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.

XTR116: Incorrect reference voltage and malfunctioning 4-20mA output/driver.

Part Number: XTR116
Other Parts Discussed in Thread: XTR115, XTR111, ISO1540, ISO1541

Hello, we created an 8 channel 4-20mA output/driver based off the XTR116. We are experiencing three problems:

1. The datasheet clearly states that the XTR116 has a VREF of 4.096 volts. The chips we received are correctly labeled as XTR116, but they are outputting 2.5 volts and not 4.096.

2. No matter what current we provide on the IIN pin, we get roughly 40mA on the current loop, it doesn't change. We have a digital-to-analog converter providing voltages/currents to the XTR116 to control this.

3. We are using the VREF off the channel 1's XTR116 for the digital-to-analog converters reference voltage. If we don't have anything hooked up to this current loop, the XTR116 no longer outputs a voltage on VREF... it becomes zero.... which causes all of the other 7 channels to fail.

Here is the circuit schematic showing the 8 channel 4-20mA drivers, the digital-to-analog converter for setting IIN, and the power supply. The idea is that any device that receives a 4-20mA signal can just hook up to 2 of the 4 terminals labeled as ISOLATED_POWER, and ISOLATED_GND. The power supply is already tied into the current loop:

Any help or advice on these 3 problems would be greatly appreciated, as we already have PCB's of these made, thanks!

  • Hi Matt,

    all the circuitry you connect to the input of XTR116 must be supplied by the XTR116 itself!!! See figure 1 of datasheet. If you want to connect something to the input of XTR116 which is powered by a different supply voltage, you must introduce a galvanic isolation as shown in figure 3 of datasheet. Only by this the XTR116 is able to generate the correct 4...20mA output current.


  • I'm not sure I am following... Are you saying I need to power the digital-to-analog converter off the VREG pin? Most examples I see has that pin disconnected and not used, so this is very confusing here.
  • Hi Matt,

    yes, everything you power on the input side of XTR116 must be supplied by the associated VREG pin. See again figure 1 and 3 and read the section "minimum scale-current" on page 6. Try to understand how A1 in figure 1 is working. Consider all the currents that are flowing.


  • Ok, I will re-design this so that the DAC is powered from the XTR116, and I will use optocouplers to get the digital I2C signal into the DAC. Is this the only change that is needed? I can still tie all of the grounds together like I already have correct?

    What about problems #1 and #3? (The XTR116 is outputting incorrect reference voltage, and the reference voltage goes to zero when nothing is connected to the current loop).
  • Hi Matt,

    no, you cannot connect all the signal grounds together! Pin 3 of XTR116 is no ground pin, but a current return pin. Again, consider how the currents are flowing in your scheme and the scheme of figure 1 of datasheet: In the scheme of figure 1 of datasheet the current into pin 2 is only flowing through R1. In your scheme this current is also flowing through R1, but then back through R2 and out of pin 3. That's totally different from the intended operation of XTR116.


  • Hi Matt,

    the first two pictures show the correct flow of currents through R1 and R2 of XTR116:

    And this picture shows the improper current flow, when pin 3 is misused as signal ground pin:


  • Hello

    Ok, I see. But what I do not understand is the following: The voltage at pin 4 is dependent of the current in the loop. At 20mA and RL of 200Ohm as example the voltage would be 4V at Pin 4, right? How can a current flow from IIN through R1 to pin 4 then when VIN is smaller than 4V?

    What when I do not connect Pin 3 and do not use VREG and VREF? Then IIN will flow only through R1 to the GND of the loop which is the same as the GND of the control circuit (=DAC).

  • Hi Kai, something I just realized... there is only 1x DAC (8 channel DAC), but there are 8x XTR116... how is the current path supposed to work for that? Originally we were talking about using the XTR116 5V regulator to power the DAC, then the DAC's output goes back to the XTR116... but I am guessing I cannot connect all of the XTR116 in parallel and then into the DAC?

  • Hi Robin,

    keep in mind that A1 always tries to keep the voltage difference between pin 2 and pin3 to zero. So, pin 2 is a true current input. And the voltage drop across R1, between pin 2 and pin 4, is "mirrored" across R2.

    When you don't connect pin 3 and do not use VREG and VREF, the driving DAC output tries to find an alternative path for the return current back to signal ground of DAC. But this again disturbs the balance between the currents running to and from the XTR116 (pin 7 and pin 4).

