Using RS485 network to supply voltage.

[TonyHunt]

Hi,

I have Pic based sensors on a 485 network. At the moment they are powered locally with replaceable batteries and they are giving a reasonable acceptable life. However, I would like to look at the possibility of powering these with the voltage on the network. The sensors have a current drain of <3mA each and there are 4 on the network. I feel certain that this is not a new idea and would be delighted to hear any thoughts on the subject.

 

[mneary]

I would wire them with a four conductor cable.

 

[dougy83]

The slaves can be powered from the master's transmit lines using a bridge rectifier. If it's a 4wire rs485 network, keep the master driving the transmit pair and the battery at the sensors can be replaced with a small capacitor. If it's a 2wire network, you'll want a bigger capacitor (or just a rechargable battery) and make sure the master is driving the lines when the slaves aren't transmitting.

 

[TonyHunt]

Thank you both.
The existing network is 2 wire and I'm commited to that for future projects.
I'd like to avoid rechargeable batteries if possible as even the best need replacing eventually and the next project will have very limited access.

Bridge rectifier and larger capacitor sounds the way to go.

What sort of capacity would be a good starting point? Or better still, could you point me to any appropriate articles?

 

[dougy83]

The capacitor size corresponds to the time that it must supply the sensor (i.e. during the time any slave is transmitting) and the amount of current the sensor draws. The supply voltage will obviously droop during this time; if the sensor is not dependant on supply voltage it's not an issue. If it is dependant, you'll need to provide some regulation/stable reference.

So far as working out the size, CV = IT (capacitance * voltage = current * time, in SI units). e.g. 0.1F (a common supercapacitor value) charged to 5V will drop to 4V in 33 seconds with a 3mA load, so 0.1F is more than enough assuming the master recharges the cap at least every 33 seconds and that the sensor works with a voltage in that range. 1000uF will drop a volt in 330ms. It depends how long a slave takes to transmit its data before the master starts driving the bus again.

 

[mneary]

An RS485 bus commonly has between 1 and 2V between the two wires, it's possible to see less especially if there's too much termination. A bridge rectifier might never turn on.

I suppose it's too late to have a common ground? It would make it a lot easier.

 

[TonyHunt]

<<between 1 and 2V between the two wires>>
I'd never considered that, I'll measure it later and maybe play with termination resistors. 120 ohms at the moment I think.

 

[dougy83]

I'd never considered that

Nor I. That's a very good point.

Foreseeable solution: just up the driver capacity or change the termination method.

What's the datarate and sensor sample period?

 

[dougy83]

What about changing the termination to 120 ohm in series with 1uF? Possibly use schottky diodes for the bridge.