Galvanic isolation in wireless sensor networks

[[email protected]]

Hello all,

I am working on a project consisting on controlling the light
intensity of a series of light bulbs using wireless sensor nodes. To do
this, I have built an experiment with five light bulbs, each of them
connected to an analog dimmer. These analog dimmers are controlled by
the wireless nodes. These wireless nodes have a DAC which gives up to
3V on their output. Since the analog dimmers require an input from 0 to
10 V, I built an inverter-amplifier circuit between each node and each
dimmer, using operational amplifiers.

The nodes are connected through a USB cable into USB Hubs which are in
turn connected to the wall socket.

The operational amplifiers are supplied with +15/-15/mass by a
regulable tension source, which is in turn also connected to a wall
socket.

The idea is to control each of the light bulbs indepently, telling to
each node per radio to change its DAC output. But, if I want to set a
node's DAC to a determined voltage while keeping the others to zero,
the voltage divides itself among all nodes.

It seems that there is no galvanic isolation between the node and its
USB connector, so that a circuit is being formed through the masses.

Could anyone give me some suggestions to help solve this problem?

Thank you very much,

- J.

 

[John Woodgate]

The idea is to control each of the light bulbs indepently, telling to
each node per radio to change its DAC output. But, if I want to set a
node's DAC to a determined voltage while keeping the others to zero,
the voltage divides itself among all nodes.

It seems that there is no galvanic isolation between the node and its
USB connector, so that a circuit is being formed through the masses.

You write good English, but you are using some French words. Voltage,
not tension. Earth (British) or ground (US), not mass. You will be
better understood if you use those words.

I think your problem is that all your wireless receivers are on the same
frequency. So when you operate the transmitter, they all receive the
same instruction. Do you know for certain that your radio system will
support ten independent channels, because that is what you need? Do you
have ten transmitters, or one transmitter that has ten transmitting
channels?

Another possibility is that you DO have a ten-channel system, but your
transmitter is so close to the receivers that it is breaking through on
those receivers that you are not intending to send to.

 

[poener]

Hello,

You write good English, but you are using some French words. Voltage,
not tension. Earth (British) or ground (US), not mass. You will be
better understood if you use those words.

Thanks for the corrections. I am actually Spanish, but I am living in
Germany at the moment, so it is sometimes complicated to find the
proper word, especially when talking about engineering, since one has
to use a very specific vocabulary. I will try to pay more attention to
it the next time I post.

I think your problem is that all your wireless receivers are on the same
frequency. So when you operate the transmitter, they all receive the
same instruction. Do you know for certain that your radio system will
support ten independent channels, because that is what you need? Do you
have ten transmitters, or one transmitter that has ten transmitting
channels?

You are right when you say that all receivers are on the same
frequency. But I am using a communications protocol which uses
addressing, meaning that I can send a message to a specific node, and
even though all nodes will "see" the message, only the node to which
the message is addressed to will "open" it.

Another thing is that if I have 5 nodes (with its 5 amplifiers and 5
dimmers) and I send the command to set one of them at 5 V, what I get
is approx. 1 V in each (Voltage/Number of nodes), so the problem seems
to lie on how the experiment has been built...

- J.

 

[Tim Wescott]

Hello all,

I am working on a project consisting on controlling the light
intensity of a series of light bulbs using wireless sensor nodes. To do
this, I have built an experiment with five light bulbs, each of them
connected to an analog dimmer. These analog dimmers are controlled by
the wireless nodes. These wireless nodes have a DAC which gives up to
3V on their output. Since the analog dimmers require an input from 0 to
10 V, I built an inverter-amplifier circuit between each node and each
dimmer, using operational amplifiers.

The nodes are connected through a USB cable into USB Hubs which are in
turn connected to the wall socket.

The operational amplifiers are supplied with +15/-15/mass by a
regulable tension source, which is in turn also connected to a wall
socket.

The idea is to control each of the light bulbs indepently, telling to
each node per radio to change its DAC output. But, if I want to set a
node's DAC to a determined voltage while keeping the others to zero,
the voltage divides itself among all nodes.

It seems that there is no galvanic isolation between the node and its
USB connector, so that a circuit is being formed through the masses.

Could anyone give me some suggestions to help solve this problem?

Thank you very much,

- J.

So your problem isn't galvanic isolation in a wireless network (which
should be automatic), but galvanic isolation in a (wired) USB network --
yes?

Without a wiring diagram it's hard to see what you've done. I assume
that the DACs are powered off of USB, and puts out it's voltage
independently of the others. You _should_ have two wires going from
each DAC to it's amplifier -- the signal and it's ground. Do you
perhaps just have the signal wire connected from each DAC? This could
cause the effect you're talking about.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google? See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

 

[John Woodgate]

Another thing is that if I have 5 nodes (with its 5 amplifiers and 5
dimmers) and I send the command to set one of them at 5 V, what I get
is approx. 1 V in each (Voltage/Number of nodes), so the problem seems
to lie on how the experiment has been built...

Yes, but without a lot more information about the circuit, it isn't
possible to say what is wrong. If you disconnect the supply voltages
from all but one receiver, do you then get 5 V or still 1 V?

 

[poener]

Hello,

I realise now I didn't make myself clear at all when I exposed my
problem. I've drawn a small diagramm to illustrate what I am talking
about. Here is the image:

http://images1.fotopic.net/?iid=yxouom&noresize=1&nostamp=1&quality=70

A few comments to it:

The dimmers are Velleman's K8064 DC Controlled Dimmers. Link:
http://www.velleman.be/ot/en/product/view/?id=354314

They take the AC Current from a wall socket and receive a control
voltage from the wireless mote through the amplifier.

