• State Resolved
  • Locked Locked
  • Replies 16 replies
  • Answers 2 answers
  • Subscribers 48 subscribers
  • Views 3026 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.

ADS1248: What is the actual ESD protection level on the analog inputs and GPIOs?

Steve DeCabooter
Prodigy 20 points
Other Parts Discussed in Thread: ADS1248, ADS1148, TPS60403

I could not find a spec for the # of kV supported by the ADS1248's ESD protection on the analog inputs and GPIOs.

Can someone help me with these numbers?

Thanks.

  • 0
    TI__Guru** 113765 points

    Steve,

    2kV HBM and 500V CDM.

    Best regards,

    Bob B

  • 0 in reply to Bob Benjamin
    Prodigy 175 points
    Bob

    We are measuring 7 thermocouples with one side grounded. The positive lead will have 1k series resistance + 10uF ceramic to ground. Would you recommend external ESD protection for the ADS1248?

    Thanks,
    Tom
  • 0 in reply to Tom Russell
    TI__Guru* 97585 points
    Tom,


    Before we go into the ESD question, I want to check something in your setup. When you say that you have 7 thermocouples with one side grounded, where exactly is the "grounded" connection? If you are using a bipolar supply (±2.5V for analog supplies) and have the grounded connection connected to 0V, that's fine. If you are using a unipolar supply (0V to 5V) and have the input connected to 0V, then the input will be outside the common-mode input range of the ADC.

    The input of the ADS1248 looks like the front end of an instrumentation amplifier, with two chopper-stabilized op-amps. There is an input range that extends from AVSS+0.1V to AVDD-0.1V. There is also an output range that goes to the same points. Since the input common-mode voltage and the output common-mode voltage must be the same, the output restricts what the inputs can be as the gain is increased. The limitation of the input common-mode voltage is given in Equation 4 on page 26 of the datasheet.

    Regarding your ESD question, I don't think I have a good answer for you. I've seen plenty of cases where there is some external ESD protection added to the circuit and many cases where there isn't any. The 1k series resistance does add some extra protection to your circuit, but I don't know of any quantifiable way of say how much extra protection it adds. In the end, you would likely have to test what this does for your circuit, with a real ESD test.

    I will say that adding the series resistance does give some error because of the analog input currents from the device. Also, external ESD devices will also draw current. If you do decide to use external ESD structures, make sure that the leakage currents are extremely low to make sure that the error from these currents reacting with the series resistance isn't large compared to the measurement.


    Joseph Wu
  • 0 in reply to Joseph Wu
    Prodigy 175 points
    Joseph

    Thanks for the excellent support. The thermocouples are indeed connect to 0V to make sure the common mode voltage is in range. We also do not have any negative input voltages from our thermocouples. I think 1k or even a bit more series resistance should be fine as the input bias current at 20 sps is only 0.5nA so we get 0.5uV of error. For type K thermocouple we have about 40uV per degree C.

    Tom
  • 0 in reply to Joseph Wu
    Prodigy 175 points

    Joseph


    Thanks again for your continuing support.  If I do add ESD diodes, I have seen two possible configurations:  1) the diodes are arranged back to back and placed across the thermocouple inputs or 2) one diode goes to ground and the other goes to AVDD.  In case 1, technically one diode is forward biased, but the thermocouple voltage maximum is only 20mV so not likely to be much leakage.  Which arrangement would you recommend?  Keep in mind that our thermocouples are wired single ended but we have a +/- 2.5V supply for ADS1248.

    I would like to use AVDD as the reference voltage using the LM4132A-2.5, so connecting REFP0 to AVDD and REFN0 to ground.  The -2.5V will come from an small inverter such as MAX1720.  Any concern with tieing REFP0 to AVDD?  I don't think the internal reference is quite good enough to give sub 1C accuracy.

    Regards,

    Tom

  • 0 in reply to Tom Russell
    TI__Guru* 97585 points
    Tom,


    I don't have a lot of experience with the external ESD structures used in these circuits. However, I don't see that back-to-back diodes across the thermocouple will do very much. Even on an ESD hit, can it generate enough voltage to forward bias either diode? Where would the ESD current get shunted to?

