• State TI Thinks Resolved
  • Locked Locked
  • Replies 12 replies
  • Answers 1 answer
  • Subscribers 48 subscribers
  • Views 1559 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.

LPV802: LPV802Electrochemical gas sensor RS electromagnetic interference

Deniszmm
Prodigy 140 points

Part Number: LPV802

We design electrochemical gas sensor circuit using LPV802 amplifier with reference to "Comparing EMI Performance of LPV802 with Other Devices in a Gas Sensor Application"，But

we failed to solve the problem of electromagnetic interferenceWalkie-talkie （10V/m, 80MHz~1GHz）.Give me some advice，Thanks.

Comparing EMI Performance of LPV802 with Other Devices in a Gas Sensor Application.pdf

  • 0
    TI__Guru**** 174576 points

    Hello Deniszmm,

    We will give you a reply Monday US time.

  • 0
    Guru 140200 points

    Hi Deniszmm,

    the best cure against electromagnetic interference (EMI) like walkie-talkie radiation e.g. is electromagnetic shielding. Put the sensitive circuit together with the sensor inside a metallic enclosure and connect it to signal ground. Sometimes even a bit alufoil or metalized paper or carton is enough to provide sufficient shielding. Or use the solid ground plane of your printed circuit board as part of the metallic enclosure. (Of course, drill some tiny holes into the metallic enclosure close to the sensor, so that the gas to be detected can reach the sensor. Best put the sensor opening directly beneath these holes.)

    Sometimes hybrid grounding of the metallic enclosure instead of direct connection to signal ground offers advantages. Hybrid grounding means to connect the shield to signal ground by the help of a HF cap (10n...100n). This can be helpful if the shield has to be connected to protection earth and you don't want to have a direct connection of signal ground to protection earth. In such a case use a HF cap with a 1...2kV voltage rating.

    Kai

  • 0 in reply to kai klaas69
    TI__Guru**** 174576 points

    Deniszmm,

    In your design did you implement the three 15pF caps on each amplifier? Is your board design similar to the board in the application note (2 layer design)?

  • 0 in reply to Ron Michallick
    Prodigy 140 points

    Hi  Ron Michallick

      There are three 15pF caps on each amplifier in my design.  My board is 4 layer design.

    Regards,
    Deniszmm

  • 0 in reply to kai klaas69
    Prodigy 140 points

    Hi 

    Thank you for your reply.

    Regards,
    Deniszmm

  • 0 in reply to Deniszmm
    Guru 140200 points

    Hi Deniszmm,

    I don't think that these 15pF caps are very effective for walkie-talkie radiation at 27MHz. May be they are for cellphone radiation in the GHz range. More, EMI can also enter the OPAmp via the output pin. So, as increasing the caps or adding caps to the outputs of OPAmps is no good idea, because they will erode the phase margin and cause instability, I would again recommend the add of a Faraday shield.

    Sometimes the insertion of UKW chokes or ferrite beads at various points of the circuit can help. But this is a time consuming trial and error procedure and can also end in instability.

    Kai

  • 0 in reply to kai klaas69
    TI__Guru**** 174576 points

    Deniszmm,

    Is your walkie-talkie running at 27MHz?  The 15pF won't be very effective at that low frequency. How close was the walkie-talkie and was it transmitting at the time?

    You have a four layer board so traces can be (and already may be) in the center two layers with low impedance ground or power covering (shielding) the traces. Of course the traces have to surface to connect to the passives, opamp and the sensor itself.

    As last resorts, the software (if any) could ignore temporary disturbances. A second identical circuit could be built without the sensor as a control signal that should be clean unless EMI was intense; thus making the sensor data suspect.

     

  • 0 in reply to Deniszmm
    TI__Guru 68676 points

    Hello Deniszmm,

    I wrote that appnote. Please keep in mind that EMI "hardening" is not the same as EMI "proof". Nothing is really EMI "proof"...there is a limit.

    When doing these tests, we found that the sensor itself was susceptible. The amp circuit was fairly robust with the sensor removed.  As Kai and Ron mentioned, the sensor itself needs to be shielded with a Faraday cage to RF ground. Wrap some copper foil tape (or aluminum foil) around the sensor and ground it at several points. The leads/traces to the sensor will be the most susceptible.

    This is important because it may take *minutes* for the sensor to recover if it is hit hard by an overload (we had to wait almost 30 minutes for one of the sensors to recover after an overload...not funny at 200$/hr EMI chamber time!). That is when we learned to sweep *down* from GHz to MHz to get the most points and stop when the errors got too big.

    Ferrite beads and small capacitors should be added where the cables enter the board.  Any large area conductors are possible antennas. Be sure to "pepper" ground vias for a solid ground plane both top and bottom. Inner layers should contain the signal paths and power traces. Avoid breaking up the ground plane, at least on one side.

    Keeping the circuit compact is the best defense to keep the trace area small. The appnote board was deliberately "spread-out" to purposely add some susceptibility. The whole circuit could have been built under the sensor.

    Holding a talkie-talkie right against the board is about the same as a 10V/m to 30V/m radiated test, which is very harsh. We used a 5W VHF/UHF walkie-talkie for "quick" bench tests before spending money on EMI chamber time. As with any RF project...YMMV..

    27MHz "CB" most likely would be conducted by the connecting cables due to the ratio of the required antenna size/wavelength and the board size.

    The point of the appnote was to show the effectiveness of the LPV's internal EMI filter, which is designed to be effective >300MHz. Below 300MHz is the duty of the external circuit protection (caps, beads, chokes) - which is why it does worse < 200MHz. You can see that the 15pF caps are helping in the 200-300MHz range.

    Below 200MHz, inductive filtering (beads/chokes) needs to be used as adding higher value pF capacitors directly on inputs and outputs can cause issues with amplifier stability (peaking - as Kai pointed out).

    Entire seminars and classes are given on EMI filtering...it is not an easy subject. Google "Bogatin EMI".

  • 0 in reply to Paul Grohe
    TI__Guru**** 174576 points

    Paul,

    Thank you for the very detailed information. Especially that part about sensor recovery time.

     

  • 0 in reply to Ron Michallick
    Guru 140200 points

    Indeed! I always enjoy reading Paul's posts. I have learned a lot from them...

    And, Paul, I enjoy your beautiful and rich English! You should write books... :-)

    Kai

  • 0 in reply to Paul Grohe
    Prodigy 140 points

    Hi Paul Grohe

    Thank you for your reply.

    Regards,
    Deniszmm

  • 0 in reply to kai klaas69
    Prodigy 140 points

    HI 

    Thank you for your reply.

    Regards,
    Deniszmm