Analog signal through a digital CMOS switch

This isn't the clearest post, but the cap coupling should not be an issue IF you someone establish a path to DC (generally ground). Floating nodes are not a good idea.

LAN? Er, as in local area network?

Please read my other post titled "Applying Pericom PI3L301D-A (10/100/1000 BaseT LAN Mux/DeMux)" from a day before.

The purpose of the circuit is to mux/demux PHY between two LANs. Usually a transformer is used between a PHY and a cable. Some older PHY's required the center tap of the transformer to be connected to a power source, however the PHY I am using (VSC8201 by Vitesse) has fully differential active drivers that can both source and sink current, so when a transformer is used with this PHY, the center tap is AC coupled to the ground. The PHY outputs are biased to about 1.5V. Now, imagine that we insert a switch between the PHY's pins and the transformer (or AC coupling capacitors if they are used instead of the transformer). The side of the switch connected to the PHY will have biasing from the PHY, while the other side will not. The question is whether it is a problem from the switch's point of view (The switch can only pass positive voltages, hence the biasing is required)? As far as I understand when the switch closes I should see the same DC offset at both sides, so I am safe, right? Or not?

Thanks, /Mikhail

No, you are not right. No one mux's the PHY lines like that...no need for it whatsoever.

Well, whether there is need or not is another question isn't it? If there was no need TI and Pericom wouldn't have parts designed specifically for this purpose.

/Mikhail

That Pericom part is intended for control & data portion of the layer, that means it goes on the side of the PHY transceiver opposite to the PHY lines, it is an interface between the control logic and transceiver.

I did check the datasheet and it goes where I said it goes.

All of those apps have a DC path between (+) and (-) of the differential outputs through the transformer, it makes no difference that the center tap is floating at DC, there will be DC on both sides of the switch because of the transformer path from the complementary side switch. This will not be the case with capacitor coupling.

Nope. Check the datasheet and the Pericom appnotes (TI doesn't seem to have any)... It is intended for the PHY lines themselves. The part numbers I am particularly interested in are TI TS3L301 and Pericom PI3L301D-A, but there are more...

/Mikhail

Enough?

/Mikhail

Right- the mux was intended to be fit into the standard magnetics configuration, it will not have a problem: View in a fixed-width font such as Courier.

. . . . . at DC . . sw . VDC ----o (+) . o---------. --------- . o | || | | . ) || ( | o . ) || ( --uuuuu-->

. high freq .----- VDC | || | jack . short | .-||------+ || +---term o . v | | || | --uuuuu-->

. --- ) || ( | . ) || ( | CM choke . VDC ----o (-) | || | | . o---------' --------- . o . . sw . .

I don't find any apps using capacitor coupling, but if they exist, they have either a high impedance choke and/or resistor across (+)/(-)...

So, in your opinion, is it a concern for the mux or for the PHY? The particular PHY I am using is a voltage mode and it doesn't require DC path through the transformer. It can happily work through capacitors. I just have to make sure the mux is happy too...

Thanks, /Mikhail

If you send me your email address I will forward to you some appnotes showing capacitor coupling. One obvious application is LAN over a backplane as in PICMG 2.16.

/Mikhail