Why are common mode chokes with 90 ohm @ 100 MHz recommended for USB 2.0 Hi-Speed data lines?

Question

When one looks at common mode chokes for USB 2.0 applications choke manufacturers always recommend CMC with a common mode impedance of 90 ohms @ 100 MHz.

Example

How does this relate to

• USB 2.0 differential impedance of 90 ohms (& 45 ohm single ended impedance)
• the "Hi-Speed" data rate of 480 MBit/s (respectively a frequency of 240 MHz)
• an approximate USB 2.0 bandwidth of ~1200 MHz (5 x 240 MHz)

Why aren't they made for, say 90 ohm @ 240 MHz? Why are they trimmed for 90 ohm @ 100MHz? Couldn't chokes with values that differ from this 90 ohm @ 100 MHz be used as well? What are the determining factors here?

Sorry the question may seem a bit weird but I couldn't quite get that topic. Maybe someone can shed some light on this.

Answer 1

I'll try to shed some light here. This question is nothing but a can of worms.

1. USB specifications do not define nor approve any passive/inductive/capacitive components along the USB transamission lines. The line must maintain 90-Ohms DIFFERENTIAL impedance over all essential frequencies to be useable and have a certifiable eye diagram.
2. So-called "recommendations" to use common mode chokes and ESD-diodes (with associated capacitance) only came in cases when USB PHY driver fails to provide balanced differential signaling and/or sufficient level of ESD tolerance at chip level, or board designer failed to trace USB lines on PCB with matching lengths, with poor return grounding etc. Normal designs do not use CM chokes, unless insane requirements for ESD and EM are imposed for the end product. They are used as last resort if the product fails FCC emission test and /or requested level of ESD protection.
3. Recommendations to use CMC with 90 Ohms (at standard test frequency 100 MHz) are coming out of sheer ignorance of chip manufacturers regarding physics/mechanics of CMC. First, 90 ohms is usually specified as a parameter for common-mode signal, and bigger value is obviously better. However, the CMC must have 90-Ohm DIFFERENTIAL IMPEDANCE to meet USB specifications, and have it over the essential range of frequencies. The trouble is that it is physically impossible for a wirewound choke to have any good definite differential impedance over USB2 (or USB3 or whatever) frequency range. Below is a typical chart of common and differential impedance of a CMC:

As one can see, the differential impedance of this CMC in the range of 100-500MHz varies from 20 Ohms to 100 Ohms. This element will present a huge differential impedance mismatch in the USB transmission line, and significant signal (eye diagram) distortions will follow. Actually, the presented diagram (common mode impedance ~800 Ohm at 100 MHz) is maybe the best possible match if you have to use CMC to correct IC/board design deficiencies.

4. More caveats: To have reasonable DIFFERENTIAL impedance in a CMC over USB2 signal range, its common-mode impedance will be in 500-900 Ohms at 100 MHz. Selection of 90-ohm choke is a blatant error. Again, the reasonable impedance can be achieved only with long wires, so the choke might acquire quite significant DC impedance, which will have a negative effect on other USB signal parameters as chirp levels and disconnect thresholds. Small-size CMC (404 or 606) usually cannot meet both requirements, so only bigger chokes do the job.

Bottom line - use of CMC in high-speed transmission lines is quite tricky.

Answer 2

Commonly these kind of small coils and ferrite beads meant for data and power supply filtering all have their rated impedance measured at 100 MHz.

The fact that a part suggested for 90 ohm differential signal happens to have 90 ohms impedance at 100 MHz can be pure coincidence for example reasons, as commonly parts like these are available at various impedances.

If you click on some of the parts in the PDF, you end up on a product web page, which in a footnote explains that the part impedance is selected based on signal frequency and how much common mode attenuation you need. So it all depends what you need to achieve for a good performance, there is a balance of filtering too little or too much, if you need to pass both USB compliance tests and conducted EMI tests.

Answer 3

Common mode chokes manufacturers' write all data in the datasheets.

The most important data in this case is Z as function of frequency.

One should look at the graph Z(f) when selecting a choke.

To answer your question:

Manufacturers, to summarize the performance of their chokes, decided to use 100 MHz as a reference frequency.

Why 100 MHz?

Maybe because it's close to 30 MHz which is the frequency cross point between conducted emission tests and radiated emission tests. To be honest, I don't know the reason, it's just a guess.

Please note that it's customary to select a reference point for summarizing the performance electronic devices.