Joe, I haven't measured Rockwell chips for minimum THL lately. In the past (18 months ago) it was 6 to 8 dB. But that is with a very clean modem and DAA design without other problems.
When the first Rockwell V.34 chips came out I manually adjusted the THL to something greater than 60 dB. The modem would not complete the V.34 handshake under this circumstance. If the THL was decreased it would handshake. I tried this on later chips and the handshake was no problem.. The use of an active hybrid also solves other problems. The single ended receiver input (unbalanced) and passive hybrid used in Rockwell reference design makes it difficult to use some transformers. i.e. if the the transformer has high winding resistance it is difficult to get resonable THL, a 600 ohm input impedance, correct receive level, decent return loss and 600 ohm input and output impedance. I always wondered why most of the reference designs used transformers with low winding resistance which means low inductance. Low inductance means poor THD. After I was asked by a client to use a high resistance transformer with a passive hybrid I understood why. It is difficult to achieve decent values in all the parameters above. The resistors in question are the transformer sourcing resistor in series with TXA1, a resistor in series with RXIN, and resistor from RXIN to TXA2. A simple divider that also includes the transformer winding resistances. The low impedances makes it difficult to get decent values. By adding an active 2W to 4W hybrid it is possible to solve all these problems and almost any transformer can be used. The temperature effect on the transformer is also reduced to a reasonable level. Regards, Duane
