But what does the varistor do to suppress the EMI? For example, Im working on a design that may be serviced in the field by technicians or by average folks who are tech savvy. Its got a USB connection, and probably would have to undergo some EMI/Wireless testing before it becomes a real product. How would a varistor help me? And, say for example, I have an AC power source nearby. Is there some way I can calculate for the EMI that I will be expecting?

Why would I use a varistor in an ESD/EMI protection circuit instead of a diode or a resettable fuse? Does anyone here have experience working with any extensive voltage suppression (see the image below) on their USB port in their projects or than the basic resettable fuse that Arduino uses on the VBUS? Could you tell me in what situations would I need any more EMI suppression than a simple inductor and resettable fuse? I found a tech note on Digikey, and Im having a hard time understanding what all they are talking about. Can someone explain in layman terms what is going on?

The Arduino Micro method does look like a better simpler method to use.

Good luck;

Thanks. 

Well perhaps if you back up a little and try to explain what your are trying solve or achieve rather then various methods one might use.

I have a device that will have a constant power supply on the pcb board (it is an AC-to-DC power supply), and will provide the option to debug/service the device in the field without turning off or removing the power supply ( it will be hardwired in). I just need to make sure its OK, or rather make sure I include a method to plug in the usb on the fly without damaging my computer or the board itself. Not sure if I need a surge protector circuit just in case plugging in the USB creates some kind of back surge into the USB for that split second before the switch (I think Im rambling here  ).

Frankly I've always felt the arduino auto-voltage selector circuit was quite a waste of board space and not as fool proof or flexible as it should be.

How flexible should it be?

I very much liked the method the very first model arduino boards used, which was a simple 3 pin male header where you placed a jumper chip on one side or the other to choose USB or on-board regulator to power the board. The Seeeduino 328p boards also have a nice manual small slide switch to choose the power source. The space taken up and the cost of the extra components could have been better spent on a small RTC chip or some other useful and common function.

Board space isnt exactly a problem with me, as the parts are extremely small, and I am quite creative with running traces. In fact, I would rather tiny SMD parts as opposed to a large, bulky 3-pin header that takes up a lot of space. Besides, I already have an RTC on board, so that isnt an issue.  

In the meantime, I am using the approach that the Arduino Micro uses (shown below), which got rid of the comparator all together, and just uses two p-channel mosfets to switch between usb and vin power sources. It seems like a much more efficient way of determining which source to use, without the use of the comparator chip. I also am using a tiny part (sot23-6 footprint) which caught my attention, which uses two reverse diodes to protect the digital lines on the usb port (not sure how necessary this is).

It has nothing to do with regulated vs unregulated. Its about voltage drop.
Voltage regulators have a voltage drop.

I know that. 

I was trying to understand what Vin was, and where to place the mosfets. This still doesnt answer my two other questions:

Is there any advantage to using a comparator, or can I just go with a set of mosfets to determine using the USB VCC vs the 5V? Also as a side question, should I use reverse diodes on the D+/D- USB signal lines?

bump!

 Vin is an 'official' shield pin name for the 7-12vdc input for external DC power that which then only feeds the on-board +5vdc voltage regulator and the resistor divider feeding the comparator opamp. Vin is the same as the voltage coming from the external DC power connector except for a series polarity protection diode added between the connector and the Vin pin.

Aha!   Ok, Vin is the unregulated power supply, so I should place the mosfet between the Raw DC in and the regulator.  I know the unregulated voltage before my 5V reg is about 22V. Im thinking that is too high for the mosfet they use. I'll just use a beefier mosfet than the one they use, such as an FDN360P.

Is there any advantage to using a comparator, or can I just go with a set of mosfets to determine using the USB VCC vs the 5V?

Also as a side question, should I use reverse diodes on the D+/D- USB signal lines?

It detects the presence of Vin and switches power supplies: U1A is a comparator. It takes Vin, divides it by 2 and compares it to 3.3v. If Vin / 2 is greater than 3.3v (aka Vin > 6.6v), U1A outputs 1 and turns off T1, which isolates USBVcc from U2. So U2 is powered by +5v (which hopefully is powered by Vin at this point).

If Vin / 2 is less than 3.3v (aka Vin < 6.6v), U1A outputs 0 and turns on T1, which switches in USBVcc.

So 6.6v is the cut off point for Vin to power the device.

They could have done a better job around U1A.

But what if my Vin is a regulated +5V? And why are they comparing the Vin with 3.3V?

So if I have a 32u4 connected to USB and a voltage regulator at the same time, that would be a no-no, correct? I'd have to implement this circuit to separate the USB voltage from the circuit voltage to prevent frying my 32u4? Cant I just use a diode on the USB voltage line, or is that still not enough?

Can someone explain to me how this circuit is working? I dont get what its doing. I see that its turning on and off the USB Vin based on some input, but other than that I dont understand how it is operating or what the application for such a circuit is? (Its part of the Uno circuitry btw)

Actually many do. On closer inspection I have 2. They can be identified by a WiFi-like logo on the card. Mastercards call it paypass or blink. If the cards have a public ID, why wouldn't the card be detected?

It could be that its not a 13.56 Mhz chip on that card. If it is a 13.56Mhz chip, then maybe the IC on your reader isnt equipped to handle reading the credit card RFID.

bump!

Ok, take at look at the chips I mentioned and see how they compare to your [what I presume is an]
I2C expander. You want a display, no?

I looked at the LM3916, and I realized the difference after I posted that. One is a multiple LED/LCD driver and the other is an I2C expander for an LCD.

What values would you suggest for the Low Pass Filter?

Well I have a PCF8574N. Will that work?

Could I use the RSSI signal to determine signal strength? Can someone suggest a simple circuit that I can visually display the signal using a LED bar graph.