    I would stick with the structure with two diodes; one that attaches to GND and the other that attaches to AVDD. A low going pulse latches the ESD pulse to GND, while the high going pulse shunts the excess current to ground. It's a common structure that is often used for ESD. Regardless of what you want to use, note the leakage current and see how much error this gives with any series input resistance on the front end of your ADC.

    I don't have any experience with the MAX1720. In general, I would generate a greater voltage and use an LDO to clean out the pulses (say generate a noisy -5V supply and use an LDO to clean it up, and lower it to -2.5V). I'm sure this switched cap supply might have some noise, but it's hard for me to quantify.

    For the 2.5V reference, you want to use it as the +2.5V supply too? Is this for the accuracy? Admittedly, the reference accuracy of 0.05% is great so the DC accuracy is good, but look at the noise. The noise from 0.1Hz to 10Hz is typically 170uVpp. This is such a large amount of noise that you probably don't need a 24-bit converter, and could go with the ADS1148 as a 16-bit converter.

    In the end, the ADC measurement is a comparison between the input and the reference. If this reference value has 240uVpp of noise, then a full scale measurement is going to be read back as having a 240uVpp noise even if it is a completely noiseless source.


    Joseph Wu
  • 0 in reply to Joseph Wu
    Prodigy 175 points

    Joseph,


    I did want to use the +2.5V as both the external reference and as the positive supply.  I can choose a much better reference such as LT6654 or REF5020.  The ADS1248 uses only 225mA for analog and REF is 30nA so even 5 of them only need about 1mA.  Of course, I can use a separate +2.5V supply and 2.048V reference if you advise that.  If I use a separate external reference can the reference connect COM to REFN1 and OUT to REFP1?  The eval board seems to have the reference COM to AVSS.  Also, I will have 5 ADS1248, so how much capacitance should I put at each one.  The eval board shows 10uF, but hard to believe I need that much capacitance at every REFP1 input and every AVDD and DVDD.


    I am going to put the ESD diodes between AVDD and GND, but take a look at Fig 44 on the last page of LTC2484 datasheet and you will see both diodes connected to GND.  I suppose this is okay because the thermocouple voltage is so low, the forward biased diode does not get turned on.

    Thanks again,

    Tom

  • 0 in reply to Tom Russell
    TI__Guru* 97585 points
    Tom,


    Since the noise performance of the ADC depends on the reference, I would try to keep the reference as clean as possible. This would include not using the reference as the supply. I'm not saying that it won't work, because it certainly may, I just would just keep it separate.

    The external reference would not be connected to VREFCOM and VREFOUT. This is used for the internal reference. I would simply place a capacitance between the two and connect VREF to a low impedance ground. Even if you don't normally use the internal reference, you might use it to generate current from the IDAC, or use the system monitor. At some point you might even decide that the internal reference is good enough to use.

    I'm not sure what capacitance you're talking about. Is this for the reference inputs? or for VREFOUT to VREFCOM? or is this for supply bypassing? Regardless of whichever this is, I'd generally use as little as possible. For bypassing, I find that it's just cleaner to have the device shut down when power is removed, instead of having a large bulk capacitance to keep the device on. If it's for the reference, then lowering it will help with the turn-on time of the reference. Once you start the board, a reference cap may take a long time to charge. If you power down the device intermittently, you'll have to wait each time.

    The back-to-back diodes now make sense - I didn't know that they were connected to the ground.


    Joseph Wu
  • 0 in reply to Joseph Wu
    Prodigy 175 points

    Joseph


    I will follow your advice and use an external reference but still provide capacitance at REFOUT just in case I need it. Sorry if I was not clear, but I intend to have a external reference which is 2.048 above GND so would be connecting REFN1 to GND and REFP1 to the external reference output.  I assume this is okay, rather than having the external reference have its negative lead to AVSS. As for the capacitance, I was asking how much capacitance I need at AVDD and at REFP1 for supply bypassing.  The eval board has 10uF, which seemed like quite a lot, particularly for the reference input.  Also, I will have 5 ADS1248 close together.

