2
\$\begingroup\$

I am working on a project using two audio transformers for two separate circuits mounted very close to each other on a pcb as shown in the image. With the transformers as mounted, there is a bit of inductive crosstalk, as would be expected.

enter image description here

My question is: ignoring shielding for the moment, is there a particular orientation of the transformers (both cores vertically parallel, each perpendicular to each other) that would improve this? I am able to separate them to some degree, but not more than a few centimeters max.

To get an idea of the field strengths, with one signal active I placed 2 ungrounded scope probes' tips at different distances from the active transformer as a test. I get a higher induced voltage on the coil side (above and below in pic) than on the core side (left and right in pic). The iron core seems to keep the flux from straying as far outside in that orientation. However, I was told that it is was common in old radios to mount power and audio transformers with cores perpendicular to reduce 50/60hz cross-talk, so curious to get some other opinions.

EDIT: A schematic with a better description may help: With music on IN1, and relay on T2 closed, I can hear quiet music at out of T2. When shorting IN2 primary, it reduces the music volume, but still faintly there. Correct me if I'm wrong, but I'm generally getting the impression that audio transformer circuits should not be left open when un-used, unfortunately I haven't been able to find much info about this.

enter image description here

\$\endgroup\$
  • 2
    \$\begingroup\$ I'll tell you something - I doubt that there'll be any inductive cross-talk unless they are the worst audio transformers I've ever heard of. Have you looked at the coil coupling percentage (or leakage inductance percentage) of half decent audio transformers? BTW your test that supposedly suggests inductive cross talk is basically meaningless. \$\endgroup\$ – Andy aka Sep 3 '17 at 19:46
  • 3
    \$\begingroup\$ Use a loop (H) probe to test for crosstalk in this case -- what you were seeing with your test is likely a stray E field \$\endgroup\$ – ThreePhaseEel Sep 3 '17 at 19:59
  • 3
    \$\begingroup\$ Your ungrounded scope probe only captures Efields. To monitor Hfields, make a 1" square or 1cm square loop, and run that to the scope probe with tightly-twisted wire pairs. Move the square loop around to explore the magnetic fields. \$\endgroup\$ – analogsystemsrf Sep 3 '17 at 20:04
  • 2
    \$\begingroup\$ Besides shielding, it might be a good idea to add a short circuit turn, to kill stray field: it may be 1 turn of copper tape around each transformer. \$\endgroup\$ – andrea Sep 3 '17 at 21:03
  • 2
    \$\begingroup\$ @Andy: Not a hypothetical. As I stated in the question, there is crosstalk (signal being picked up in one line with music coming in on the other line.) This happens with the secondary of the music line open and with the primary of the non-music line shorted, but is much worse with non-music primary open also. A thick layer of aluminum foil between the transformers attentuates the higher frequencies. The scope idea was just to try to determine how re-orienting transformers might help. I'm clearly still learning, a "WHY" something is meaningless as others have tactfully done would be helpful \$\endgroup\$ – User7251 Sep 4 '17 at 0:54
2
\$\begingroup\$

As per Brian Drummond's suggestion, I mounted one transformer on wire and played around with the orientation. I found that at about 1" separation between the transformers, I can crank up the volume all the way with no audible induced voltage from the other. The more separation, the quieter the induced sound. Rotating them parallel/perpendicular to each other didn't seem to make a noticeable difference.

\$\endgroup\$
  • \$\begingroup\$ That's a good result. Don't forget, you can accept your own answer to mark it as correct and highlight the solution to others. \$\endgroup\$ – Transistor Oct 21 '17 at 8:59
1
\$\begingroup\$

The suggestion that these are 'the worst audio transformers' means that it is possible that there is no conductive outer wrap around the transformers. For audio, to keep signal leakage low, it is common to see a copper foil tape wrapped around the outside of the core (so it encloses both the windings and the steel core outside the windings), with the copper soldered to form a conductive belt.

Instead of just foil between the cores, a conducting wrap around one or both might be an improvement. example of shield wrap on transformers

\$\endgroup\$
  • \$\begingroup\$ The transformers used actually have decent frequency response for the price, but you're right, there is no shielding at all, neither against stray electric fields nor magnetic fields. Adding one would no doubt improve the cross-talk. Re-laying out the circuit so that the unused input primaries were shorted helped reduce cross-talk as well. \$\endgroup\$ – User7251 Oct 22 '17 at 19:54
0
\$\begingroup\$

Only the flux that cuts a conductor at an angle (any angle) will induce an EMF in it; with 90 degrees being the angle which gives the max magnitude of EMF. Therefore if you want to have zero have the flux parallel to the conductor. Placing the two transformers with one core at a 90 degree angle to the other will reduce inductive coupling. As far as capacitive coupling distance is the answer. This is why you see on audio amps that have power trans and audio trans on the same mounting plane; with their cores at a 90 to each other. I know this is simple compared to all the other "fancy" answers; but sometimes going back to the basics is overlooked. Your original question is about orientation; so here is the answer. Yes all the other things said are true, but get the obvious out of the way before you approach the esoteric. Not knocking the "fancy" stuff just trying to help. If you really want to get exact, inject a signal of the median frequency of your device into one transformer and then your oscope probes connected to the other and play with the angles to obtain minimum inductive coupling; the angle might be slightly off 90 but close to it.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.