Post Go back to editing

(AD8418A)Vout ≠ Vref

Hi !

I have a question about AD8418A.

This is my test circuit.

I measured the Output voltage but it is not Vref level.

=> I don't have AD8418A so I measured AD8418.

Why this happens?

Our customer said that  this won't happen when they are using AD8206

but I measured AD8206 at my Lab and same thing happened.

But the output voltage offset is smaller than AD8418.

I think that maybe the +IN pin leakage current is big 

Can you explain that why this happens?

Best regards

Kawa

  • HI !

    Can anybody help about this?

    Our customer is using this device and already mass-production.

    So if somebody can help about this it would help our customer and us.

    Best regards.

    Kawa

  • Hi Kris

    Thank you for your kindly support!!

    Your support help us very much.

    >Can you try shorting the inputs together and connecting to ground?

    I got it.

    I will try that.

    >Is the output accurate when you measure with a reference voltage?

    Sorry I couldn't understand what you said.

    Do you mean did i measured the output voltage with Voltmeter?

    For that , the answer is yes.

    Also I got some AD8418A samples from our customer which was failed at the mass-production line.

    So I will try the same thing.

    >Why does the design call for such a large shunt?

    At the real application, the will use 5mOhms.

    But at the mass-production test , they are using 220Ohms to limit the current.

     

    Best regards.

    Kawa

  • Hello Kawa, 

    Yes, the AD8418 can have high bias currents at high common-mode voltages (See Figure 8 on the datasheet). Using a shunt resistor that's too large would cause a high offset when that bias current goes through the 4.7k resistor.  

    I have been trying to go through the results and your setup, and the measurements with the 4.7k resistor do not really make sense to me. The output is much too high, even when the possible errors are accounted, i.e. bias current, offset voltage, and the error from the reference voltage. With your setup, you would still be operating close to 0Vcm, and Figure 8 from the datasheet shows you that the bias current will be very small at this point. It is usually around 1uA, based on experience.

    The second set of results makes more sense -- you are still seeing the reference voltage coming through correctly, albeit with some error. Can you try shorting the inputs together and connecting to ground? Is the output accurate when you measure with a reference voltage? I just want to make sure that the device is working correctly in a known configuration. 

    I will check the AD8206 bias current and see how it compares to the AD8418, and see whether we can conclude anything from there. I must reiterate though, that in the applications the AD8418 is designed for, large shunts are not really used due to the error they can contribute. Why does the design call for such a large shunt?

    Thanks, 

    Kris

  • Hi Kris!

    Thank you for your support everyday.

    The result I attached at above, the resister was 47k.

    I'm very sorry.

    I measured the samples(AD8418A) which my customer gave me.

    I think the shunt resistor which they are using for the mass-production line is too large.

    But they can not change the resistor at this timing.

    So they want to solve this problem with other way.

    Q1)

    I think this happens because of the +IN pins leakage to outside.

    How do you think about this?

    Q2)

    Which current pass is it caused by?

    Can you explain with some inside circuit?

    Q3)

    Can you tell the dispersion for the +IN & -IN pins leakage current?

    Or is there any specification for this current?

    Best regards.

    Kawa

  • Hi Kris

    Thank you for your support everyday.

    I talked about this with ADKK and we think the shunt resistor is too big as you said.

    So I told the same thing to my customer and maybe I can close this issue.

    If I get any additional question for this maybe I will ask here but 

    anyway I will let you know that can I close this session or not.

    I will report you very soon!

    Best regards.

    Kawa

  • Hi Kris 

    Thank you for you kind support everyday!

    Your data help us very much.

    So the data you attached is for AD8418? Or AD8418A?

    I'm asking this because you gave me the bias current VS Vcm graph and the result of Vcm was

    Vcm[V]    Ibias+[mA]     Ibias-[mA]
    0               -6.35E-03       -1.63E-03

    It look that Ibias+ is bigger than the result which you attached today.

    Best regards

    Kawa

  • Hi Kawa, 

    Yes, I agree. The shunt resistor is too big. If, as you say, the resistor is 47kohms, with a 1.5uA bias current at the inputs (which you would very likely get), your numbers now look valid. 

    I just wanted to provide some additional information on your questions. 

    The current you suspect is not leakage, but is inherent to the part. Figure 8 gives you an idea of the bias current over common-mode voltage. 

    The inputs of the AD8418 are tied to an internal floating power supply, which powers the device at high common mode voltages. This is the reason the bias current increases largely as the common-mode voltage increases. At low common-mode voltages, on the other hand, there is still a leakage current from the floating power supply, and this is what you see as bias current. 

    In your configuration, you would be operating roughly at 0V common-mode. Fortunately, we do have data available to give you an idea of the spread of the bias current at that common-mode voltage. See below:

     

     

    Based on this data, you can see that the Vin+ pin can have a little over 4uA worst case, while at the Vin- pin, you would roughly have 1uA. That's 3uA going through a 47kohm resistor worst case, and you get 2.8V at the outputs, so this definitely will not work. 

    Let me know if you have additional questions. 

    Thanks, 

    Kris

  • Hi Krisf

    Thank you for your reply!

    I understood.

    Anyway , the value of bias current is not specification so I should explain to my customer that 

    don't use the big value resistor .

    Best regards.

    Kawa

  • Hi Kawa, 

    The data attached is for AD8418 only. The AD8418A will have slightly higher bias current.

    I think there is not enough resolution on the graph of Figure 8 to get an accurate value for bias current. The histograms will be more reliable. 

    Regards, 

    Kris

  • Hi Kris.


    Thank you for your support every day.

    Our customer are using AD8206 at the same application which is already mass-production past.

    So if you have AD8206's Vcom VS Ibias+ Ibias- can you share that?

    You send me AD8418A Vcom VS Ibias + Ibias- by e-mail few month ago so if it is difficult to up to this session, e-mail is OK.

    I tried to measure this at my Lab but my equipment is not good enough so I failed to measure uA level.

    Best regards.

    Kawa