multiples of the modulator sampling.

    For the ADS1148 and ADS1248 families this corresponds to 32kHz, 128kHz, 256kHz and 512kHz depending on your selected data rate. Even small <5mV ripples at these frequencies can cause degredation without adequate decoupling. It will be very important to provide a significant decoupling capacitance near the ADC and output of the supply source.

    Regards,

    Mike Beckman 

  • Hi Mike,

    Thanx, I didn't knew that switching frequency can be so much important. I am attaching the schematic diagram for your reference.  We are using the bipolar supplies AVDD and AVDS. The regulator used are tps72301 as negative LDO, tps73001 as positive LDO and the charge pump used is tps60403. The switching frequency for charge pump is variable but for the LDOs I am not sure.

    I would like to know as how do we switch the frequency of the charge pump or LDOs. As can be seen from the schematic, we have already placed the decoupling capacitors near the supply line.

    Thanx,

    Himanshu.V.Ajmera

  • Hello Himanshu,

     The TPS72301 and TPS73001 are low noise linear regulators; so they don't produce switching noise. 

    The TPS60403 charge pump has a switching frequency of approximately ~180kHz to 240kHz depending on input voltage and temperature; referring to Fig19 on page 11 of the datasheet.  The other options are the TPS60402 with ~50kHz and  TPS60401 with ~ 22kHz. You may post in the Power Management forum if you need more details for these devices.

    The modulator frequencies for the ADS1148  vs data rate are given on Table 7 page 17. As Mike mentioned, you can make the decision of the charge pump depending on data rate that you expect to use, choosing one that switches at frequencies that are not integer multiples of the modulator sampling.  As mentioned before, using decoupling capacitors at the supply pins of the converters, at the outputs and inputs of the LDO's and at the outputs and inputs of the charge pump will help.  Good board layout considerations (isolating the charge pump) and good grounding techniques are also important.  Using appropiate anti-aliasing filters at the inputs of the ADS1148 will help.  Since you are targeting a 16-bit resolution requirement with the ADS1148, you should be able to obtain optimal results; higher resolution measurements could be more challenging.

    Regards,

    Luis

  • Hi Luis,

    That really helped me.According to the last statement, with 16-bit resolution I would get an optimal result but what if the ADC has higher resolution, maybe if I use ADS1248.  Would 1248 be able to provide same result or would I need to make any changes?

    Thanx,

    Himanshu.V.Ajmera

  • Hi Himanshu,

    As we already discussed on all the posts above, in order to obtain optimal results the user must provide a quiet and stable signal source, reference source and quiet power supplies.  Switching supplies will always have more noise than linear supplies. In your previous posts, since you needed to build a 16-bit system using the ADS1148 and measure thermocouples referred to GND, the solution required bipolar +/-2.5V supplies.  Since your system only had a +5V unipolar supply available,  your design relies on a charge pump to generate the -2.5V supply and a couple of linear voltage regulators. 

    The ADS1248 (24 bit converter) is a higher resolution device; therefore it is able to reveal more subtle changes in your signal and produce more accurate results; provided your overall system is also quiet and stable.  As you try to perform higher resolution measurements, the noise requirements on the overall system become more stringent.   It may be possible to obtain high resolution measurements using switching supplies; but the results will be heavily dependent on providing a quiet, stable signal source, optimal layout, grounding, filtering, appropiate bypass capacitors and how well the switching noise of the supplies was isolated from the A to D converter.  In the past, we have been able to obtain okay results using 24 bit converters using switching supplies provided careful design and layout was implemented. 

    Regards,

    Luis     

  • Hi Luis,

    Thanx for the information. Anyways, we are not planning to use ADS1248. I had written to you to have the knowledge, which could be useful, if in case in future, we would like to upgrade the number of resolution bits.

    Thanx,

    Himanshu.V.Ajmera