    The only way to make the XTR116 work properly, if you want to connect something to the input of XTR116 which is powered by a different supply voltage, is to introduce a galvanic isolation as shown in figure 3 of datasheet.


  • Hi Kai

    Ok I understand. So as Matt wrote above, even when the x channel DAC is isolated from SPI it is not possible to use multiple XTR115 in "parallel" controlled from 1 DAC, right?

    Is there another component to achieve such a circuit with 1 DAC controlling x current loops?

    Thanks for your help

  • Hi Robin,

    yes, "using multiple XTR115 in "parallel" controlled from 1 DAC is not possible", unless you provide a galvanic isolation.

    You could add a single channel DAC to each input of the XTR115 and use optos in the digital control lines.

  • Hi Kai, thanks again for the help.

    I put in the 8x DAC for 8x XTR116, am using a I2C multiplexer, and using an optocoupler for digital isolation, does this look correct?

    This DAC (MCP4725) does not have a reference pin; it uses its VDD as the reference. Because of this I am powering it directly off the VREF pin on the XTR116 since its more accurate than the VREG pin. It was not clear in the datasheet how much mA this pin could support, but it seems to be able to at least support this DAC and pullups? It sounded like as long as the DAC and pullups don't exceed 4mA then it should work, and I can leave VREG disconnected.

    p.s. This adds a lot of extra components to the PCB. Are you sure there is no way to just use a single 8 channel DAC for 8x XTR116's? You mentioned it might be possible with "galvanic isolation", but I am having a difficult time imagining how that would work?

    Thanks again for the help and advice!


  • Hello Matt,

    I believe your best option is to use a single DAC for each XTR116 and control their interfaces with optos, since this is a 4-20mA 2-wire transmitter and each DAC driver circuitry is intended to be powered by the XTR and the DAC potential is referred to the IRET pin of the XTR116 they drive; as it was suggested by Kai on the previous post. The reference accuracy is specified with 0mA to 2mA load.

    If the 4-20mA current loop application allowed the use of 3 -wire transmitters, an option is to implement voltage to current converter using the XTR111. Please refer to figure 36 of the XTR111 datasheet; where all the circuitry can be biased with respect to the current loop ground.

    Thanks and Regards,

  • Hello Luis

    With the XTR111 I can use several ICs in "parallel" to control multiple loops with one DAC, right?

    Best regards
  • Hi Robin,

    any non-loop powered, 3-wire voltage-to-current converter allows this.

  • Is that an acceptable solution though? The schematic I posted above with the 8x DACs?

  • Hello Matt

    If you want to use the XTR116 your schematic with 8 DACs is the only working solution. With a 3-wire transmitter you can do it with just one DAC but the function is a different one. Check for the difference between 2-wire and 3-wire loops.

  • Yes but what about only using the VREF (4.096v) to power the DAC and not the VREG in my new schematic?
  • Hi Matt,

    you might also want to read this:


  • Hi Kai, thanks for the article. Isn't this what I implemented above? And to re-iterate my question again, is it ok for me to power this stuff off the VREF pin instead of the VREG pin? (leaving the VREG pin disconnected). Because as I said above, the DAC doesn't have a reference input voltage.... and the VREF on the XTR116 will be more accurate than the VREG pin. I will post my schematic again:

  • Hi Matt,

    Luis already mentioned that the reference accuracy is specified with 0mA to 2mA load. So, supplying the DAC should work. But have a look at figure 5 of datasheet to see how your C24 (10µF) has to be connected!

  • Hi Robin,

    Since the XTR116 is a 2-wire transmitter,  the DAC output potential needs to be referred to the Iret pin of the specific XTR116 device that is being driven.  For this reason, you will require independent DACs driving each XTR116; and the DACs digital interface connections will need to be isolated as previously suggested by Kai.  The reference pin of the XTR116 can maintain its accuracy while sourcing up to ~2mA; so it could be used to power the DAC as long as the DAC current does not exceed 2mA.  Please review the required REF compensation capacitors on Figure 5 of the XTR116 datasheet,

    If you want to use a multiple channel DAC to drive multiple current loop transmitters, you will need to change your configuration to a 3-wire configuration, and use a 3-wire transmitter like the XTR111.  In the 3-wire configuration, the XTR111 and its driving circuitry is referred to the same ground as the load or loop ground potential. Also, 3-wire transmitters are locally powered.