    Again,

    Thanks,

    Tom

  • 0 in reply to Tom Russell
    TI__Guru* 97585 points
    Tom,


    Connecting the external reference to ground (so that the reference output is at +2.048V) will be fine. Both reference inputs will be within the range of operation. As for the capacitive bypassing, I would also say 10uF is a bit much. I'd go with 1uF or lower for each.

    If you're willing to share a schematic, or part of a schematic, I'd be willing to review it when you're ready.


    Joseph Wu
  • 0 in reply to Joseph Wu
    Prodigy 175 points

    Joseph


    Here is a portion of the schematic.

    There are 32 thermocouples which are wired single ended.  They are wired with 1k + 10uF filter and ESD diodes. The ADS1248 gets +5V from ADuM5401 isoPower with 10uF output capacitor.  The AVDD, AVSS and REF supplies are shown below.  There are a total of 5 ADS1248 which all share the same analog supply and REF.

    My question is do I have the right values of capacitance?  Anything else I need?

    Thanks,

    Tom

  • 0 in reply to Tom Russell
    TI__Guru* 97585 points
    Tom,


    Did you forget to attach the schematic?

    Without seeing it, it sounds like the capacitance you have for input filtering is very high. I don't know what data rate you're using, but the filter for anti aliasing needs to go out to 1/2 of the sampling rate (same as the modulator clock frequency), not the data rate. In a delta-sigma converter. the sampling rate and the data rate are not the same.

    Regardless, I think 10uF is too high. I'd stick with 1uF for even bulk capacitances, and even less for filtering.


    Joseph Wu
  • 0 in reply to Tom Russell
    TI__Guru* 97585 points
    Tom,


    Sorry about missing this reply. I didn't get a chance to answer this at the end of last week.

    I'll throw out a few comments on for this schematic. First, I've never used this TPS60403, so I'd be a bit concerned that there's would be some ripple associated with this device generating the AVSS line. You can compare it with the PSRR spec to see if this might be an issue.

    I don't know the Maxim reference, but I'm sure it's fine. It looks sufficiently low noise to use with the ADC.

    How do you have the grounding setup? I noticed there's a GND to ISOGND connection connected by a 0 Ω resistor. In general, we recommend one single ground plane, but consider the return currents for everything. If it helps, think of it as separate analog and digital grounds and connect the two together under the ADC.

    You show one input with a filter of 1k and 10uF. I still think the 10uF cap is high. Remember that the device is oversampling, so the digital filter provides some filtering for lower frequencies. The Nyquist frequency will be at half of the modulator rate, which will be much higher than the filter bandwidth. The BAV199 should provide some modest protection. If you do use this, the main issue will be the leakage current from these diodes. I've seen the BAV199 and the BAS70 devices used for this purpose. Just remember that this leakage current will get pulled through the series filter resistance.


    Joseph Wu
  • 0 in reply to Joseph Wu
    Prodigy 175 points

    Joseph

    You have a good point about the ripple.  I chose the TPS60403 because it runs at 250kHz and figured the ripple would be beyond the sampling frequency.  I think I might add a 10 ohm in series with each ADS1248.  The voltage drop is tiny and combined with the 1uf should provide a good deal of filtering of the 250kHz ripple.  I agree that the 10uF on the thermocouple inputs is much larger than is needed.  My customer used it in a previous design so I did not want to remove it.


    Thanks again for all your help.
    Tom

  • 0 in reply to Tom Russell
    TI__Guru* 97585 points
    Tom,


    One other possibility would be to use the TPS60403 on the +5V supply to generate a -5V supply and use an LDO to drop the supply to -2.5V. That should remove most of the ripple, and make a clean -2.5V supply. I don't have a particular LDO to recommend, but I'm sure there's a simple solution.

    Let me know if you have any further questions.


    Joseph Wu