Common Mode Transformer Effect

Discussion in 'General Electronics Chat' started by seanstevens, Apr 22, 2015.

  1. seanstevens

    Thread Starter Member

    Sep 22, 2009
    Hello all,

    Wondered if someone could help me with this issue; please see attached PDF file containing part of my simple circuit (part of an LED flasher). the question I have is regarding the effect of common mode transformer I use in the circuit.
    See circuit (section of an LED flasher), P1 feeds a DC supply to the circuit, S1 or S2 when connected to the ‘-‘ side should change the flash mode.

    D2, D3 and R4 and R5 form the input conditioning and protection for the flash pattern change circuit which eventually feed a couple of pins on a PIC.

    Now, when T1 Common Mode transformer is in circuit which separates the nets PGND from VSS (as shown in the attached circuit) the change flash pattern inputs (S1 or S2) don’t work, they seem very noisy causing random flash patterns. It is obvious that this is the effect of T1 CM transformer, as it practically separates the grounds (PGND – VSS). If T1 is by passed, then it all works as normal.
    I have a rough idea of how common mode transformer work, I use it to block noise from the right hand side of the circuit getting in to power supply (conducted EMC) however, using T1 renders the flash patter input useless.
    My question is; is there an easy way around this problem, obviously the grounds or ‘-‘ (PGND and VSS) are not identical once passed through T1.
    Some may say connect S1 or S2 to VSS rather than PGND but logistically that is not possible as the flash patters will be remotely selected and the VSS cannot be taken to the remote panel to be used that way. The normal way of operation is if ‘-‘ is connected to S1 or S2 then the flash pattern should change, when disconnected, it will go back to the original pattern.

    Any help would be appreciated.

  2. DickCappels


    Aug 21, 2008
  3. seanstevens

    Thread Starter Member

    Sep 22, 2009
    [Since your question does not involve AC mains - as clearly stated in your post and shown on the schematic - here is an answer.

    First, there is no input or output to your filter section. It is symmetrical, and presents the same impedance to signals going through it in either direction.

    Second, don't think of T1 as a transformer. In this application, it is not. It is a common mode choke, and exhibits a large common mode inductance that is intended to act as the series element in a balanced lowpass filter. As such, it presents an impedance to both of the input power lines, and this can cause problems if something (like a control signal) on each side of the filter needs to be at the same potential. Everything is fine at DC, but noise in the power return can appear as a additional voltage on signal lines.

    Since you are not using a switching regulator or transformer, two common sources of common mode noise, are you sure you need T1 in the first place? I would replace T1 wit a single inductor in the + power input line only, and tie your GND (U3 pin 4, etc.) directly to P1 pins 3 and 4 (- power), and see if that gets you the right combination of filtering and reference potential.


    Above, AK post on my question...

    Thank you very much for your input. Yes, as you can see I have encountered problems...
    Initially the design was with the CM choke (T1) it is very effective, I have an LED driver which is really a switch mode PSU up stream from the chip you see in the (cut) circuit and it generates a lot of noise. Currently I am using your suggested option of a an inductor and two caps as a standard low pass filter which also works well not as good as T1 however, when it comes to running some surge test and burst transients T1 in combination with other protective components such as varistor seem to save the driver board but my standard filter configuration is not always as successful.
    So T1 is more effective as filter and as a secondary function due to its high inductance slows down surge and burst but on the down side it interferes with the flash patterns. On the other hand my simple filter isnt as successful in holding surge and burst but doesnt give me any problems with patterns.

    I have had this issue before and now it has become a challenge and I just wanted to see if there are any solutions to this.