    In summary, the main difference between 2-wire and 3-wire transmitters are as follow:

    2-Wire Transmitter:  Loop powered (XTR116):

      Transmitter and sensor remotely located

       Local power supply not practical

       Input circuitry floats with respect to loop supply ground

    3-Wire Transmitter: Locally powered (XTR111):

       Transmitter located close to local power supply

       Input driver circuit and Rload output are referenced to same ground.

       Three wire transmitters are locally powered, therefore the current output is not dependent on how much current is consumed by the input circuit.

    Please find attached a detailed presentation explaining the differences between 2-wire and 3-wire transmitters.


    Thank you and Regards,


  • Hello Luis

    Thank you for this nice summary and the helpful presentation.

    Regards Robin

  • Hello, I created the 2nd circuit board with the above mentioned schematic and suggestions, and it has a lot of problems and is still not working right. I could definitely use some help.

    And for reference, this is my schematic (I am actually using MCP4726, so pin A0 is actually the external reference of the MCP4726):

    The first problem I noticed is that the I2C communication is not coming across the isolation properly. This is the optocoupler I am using "TLP291-4(TP,E)". Here is what it looks like going into the optocoupler:

    And here is what it looks like coming out of the optocoupler and going into the MCP4726:

    The LEDs inside the optocoupler are powered from 5V, which goes through a 470 ohm resistor, then through the optocouplers LED. I even tried reducing the resistor down to 235 and it made zero difference. It almost seems related to the VREG of the XTR116.

    During my testing, the VREG would occasionally drop down to only 1.6V. I would have to disconnect the power supply, then reconnect it for the VREG to come back to normal (and by normal I mean 4.8V which is not quite normal, its supposed to be 5.0V). This also further reinforces that something is not quite right with the VREG voltage.

    Another problem I noticed is that this isolation only allows for one way communication. The MCP4726 would not be able to send data back. In this case, I think it might still work since I don't need to read anything from the MCP4726, I just need to set DAC values. It is a bad design and shouldn't be like this.... but it seems the I2C should still come across which it doesn't.

    And for reference, here is the isolated power that is being used in the current loops for the current transmitters:

    So I am pretty stumped on what the problems are, and how to troubleshoot this thing to get it working. Thanks again for the help.


  • Hi Matt,

    unfortunately, I cannot read the component values.

    In any case, you must not exceed the allowed maximum current of Vreg. Read the section "MINIMUM-SCALE CURRENT" of datasheet of XTR116.

    Another issue is the decoupling of Vreg pin. Follow the recommendations given in figure 5 of datasheet.

  • Hi Kai, I was careful to not exceed the allowed maximum current of Vreg. The MCP4726 draws less than 1mA while in use. The pull up resistors are 10K. Do you think these pull up resistors are drawing too much current? It looks like the XTR116 Vreg can support 25mA (this would skew the current transmitter, but technically would still work). While it recommends to stay below 3.7mA for accurate transmissions.

    I followed the datasheet for decoupling of Vreg pin, and don't see anything wrong with it? Figure 5 says that if I am putting capacitive load on the VRef pin (which I am not), then I need to do that. There is no capacitive load on the VRef pin. So I am not sure how my decoupling is incorrect here?

    And here are the component values:

    R1, R2: 470 ohms (also tried 235 ohms with no success)

    The pullup resistor array is 10K

    C9: 0.1uF

    C10: 10uF Tantalum

    R4: 20K, 0.1% tolerance

    T1: FCX690BTA

    OK1: TLP291-4(TP,E)

    Thanks again for the help.

  • Hi Matt,

    table 10 and figure 10.1 of datasheet of optocoupler TLP291-4 demonstrate that the optocoupler isn't fast enough with a 10k pull-up. You should look for a faster optocoupler.

    By the way, have you thought about using an I2C compatible digitial isolator, like the ISO1540/1 e.g.?


  • Thanks for this info. Yeah I think that will fix the I2C problems, I will give that a try.

    But what do you think about the Vreg problems though? The Vreg would occasionally drop to 1.6V like I mentioned above. And Vreg was never 5V, it was always a little lower, like 4.8V.
  • I looked up those ISO1540 chips. In quantities of 100, these things cost $3.50 each! This along would add almost $30 to the cost of the circuit board. Surely there has to be a cheaper way of accomplishing this?

    Also, do you know how my design isn't decoupled properly? I double checked the datasheet and I don't see the problem? I still can't find an explanation on why the Vreg was dropping so low on the XTR116.

