R485 communication? Ground?

[viper]

Since my last thread went nowhere, I will slim this down.  Does having a ground wire connected between a 485 to USB adapter matter?  Im using Qmodmaster trying to poll for any response at all. Nothing with 2 wires and countless setting changes.  I have another RS485 module on the same device that seems to communicate without need for the ground. 

[Doctorandus_P]

Yes, ground matters. RS485 has a common mode range between -7 and +12V, and without the GND wire, the leveling has to be done with the data lines, and that adds distortion.

Opto isolated RS385 transceivers often (seem to) "work" without the GND wire, but you're still eating into the margins.

Sometimes the GND wire is handled in another way, for example via the mains earth wire

[viper]

Yes, ground matters. RS485 has a common mode range between -7 and +12V, and without the GND wire, the leveling has to be done with the data lines, and that adds distortion.

Opto isolated RS385 transceivers often (seem to) "work" without the GND wire, but you're still eating into the margins.

Sometimes the GND wire is handled in another way, for example via the mains earth wire

Thanks!  This is where I am, right or wrong.  It is said I need to see about 200mv of A-B separation.  I tested at 80mv.  The little dongle that will certainly work in that db9 port I measured 2kohms resistance between the gnd and B, then .24Mohms between gnd and A.  I also checked my adapter and see 4ohms gnd to B and 24kohms between gnd and A.  So it seems connecting the gnd will give the reference and a proper pull down on B?  Prob snub some noise too? 

[KrudyZ]

RS485 is bi-directional.
You will only see valid signals if one side is transmitting...

[uer166]

Yes, it matters, for similar reasons why USB needs a GND as well, and other pseudo differential buses like CAN.

[viper]

I might have realized I F'd up......  Stumbled into a vid that says that Qmodmaster has a quirk where you write the start address 1 number higher than desired.  Duh.....  I noticed to started at 1 and I had to go into settings to get it changed to a zero, in which who knows what register that is trying to poll but there ain't a 99 or 999.  Hopefully that will bring some action tomorrow.  What else did I miss.......lol

[5U4GB]

I think there's two answers here, a ground is a good idea with RS485 in general but in your case if you're seeing absolutely nothing then I think there's a bigger problem than a lack of ground.

For the wiring I've done at home, all optoisolated RS485, I don't use a ground because it's running to locations that may have a different idea of what ground is, and have never had any problems. I know this isn't good practice but just pointing out that your problem probably isn't lack of ground.

[5U4GB]

I might have realized I F'd up......  Stumbled into a vid that says that Qmodmaster has a quirk where you write the start address 1 number higher than desired. 

That's not a quirk, if you're using modbus then some things start the register count at 0 and some at 1.

For testing modbus comms, get some cheap modbus devices with known characteristics and stick that at the other end of the link.  I use a $5 4-digit LED display off Aliexpress, it both has fixed, known registers and I can see the effect of any commands immediately, no need to faff around with modbus emulators that don't quite behave as expected.

[PGPG]

Does having a ground wire connected between a 485 to USB adapter matter?

It doesn't matter in a sense 485 works or not during your experiments.
It matters in a sense of emitted disturbances and being sensitive to strong external disturbances.

For the wiring I've done at home, all optoisolated RS485, I don't use a ground because it's running to locations that may have a different idea of what ground is

For optoisolated RS485 I understand the GND connection between them as connecting insulated side GNDs and not what you can have the idea as device ground.
In my opinion....
In optoisolated RS485 you have 2 disturbance sources:
- DCDC converter supplying insulated RS485 driver,
- RS485 driver itself.
When I used DCDC with 4kV trafo with very small capacitance (4.2pF) the DCDC disturbances and disturbances generated by RS485 driver (with about 250kbps limited bandwidth) are effectively suppressed by these small capacitance inserted in ground-loop so practically adding GND connection between RS485 drivers have no effect.
When I used just standard DCDC that needed be blocked by 1nF to limit its emissions at insulated GND side than I lost the suppressing action of 4pF for driver generated disturbances. The disturbance generated by driver (mainly because one line changes voltage little before the other generating common mode pulses) are now traveling in a loop through these 1nF. Adding GND connection (not devices GND but insulated RS485 drivers GND) gives to these disturbances the short way back along the A,B wires making the loop emitting them as small as possible.
The same way you can look at sensitivity to external disturbances. Having two 4pF in series in loop limits their current (flowing through A,B) much better than having there 1nF. When (thanks to 1nF) currents are bigger adding GND connection between insulated RS485 drivers makes these current be flowing through this GND connection making their effect on your A,B signals weaker.