• State TI Thinks Resolved
  • Locked Locked
  • Replies 25 replies
  • Answers 3 answers
  • Subscribers 17 subscribers
  • Views 9135 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.

Analog Switch recommendation for 4 Analog Input for Industrial Outputs?

Noam Weidenfeld1
Intellectual 780 points
Other Parts Discussed in Thread: MUX36D04, MUX36S08, ADS1248, INA826, INA159, MUX509, ADS8664, TVS3300

Hi,

I need to multiplex 4 Analog Input Module for Industrial Outputs to a single A-to-D.

Requirements:

1. Analog Input Module for Industrial Outputs: +/- 10V or 4-20 m"a.

2. Since it is connected to outdoor sensor,with line protection elements. Otherwise external protective elements?

3. Isolation between inputs?

4. Isolation between analog line and digital control lines?

I'll appreciate your recommendations.

Thanks

Noma

  • 0
    Intellectual 780 points

    Hi,

    Adding the application schematic.

    Noam

    analog input.pdf

  • 0 in reply to Noam Weidenfeld1
    TI__Expert 3900 points
    Hi Noma,

    Are you looking for a 4:1 mux for analog input modules?

    We have some parts such as the MUX36S08 which is a high voltage, low leakage, 8:1 single ended mux that may be a little more than you need but could still be beneficial. We also have the MUX36D04 which is a 4:1 mux with differential inputs.

    Are these parts helpful for your application? I didn't see a part number included in post but looking at your schematic you are using the ADS1248 which is a 24-bit ADC with an internal mux. Are you looking for advice on the ADS1248?
  • 0 in reply to Dakotah
    Intellectual 780 points
    Dakotah Hi,

    I need to measure 4 Differential Analog input. I know that the ADS1248 has an internal MUX but I need to add a pre-sampling stage that include the INA826 (High impedance buffer) and INA159 (Set the appropriate voltage)

    So instead of using 4 pre-sampling channels I prefer to use a single channel and multiplex the inputs. (Any way the ADS1248 can sample one channel at a time)

    So to understand if the I can use the MUX I need to understand the following:

    1. Does it include protective elements or I need to add it externally? Any recommendation?

    2. Since my device is powered by a battery, what will happen if I'll disconnect the power rails +/- 15V? Can it damage the mux if the analog inputs are still connected?

    Thanks

    Noam
  • 0 in reply to Noam Weidenfeld1
    Intellectual 780 points
    related to question 2.

    What will be the total current if the EN signal is LOW i.e. all switches are in OFF state?
  • 0 in reply to Noam Weidenfeld1
    TI__Expert 3900 points

    Noam,

    Thanks for clarifying your design intent, I wanted to make sure you were looking for an additional external MUX.

    If you are looking to multiplex 4 differential analog inputs the MUX36D04 seems like a good device. This part is used in industrial PLC settings and is even used in some TI Designs. The TI Design TIPD151 uses the MUX36D04 to multiplex 4 differential inputs to a single ADC and includes a complete reference guide, test results, and some relevant schematics. Additionally, the TI Design TIDA-01333 uses the MUX36S08 which is the single ended version of the MUX36D04 but has a reference guide that goes in depth into protection circuitry examples for the design.

    1. Protective elements should be added to satisfy you design needs but examples are provided in the above TI Design links.

    2.In the case of power being disconnected from you mux some of our devices have powered-off protection which states the current into the device when VCC=0V. Unfortunately the MUX36 device does not have this feature, however this can sometimes be worked around by adjusting power sequencing or other isolation devices. I have attached a link to another E2E thread that talks about this for the MUX36 devices.

    Related to question 2. The total current consumption of the device if EN is tied low (i.e. all switches off) should be the summation of the ICC current and input off-leakage current for the channels.

    The first figure shows the Power Supply currents for the Vdd=15V and Vss = -15V case.

    while the second figure shows the off-leakage currents for the Vdd=15V and Vss = -15V case.

  • 0 in reply to Dakotah
    Intellectual 780 points

    The first table shows the power consumption from the power supply when Ven = 3.3V. Does it change when Ven = 0?

    Noam

  • 0 in reply to Noam Weidenfeld1
    TI__Expert 3900 points

    Noam,

    The power consumption should not change based on Ven = 0 or Ven=3.3V for this device.

    Additionally, I suggested the MUX36D04 devices to you since you mentioned being powered by a battery. We also have the MUX50X family besides the MUX36XXX family I have already mentioned. The MUX509 has very similar performance to the MUX36XXX family with the biggest differentiator being leakage currents.

    We have a number of other switches that could potentially be used such as our CD40XX family of devices that aren't as high-performance but may be suitable for your needs.

    Let me know if you would like some more recommendations.

