It's fine to have the master between slaves. The most common problem I see is lack of DC bias on an RS485 network. Some manufacturers put the DC bias in the slaves, some put it in the masters. If you get a mix with neither having it, you can get intermittent or marginal communication.
Then, you may want to add a pull high/pull low module to provide the bias such as http://www.icpdas-usa.com/products.php?PID=3850 . You can check the DC bias easily with a volt meter, look for .2 to 5V to be good. I prefer to see at least .6V.
1) Very interesting. I'd love to think that I could solve bias issues with a commercial box. But I'm really confused about how this works; how the resistor in this box acts as a pull up resistor.
It shows a pull down bias resistor between ground and the A line and a pull up bias resistor between +5V supply and the B line.
The access to A and B lines is easy, every device has them.
Access to the signal ground line is more problematic since many 2 wire devices don't have a terminal for signal ground.
But access to the +5V driver supply line? Where's that coming from?
The cut sheet says "powered from the terminal block 10-30Vdc". OK, but is the biasing resistance connection from the B line to the DC power on the terminal block?
Unless I break open the RS-485 port and figure out where the V+ line is and bring a line out?
Or does is V+ on the box providing the bias voltage?