  • Hi Matt,

    if Vreg is provided by a typical LDO topology, then the capacitance and ESR of output cap can play a relevant role. I would supply the MCP4726 not directly by the Vreg pin, but via a RC-low-pass filter. This will increase the effective ESR seen by the Vreg pin and by this keep the Vreg output stable. At the same time the RC-filter can decrease the inrush currents or dynamic supply current spikes eventually drawn by the MCP4726 which the Vreg pin had otherwise to source. These current spikes (if too high) could make the Vreg go out regulation.

    Just play a bit with the involved parts and try to find out why the Vreg goes down. You could use a small shunt resistor (1R or so) and directly measure the dynamic supply current drawn by the MCP4726.

    Divide and conquer!


  • Thanks again for the advice. Do you know what might be a good value resistor and capacitor to use in the RC-low-pass filter? I don't see any of this stuff mentioned in the datasheet, so its odd that its required to get this working properly.
  • Is it possible there is a malfunction in the current loop side of things that's breaking this design?
  • I found a similar problem here regarding the XTR116 Vreg not working properly, but there was no solution to this?!

    Can I please get some insight on this problem?
  • Hi Matt,

    it IS mentioned in the datasheet. See figure 5 of datasheet and read a bit between the lines. There you will find an RC low pass filter. I would start with 10R and 100nF.

    Eventually, the issue is not the XTR116 but the MCP4726 and the optos you want to connect to it. The XTR116 is not designed to have complex digital circuitry connected to its input but to allow a very autarchic little circuitry, best consisting of only a bridge and a tiny analog OPAmp, like shown in figure 1 of datasheet. This OPAmp can be bypassed with one simple 10n...100nF decoupling cap and will not draw dynamic current spikes either.

    Have you confirmed that the MCP4726 isn't drawing dynamic current spikes?


  • Hi Kai, thanks again for the response.

    Like I mentioned earlier, Figure 5 is referring to the VREF pin, not the VREG pin. Nowhere in the datasheet does it mention putting a RC low pass filter on the VREG pin.

    I went through the datasheet again on the MCP4726, it says the typical input current draw is 210uA, with a max of 400uA... about 75 times less than the maximum that the VREG pin is rated for.

    Look at Figure 3 of the datasheet... it has a full microcontroller being powered from the VREG, which is complex digital circuitry. This, along with the other application circuits with DAC's implies that the XTR116 is designed to perform for these functions doesn't it?

  • Hi Matt,

    yes, I have seen that the decoupling scheme in datasheet is originally meant for Vref and not for Vreg. But I asked you to "read between the lines" and to realize that Vreg might need a similar caution.

    Even if the datasheet of MCP4726 specifies a supply current of no more than 400µA, this might only be an average value, eventually even with the MCP4726 in idle mode without having the digital bus running. It does not say anything about the occurence of short lasting dynamic current spikes which can be orders of magnitudes higher. Such current spikes are not unusual for digital chips. Especially in CMOS circuits when the input signal slews from rail to rail the internal PMOS and NMOS stages are partially turned-on at the same time causing a "shot through" current from the rail to ground.

    A RC low pass filter can help to handle these current spikes. But this does not mean, that the datasheet must recommend the use of a RC low pass filter for all other applications.

    Of course can a microcontroller circuit be connected to the XTR116. But it must be a circuit which does not draw much supply current. Have a look at this link where all this is dicussed:

    In the schemes which are shown there you will not find a huge decoupling cap at the Vreg pin. Read this in such a way that only a tiny decoupling cap should be used at the Vreg pin and that only micropower digital circuitry should be connected to the XTR116, which does not draw nasty short lasting dynamic current spikes.

    Also note that Collin is recommending the use of a digital isolator, if externally powered digital circuitry shall be connected to the XTR116.


  • Hi Kai, thanks again for the info. I updated the design using the ISO1541, and I incorporated a 10ohm resistor for part of the RC low-pass (1/10th watt, I think this is ok?). The VCC2 of the ISO1541 will be right next to C9 and C10. Similarly VDD of MCP4725 will be close to C9 and C10 as well, so I think this decoupling should work? I left the 10K pullup resistor array to reduce current.

    Does this seem like a good implementation?

  • Hi Matt,

    looks good. To reduce the risk of a too high inrush current during power-on giving the XTR116 the feel of a short circuit condition at the Vreg pin, I would think about decreasing C10 a bit. You could have a 100nF cap directly at the VCC2 pin of ISO1541 and another 100nF cap directly at the VDD pin of MCP4725.

    Important is that the total supply current of ISO1541 and MCP4725 never exceeds 3.7mA.