  • 0 in reply to Dakotah
    Intellectual 780 points
    Dakotah,

    The thing is that my system is in sleep time most of the time and I need to keep its total power less then 90 uWatt.

    So one option is to shut down all the power rails when the system is in sleep mode. So I need to know if the MUX can get damaged if it has no power but the analog inputs are connected?

    Second option is to configure the chip in a way that its power consumption will be minimal. Check for example Analog Devices chip ADG1208_1209. If its digital input are all set to 0 then its power consumption is less then 1uA.

    Noam
  • 0 in reply to Noam Weidenfeld1
    TI__Expert 3900 points
    Noam,

    Thanks for letting me know more about your system, those two option really help understand your design constraints.

    Unfortunately are only devices that can support +-10V are our MPC50X, CD405X, MUX50X, and MUX36XXX families. These devices cannot support this powered off protection feature needed by your system or have too high of a power consumption. This is great feedback and will help in developing future parts.

    However, you initially mentioned multiplexing the 4 inputs to a single ADC. If this ADC hasn't been chosen yet we may have a potential "all-in-one" solution. A device like the ADS8664 may work in your application. It is a 12-bit ADC that has an internal mux. It shows a power-down supply current of 3uA at a supply voltage of 5V.

    Is there potential to use a system level solution like this? It may not be that specific ADC but if you don't NEED and external mux we may be able to help find a solution that fits your design.
  • 0 in reply to Dakotah
    Intellectual 780 points

    Dakotah,

    I am already using the ASD1248, I have some in stock from previous design.

    The thing is that I need a very high impedance input for the analog channels (+/- 10V or 4-20 m"a), therefore I have a pre-sampling channel that includes the INA826 and the INA159. attached is the schematic.

    So even if I can protect the Analog Mux with external chips it is OK, but I have to keep a low leakage current (High impedance).

    Noam

    analog input ii.pdf

  • 0 in reply to Noam Weidenfeld1
    TI__Expert 3900 points
    My apologies Noam, I was looking at your schematic but I misunderstood you originally. I thought you were looking to add in another ADC in addition to your current one.

    We are working to come up with a potential solution but would like to test it in the lab first.

    I do have three questions for you that can help us come up with a solution:

    1) Does a current flowing back on your VDD rail cause potential issues? We may use an external diode to mitigate the problem.
    2) Is your sensor that provides +-10V powered from a different source than the mux?
    3) Do you currently have a low pass filter before your mux to help with a noisy environment?
  • 0 in reply to Dakotah
    Intellectual 780 points

    1. Yes, attached is an updated schematic. I added diodes on the +/- 15V and a series resistor on the analog inputs. I am considering adding a TVS on the differential pairs as well however it has a leakage current of 1 uA so I am not sure.

    2. It can be powered by an external power supply.

    3. No. I have a filter on the ADS1248 inputs.

    Noam

    Analog Input iii.pdf

  • 0 in reply to Noam Weidenfeld1
    TI__Expert 3900 points

    Thanks for the info! We will work on coming up with something and let you know! 

    I had looked into some TVS diodes before and we have some TI solutions such as the TVS3300 with a leakage of 19nA typ. but is at a standoff voltage of 33V. But they may have other solutions at your required voltage. Also this TVS diode family may have low specs. I haven't evaluated it for your case but it was saved from when I was looking through TVS diodes and I believe it has lower leakage specs. 

    Let me get some lab results and I can get back to you!

  • 0 in reply to Dakotah
    Intellectual 780 points

    Thank you,

    Waiting for you input.

    Noam

  • 0 in reply to Noam Weidenfeld1
    TI__Expert 3900 points
    Hi Noam,

    I wanted to keep you updated on the status of our tests. We have been waiting on some components but are beginning to get some results.

    Our current plan is working on a way for you to disconnect the supply pins of the MUX but continue to have signals present on you input pins using discrete components. Our solution would disconnect the mux from the +15V supply that is powered by your battery to consume less power. However, signals at the input could then turn the device back but would be powered by the source of you sensors.

    You mentioned the +-10V signals are supplied by another source but I want to confirm that are not limited to the 90uW power limit?
  • 0 in reply to Dakotah
    Intellectual 780 points

    Dakotah Hi,

    Thank yo again for your efforts.

    I don't know what are the sensors power limitation.

    The application is measuring pipes status. It uses a "Cathodic protection" so actually our costumer requests that no current will flow through our board when it is not connected.

    He actually asks for mechanical relays at the input of the board so the analog lines will be totally disconnected from the board when no measurements done.

    I though that by using chips with very low leakage power at the input we can get the same results. However I am not sure that it will work.

    So for now I will use the relays unless you have other thoughts

    Noam

  • 0 in reply to Noam Weidenfeld1
    TI__Expert 3900 points
    Hi Noam,

    Due to the requirement of no current flowing through the board when not connected I do not think what we were coming up with will satisfy that need.

