Post Go back to editing

AD8236 Level-Shift Output (pin 7) offset

Our AD8236 level-shift output at p7 has an offset relative to a stable and accurate Vref at pin 6. What conditions or issues could cause the AD8236 to (pin 7) output level shift to not match the provide REF at pin6? Could capacitive loading on the AD8236(pin7) cause this? Our AD8236 has a low pass filter on the output with 100 Ohms feeding 22uF RC LPF.

I have provided the circuit in hopes of any recommendations. A rework was done on the circuit that set R23 to 0 Ohm and removed C15. The AD5160 sits at  ~100Ohms during production and this is where we have seen PCBs drop out due to a AD8236 p7 output level shift should be 1.65V. Many PCBs pass, but 5 have dropped out (out of 100) due to pin 7 level shift starting as high as 2.5V instead of 1.65V...I have also noticed tha the starting level shift, when high or low will capacitively discharge or charge to the neighborhood of 1.8V, 1.73V etc

Could a hot-air re-work to pull C15 and change R23 have damaged the AD8236 on these particular PCBS? Since 95 have passed? R23 and C15 are right next to the AD8236 chip.

Also I thought 100 Ohms should be enough isolation from the 22uF capacitor C16...we never saw this issue until production :-)

  • Hello Joe, 

    I would like to make sure I understand what you are seeing correctly:

    With the LPF of 100ohms and 22uF connected, you are seeing 2.5V on the output. Is this correct? Does it stay at this level?

    When the 100ohm resistor is replaced with 0ohm and the capacitor is removed, you still see the 2.5V but it goes down to about 1.7 to 1.8V eventually. Is this correct? 

    Have you tried measuring the output when the AD8236 is electrically isolated from the digipot? (Remove resistor connected at R23). Also, what is the input voltage you are applying to the in amp when you measure the output?

    The absolute maximum rating for soldering temperature is 300degC for 10s, so that might be useful in determining whether the device was damaged or not. Did you only observe the incorrect output after you reworked the boards?



  • Hi Kris,

    Thanks for responding.

    The nominal level-shift offset I am experiencing on these PCBs only occurred after PCBs were sent out for re-work - then we re-testing the PCBs in production. In other words, the pre-rework circuit was passing production with AD8236 p7 giving the expected level shift of 1.65V.

    We had some customer for a new batch of PCBs, we adjusted the LPF 3dB and step response. On these new batch re-works, on the 5 or so I have in my office, we noticed this offset on what is supposed to be a 1.65V level shifted out on p7 of AD8236.  We never saw this with the nominal circuit shown in the attached schematic from my first message.

    When these re-worked PCBs are powered up (the 5 I have) they will overshoot or undershoot the 1.65V and then while still powered and measuring, they will decay or increase to the ballpark of 1.65V...some stop at 1.7, some at 1.8 and if they are powered all night, I've seen them, get to 1.66V.

    I did not remove R23, but that is a good suggestion and a good test. What I DID do, however, is set the digipot to 1K ohm to isolate AD8236p7 from  C16. I did not notice any improvement in the offset from 1.65V. The decay (or increase from initial power-on U6p7 voltage) does surely look capacitive....but since (in the original design before rework) C15 was only isolated by 499, I would hgave expected the rework board to show improvement (if capacitive drive is really the issue) when setting the digipot and isolating U6p7 from C16 by 1k...?

    I hope this clarifies..and thank you again for helping!

    Saggio Technical Services

    Joseph A Saggio, Jr | Chief Engineer

    Phone: (714) 600-3516<>

  • Hello Joe, 

    Thanks for the explanation, that clears things up a lot. 

    One thing I would check, actually, is to make sure that the components that were reworked were actually soldered on properly. I might expect some varying offsets if there are some nodes that are floating.

    If the problem really is capacitive loading, you could always check the output of the AD8236 on the scope to see if you have any oscillation on that line. However, if you've gone so far as to remove C15 in your setup and isolated C16 as well, then I doubt this is the issue. It would not hurt to check, though. 

    Again, we do have an absolute maximum spec for the soldering temperature that can be applied to the device. If the soldering temperature was applied for longer than the specified period of 10s, then there is a chance that the device could have been damaged during rework.