AD8293G160 EMC Considerations

We use the AD8293G160 to amplify a strain gauge that measures water pressure in a household or commercial environment.  It's a product that has been out there for a while now, but we've had a number of suspected EMC issues.  The effect of these failures is that the amplifier output freezes at a value less than 1V (Vcc=5V, Ref=1V), usually somewhere from 0.4 to 0.8V.  There are no parametric indications that the part has failed (diode measurements on all pins to all rails) when compared to a known good.

The product passes the standard panel of IEC 61000-4-* tests.  We even went back to repeat surge testing up to 6kV as our primary suspect was nearby lightning strikes.  The product will fail (when powered) with an ESD hit > 4kV.  Our feeling is that the problem is speed of the pulse more than energy.

The datasheet is very light on ESD protection.  What is the ESD spec on this part?

What would be the best method of supplemental input protection?  I currently have a 0.1uF X7R cap differentially on the inputs, the bridge impedance is about 10k/leg.  Bandwidth is not an issue for me, but simpler the better - I have very little board space available.

Any help or advice would be appreciated!

  • 0
    •  Analog Employees 
    on Apr 2, 2013 7:23 PM

    Hello BHallen,

    In the protection you describe, you don't mention anything that would limit the common-mode voltage seen by the part during such transients. How are you protecting in that case? Most system transients have a way for showing up as a common-mode signal, as percieved by the amplifier. Also, is the bridge far away from the amplifier?. In addition, although they are the usual suspects, not all the transients show up at the inputs (for example, could the transient come in through the supplies? do you have provisions against rail inversion?). It is very hard to predict how transients will flow through your system (plus due to the installation requirements, this may be an isolated system and transients can flow through power supplies, ground, etc). If you feel comfortable sharing a schematic or block diagram, that would be helpful.

    We don't specify ESD for this part, but I can tell you that we do perform tests on every part we release to verify its robustness against damage and latchup conditions. However, there's a thing or two that I need to mention regarding ESD...

    - The ESD ratings for ICs are generally only valid for the IC only, bare inputs, etc, not in a system like most customers will have. The good news is that in a system, you can expect these to improve, as you will have additional capacitances, resistors, etc.

    - The IEC standards are written as they pertain to the system, not the ICs. On top of that, ESD conditions are based on the human body model, and may not be sufficient for harsh conditions such as the large transients you mentioned. The CE compliance guidelines also require surge immunity testing, which is meant to simulate the type of surges generated by lightning. But even that may not cover your actual operating conditions.

    In my opinion, even though ESD ratings provide good information regarding the robustness of an IC, you cannot rely on ESD ratings alone (as provided by any IC manufacturer). The reason is that nobody but the system designer can control the end use case (and sometimes is really down to the end user), and the root cause may not even ESD. Plus, lighting strikes are well known for their destructive characteristics.

    If you wish, we can further discuss your protection scheme if you can share more details.



  • Gustavo,

    Thanks for the reply.

    One the first things I was going to try was adding some common-mode protection once I get some units in hand.  I should have covered some of the low hangers in my original post.

    The system run from an offline buck power supply, so it is not isolated.  The power supply is well filtered with common mode and differential chokes, as well as X-Caps and a MOV.  I use 5V from that to power the amplifier, a micro, and a 2 segment display.  The sensor is mounted in a brass fitting about 3 inches away and connected by a ribbon cable.  The back side of the sensor is potted for moisture resistance and electrical isolation.  The sensor substrate, by calculation, has about 5.5kV of dielectric strength.

    I say ESD a lot only because it's the only method I've found to reproduce the failure event.  Having gone through extended surge testing, it seems that the dv/dt of an ESD pulse it what's required to cause the amplifier to fail.  Even so, I would think it's only coupling through the sensor as the sensor has held off a 6kV surge.  Do you think a 100-200V pulse (coupled down from 4kV ESD pulse) could damage the amplifier?

    Additional EFT testing is also on my short list of tests to revisit as it has a higher dv/dt than surge, although I though EFTs were meant to reproduce coupling from nearby switching equipment when they're starting up.

    Overall, I'm trying to probe for recommended protection techniques for this part so I can work it into my test plan.



  • 0
    •  Analog Employees 
    on Apr 2, 2013 11:02 PM

    Hello Brandon,

    You are correct, In my previous post I really meant surge immunity testing (IEC 61000-4-5), not EFT. I have corrected this to avoid further confusion.

    I am a little puzzled by your description of the system. If you are using an offline supply, wouldn't that require to be isolated for safety? Unless I imagine a system that's different from what you have. Or maybe that's where the isolation resides. My point in mentioning the isolation was that the surge does not necessarily need to return through your common to reach earth ground.

    As you mentioned before, the question is how much energy is in your 200V pulse. 200V applied directly to the inputs of the amplifier will definitively damage it, but the reality is that the pulse will get clamped to the supply through the internal ESD protection and then it is up to the supply to keep that voltage below the max absolute rating for the supply. If it is low-energy, decoupling capacitors may be enough; higher energies may require additional clamps or additional protection devices. In addition we sometimes tell customers, as a general guideline, that the ESD protection can take up to 5mA (continuous) at room temp. Under pulsing conditions, this current may be much higher, as long as it is a short impulse, but it is hard to quantify.

    Many protection schemes are possible. A simple and inexpensive way is to use a TVS and diode clamps on each input. I am making this suggestion, but I don't have enough information to be sure if it would suit your needs:

    Some of these components may already be in your board. I would add something like a zener across the supplies to help with the clamping (if not already there; maybe that's what the MOV is doing).  This may be overkill for you, and you may not have space for all these components, but notice that I omitted values, because I really don't know much about the surge. Do keep in mind leakages introduced by some of these components, as they tend to vary a lot with temperature. These leakages can degrade the accuracy of your system. There are other options such as GDTs if you cannot find a suitable TVS (and some of them are very compact). I know a few companies like Bourns and Littelfuse have extensive resources on protection, you may want to check their selection and recommendations.

    You may be right in that the speed of the transient could be the cause. I have ran into similar situations in the past. The key is to build a protection circuit that can turn on faster than the surge, to keep the amplifier happy.



  • Thanks for all your help Gustavo.  I don't have TVS diodes in the circuit because of leakage concerns as the circuit needs to operate over a wide temperature range.


  • 0
    •  Analog Employees 
    on Apr 4, 2013 9:31 PM

    Hi Brandon,

    If your source resistance is large enough that leakage is a problem, you could consider gas discharge tubes. They are slower than TVS, but if your problem is surge, not ESD, they are likely to help. In my previous schematic, replace the TVS with GDTs without any resistors in front. Then, use low-leakage diodes such as BAV199 for D5 and D2. Diode-connected bipolar transistors also make good low-leakage diodes.