    The mechanical relays on the inputs is one way to isolate the downstream like you mentioned.

    I agree with your plan on using the relays for at least the time being. Since the MUXs with your requirements do not support the powered off feature, coming up with a discrete solution that also has not current flowing from the sensors is proving very challenging. If we can come up with something then we will let you know.
  • 0 in reply to Dakotah
    Intellectual 780 points

    Dakotah Hi,

    What will be at the output of the MUX if the selected channel is floating i.e. nothing is connected?

    Noam

  • 0 in reply to Noam Weidenfeld1
    TI__Expert 3900 points
    If the analog input signal pin of your MUX is floating the channel will be in an indeterminate state. You can always attach a pull-up or pull-down resistor if needed. However, the best solution may be to simply disable the device with the enable pin.

    Make sure you never float the control signals though. These CMOS pins if left floating can lead to excessive current draws.
  • 0 in reply to Dakotah
    Intellectual 780 points
    But how can I recognize this state?

    Assuming that the sensor was disconnected from some reason, How can I know?

    Do you mean that the MUX output will be in an unknown state or will it be pull to one of the MUX powers (+15V or -15V)?

    Noam
  • 0 in reply to Noam Weidenfeld1
    TI__Expert 3900 points
    If for some reason the sensor input to your MUX became disconnected then your signal input would no longer be present.

    If your system had no detection means for this then your mux would not know the difference and still be conducting a low impedance path from input to output pin. If your input pin is then floating your output would be the same voltage present at the input pin. It would not be pulled to one of the voltage rails.

    What you can do is attach a pull-up or pull-down resistor of a high resistance (say 10Mohm to minimize leakage current) to the output pin. This way if your sensor becomes disconnected your signal would be pulled to a known state (either +15V or -15V) and the ADC/software could recognize this issue in the system.
  • 0 in reply to Dakotah
    Intellectual 780 points

    Dakotah Hi,

    Attached is a circuit according to your suggestion. Don't you think that the pull-up resistor will have some effect on the voltage measurement during normal operation?

    Note - The "-" input of the INA826 is connected to GND to measure the +/- 10V.

    Noam

    Analog In.pdf

  • 0 in reply to Noam Weidenfeld1
    TI__Expert 3900 points

    Noam,

    Will the pull up resistor have SOME effect? Yes. But does this impact your circuit, that depends on the accuracy to are trying to achieve.

    The effect of the pull up resistor is VERY minimal compared to your +10V input. By analyzing the circuit, or applying superposition from the two voltage sources, the drop across your combined series resistors of 1K S1A + 125ohm mux R + 249 ohm would be 0.0007 volts. The majority (4.999V) of the +15V supply is dropped across your 10M pullup. 

    One way to look at is thinking of the impedances from the two supplies. The +15V supply sees a very high impedance (10M) and is a weak voltage source. The +10V input signal sees a low impedance (~1375 ohm series combination) and is a very strong voltage source that will dominate the circuit. Or you can simply apply superposition across various voltages to verify for yourself.

    Hope that helps!

  • 0 in reply to Dakotah
    Intellectual 780 points

    Dakotah Hi,

    Unfortunately I am not an analog expert, so I need your help to understand the impact of the input element on the differential analog Input impedance: 

    1. Directly connected to the INA826 the input impedance is very high (~10-20 G OHM).

    2. Does adding the analog MUX have an impact on the input impedance? in what way?

    3. Does adding the pull-up resistor have an impact on the input impedance? in what way?

    Thanks

    Noam

  • 0 in reply to Noam Weidenfeld1
    TI__Expert 3900 points
    Noam,

    The analog mux has an Ron value of 125ohm, 250ohm max across temp. The mux is in series with the the 1K and other 249ohm resistor then into the INA. Your input impedance seen by the sensor is then 10G+1K+250+249 =~10G. Very little impact seen from the sensor side.

    The Pull up resistor will effect the input impedance as seen from the sensor providing +-10V. While the INA will still have 10G input impedance the sensor will now see a 10M input impedance (plus all the series resistors like last time) which is =~10M. This 10M however is still very high compared to the series resistors of 1K+250+249, hence why we can ignore them by saying ~10M. But if we do not ignore the resistors it can be modeled as your sensor feeding into a 1.5K resistor, then a 10M pull up to +15V while also going into the input of the INA. The INA input impedance of 10G is so high it essentially looks like an open circuit. The voltage at that node is changed very little by the 10M pullup because it is so much larger than 1.5K. The drop across the 10M is equal to 99.98% of the value with 100% being ideal. If 0.02% is too much error then you can try increasing the pull up value to reduce the error.