Which way of winding the toroid ferrite ring is correct?

Following my previous question I'm trying to prepare a test to see the effect of a torodial ferrite ring. The setup will be simply as below:

simulate this circuit – Schematic created using CircuitLab

But I'm not sure which way the wires should be wound around the toroid.

Which one below is correct:

Photo 1(wires wound together):

Photo 2(separately wound):

I want to vary the sine wave of the function generator from 1MHz upto 10MegHz and observe the attenuation on scope. If the setup is fine, which way of winding if correct? Photo 1 or Photo 2?

Picture 1 shows the winding for a common mode filter. This will attenuate the sum of the current flowing on the wires. This is useful when something like a PC is sending RF along the mains lead or a USB lead. It will not attenuate differential signals.

Picture 2 shows the winding for a differential mode filter. This will attenute your signal. However, it won't attenuate it very much, as the scope load is so high. The resistive part of the load is 680k, which might as well be infinity compared to 70 ohms.

You've not defined the capacitive part of the scope load. Depending on what you mean by 'scope probe', it could be a few pFs, if it's a good 10:1 probe, it could be around 30pF if it's the basic scope input, it could be 100pF or more if you have a length of 50ohm coax connected to the scope. This is what will cause signal attenuation at high frequencies.

• If I add a 3.3u capacitor between the node A and B, would that form a LP filter for differential mode or common mode case? Because in real the load will be a capacitor(1u or 3.3u) followed by a DC-DC converter. – cm64 Dec 5 '18 at 12:37
• I understand now I think what you mean. For differential mode 3.3uF becomes 0.05 Ohm. I can add a 100 Ohm resistor for differential attenuation. But for common mode I will use Picture1. Should I use two probes to see CM attenuation and subtract the channels(hot and gnd) on scope? – cm64 Dec 5 '18 at 12:53
• To measure CM attenuation, twist A and B together, and the two terminals that go to the function generator together, and measure the attenuation of the two wires in parallel. Obviously you will need to lift the ground at B, and use a different ground path. Doing meaningful things with the grounds, understanding what you've done, what effect it has on measurements, and whether that's the right thing to do, are some of the problems with trying to make CM measurements in a single-ended (one terminal grounded) circuit. – Neil_UK Dec 5 '18 at 13:42
• I didn't get what you mean can you provide a diagram for such test?otherwise I can ask as a separate question. – cm64 Dec 5 '18 at 18:45
• And here the CM choke the wires are not wound together richtek.com/Design%20Support/Technical%20Document/AN008 Look at Figure 2 there it says common mode even the wires are seperately wound. – cm64 Dec 5 '18 at 18:56

From your schematic, it appears that you are using the toroid to form a common mode choke. While the first picture shows the correct winding direction, the pairs should not be twisted while winding around the core. Instead, they should be wound in a bifilar fashion as shown below. However, unlike this picture below from Wikipedia, do not overlap the final bifilar turns.

• Are there other type of CM ferrite that one doesnt need to wound? – cm64 Dec 5 '18 at 13:25
• There are split cores that are snapped onto the wires. This is typically equivalent to a single winding in a toroid. You may be familiar with this type of core as they are often on computer cables. – Glenn W9IQ Dec 5 '18 at 13:26
• If I add two in series, is that equivalent to two rounds? – cm64 Dec 5 '18 at 13:30
• Yes assuming the basically the same ferrite material and core dimensions. Sometimes the hole in the split core is large enough that you can make a few windings before closing up the core. You can combine the two techniques as well. – Glenn W9IQ Dec 5 '18 at 13:35
• Thanks if I use something like this media.rs-online.com/t_large/F1232486-01.jpg, is that fine if the wires are twisted(plus and ground wires as in my diagram)? – cm64 Dec 5 '18 at 13:36