SN65MLVD206: Half Duplex M-LVDS ac coupling

Hi Aitor,

Aitor Garcia said:

In a harsh industrial environment I am having problems with a M-LVDS communication where the signals shared between transmitter and
receiver are A,B and GND.

The AC coupling solution is mainly to reset any GND shift issues. If your problem is noise, the AC coupling may not help. 

Aitor Garcia said:

Most solutions I've found is an ac coupled transmitter with a dc bias receiver.
But I am not sure if this can be done in a half duplex configuration.

It's not common to use an AC coupled set up with a bi-directional multi-node setup. This is more common in point to point communication between just two nodes.

Aitor Garcia said:

Is it possilble to use an ac coupled solution in a half duplex M-LVDS configuration where tx and receiver
lines are shared ?

I do think it is possible. If you have a point to point MLVDS set up, then the AC coupled solution would be easier to implement. If you have multiple nodes using MLVDS, it becomes trickier to set up. In a point to point case of MLVDS, you would need to have both sides/points use a DC offset (in an LVDS set up it would only be the receiver node). In the multi-node setup, you may need to put series caps on the nodes that see GND shifts and add the DC offset to it (termination resistors shouldn't be present on the mid nodes).

-Bobby

Hello,

Thanks for your response.

In case of a point to point mlvds setup, the schematic below would it be correct ?.

The dc bias polarises the bus to 1 V common mode voltage which is in the middle of the inpout common mode range.

Should I polarise A input a bit higher in order to guarantee a default state in case I chose a Type 1 MLVDS Transceiver ?

Is it needed a common mode choke in ac coupling ?

Regards;

Aitor

Overall, this looks like what I would expect both points of an MLVDS AC coupled set up to look like. 

Aitor Garcia said:

The dc bias polarises the bus to 1 V common mode voltage which is in the middle of the inpout common mode range.

This looks correct to me. The datasheet for the SN65MLVD206 shows the Vos(ss) is min 0.8V and max 1.2V so 1V being the mid point would be the best to select. 

Aitor Garcia said:

Is it needed a common mode choke in ac coupling ?

I don't think it's required but you can keep it as a DNP part and then have 0 ohm resistors between position 1-2 and 3-4 of V2. Try to reduce any stub lengths.

Aitor Garcia said:

Should I polarise A input a bit higher in order to guarantee a default state in case I chose a Type 1 MLVDS Transceiver ?

This will burn a little bit more power but yes, this would help if the driver isn't actively driving data. You could set the resistor divider you have on the A net to be 1.1V. The current transceiver you have in the schematic is a type 2 though. This would be a easy change/fix to the resistor divider values that you can modify later if you need/want to.

I see J4 looks like you want the possibility to shunt the 100 ohm resistor. If you go through J4, the characteristic impedance will change a bit from the trace into the jumper shunt and back into the trace again. It may also add longer traces on the 'A' net which adds more differential loading imbalance. In the case where you don't use the shunt, you end up with a stub with a 100 ohm resistor hanging there. 

It may be better to just populate the 100 ohm resistor between the A-B connections and depopulate it later if you think you don't need it. This would keep the differential load more balanced and reduce the stub. 

I didn't mention this in my earlier post but in an AC coupled set up, your data packets you send will need to be encoded in a way to keep all 1's and 0's balanced. (You wouldn't want to send too many 1's or 0's in a row and both 1's and 0's need to be balanced/equal to avoid letting the series cap saturate).

-Bobby