The nodes are MoteIV's Wireless Tmote Nodes. Link:
http://www.moteiv.com/products-tmotesky.php

They generate the control voltage. Their DAC and GND pins are connected
to the amplifier circuit. They also have a USB connector from which
they are connected to a USB Hub, which is connected to the wall socket.
Eventually the USB Hub will be connected to a computer, to be able to
communicate with the nodes via USB from the computer.

The colours on the diagramm (red, yellow, blue) represent the different
voltages (ground, +15, -15) which are generated by a voltage generator,
also connected to a wall socket.

Yes, but without a lot more information about the circuit, it isn't
possible to say what is wrong. If you disconnect the supply voltages
from all but one receiver, do you then get 5 V or still 1 V?

If I work with one node only, the voltages are correct.

These nodes have the possibility of working with 2 AA Batteries instead
of USB power supply. If I disconnect all USB cables and put batteries
on each node, they also work properly. It is when they are all
connected per USB that the circuit closes and the faulty behaviour I
describe happens.

 

[John Woodgate]

The colours on the diagramm (red, yellow, blue) represent the different
voltages (ground, +15, -15) which are generated by a voltage generator,
also connected to a wall socket.

The way you have your amplifiers configured, the RED wire from your
'voltage generator' must be GROUND. It seems that you may have the BLUE
wire as GROUND (and connected to mains earth), so that when you connect
everything else, half your supply voltage is short-circuited.

 

[poener]

The way you have your amplifiers configured, the RED wire from your
'voltage generator' must be GROUND. It seems that you may have the BLUE
wire as GROUND (and connected to mains earth), so that when you connect
everything else, half your supply voltage is short-circuited.

Sorry if it looks otherwise, but red is ground; yellow and blue are
+15V and -15V.

The voltage generator is also connected to the walls sockets; this
means that the USB Hub and the voltage generator are sharing the same
ground.

 

[John Woodgate]

Sorry if it looks otherwise, but red is ground; yellow and blue are
+15V and -15V.

The voltage generator is also connected to the walls sockets; this
means that the USB Hub and the voltage generator are sharing the same
ground.

OK, then it looks like the red connection at the node is not at the same
potential as the ground of the USB connector on the node.

Disconnect all your USB cables except one. Unplug that from the node.
Everything works? Now measure to see if there is a DC voltage between
the shell of the USB connector and the red connection on the node.

 

[poener]

OK, then it looks like the red connection at the node is not at the same
potential as the ground of the USB connector on the node.

Exactly. And since there is no galvanic isolation between the USB
connector and the output pins the circuit closes.

I suspected the problem could be solved if the power generator and the
USB Hub were plugged in to wall sockets which had isolated grounds from
each other, but I tried that and it still doesn't work. Perhaps the
circuit is closing in the USB Hub?

- J.

 

[John Woodgate]

Exactly. And since there is no galvanic isolation between the USB
connector and the output pins the circuit closes.

I suspected the problem could be solved if the power generator and the
USB Hub were plugged in to wall sockets which had isolated grounds from
each other, but I tried that and it still doesn't work. Perhaps the
circuit is closing in the USB Hub?

A closed circuit doesn't close in one specific place.

Where are you taking the 5 V output from? The mote data seems a bit coy
about outputs.

 

[poener]

A closed circuit doesn't close in one specific place.

What I meant is that that the group formed by Node 1 - Amp 1 - Dimmer 1
and Node 2 - Amp 2 - Dimmer 2 should be independent from each other
(electrically isolated from each other), but if giving voltage to Node
1 causes Node 1 and Node 2 to have 1/2 of that voltage each, then they
are connected with each other...

Where are you taking the 5 V output from? The mote data seems a bit coy
about outputs.

The mote gives 0V to 3V (The difference between the DAC0 and the GND
pins). The Dimmer needs from 0V to 10V, that's why I'm using an
amplifier.

- J.

 

[John Woodgate]

In message <[email protected]>,

The mote gives 0V to 3V (The difference between the DAC0 and the GND
pins). The Dimmer needs from 0V to 10V, that's why I'm using an
amplifier.

With one node only connected to the hub, and no amplifier connected to
DAC0 and GND, address the node, so that the output on the red and black
wires is 3 V, and measure the voltage between the USB connector shell
and DAC0. If it's 3 V, connect the DAC0 and GND wires from all nodes
directly to all dimmers. Do they then work with the limitation imposed
by the 3 V output?

If they do, then your amplifiers are probably the cause of the trouble.
Have you put 100nF capacitors from each supply pin to ground on each
op-amp? Which type number op-amp are you using?

 

[poener]

With one node only connected to the hub, and no amplifier connected to
DAC0 and GND, address the node, so that the output on the red and black
wires is 3 V, and measure the voltage between the USB connector shell
and DAC0. If it's 3 V, connect the DAC0 and GND wires from all nodes
directly to all dimmers. Do they then work with the limitation imposed
by the 3 V output?

I will check this when I get to the lab. But like I said, if I power
the nodes with batteries and disconnect the USB wires, it all works.
The problem is when I connect the USB wires to the nodes.

If they do, then your amplifiers are probably the cause of the trouble.
Have you put 100nF capacitors from each supply pin to ground on each
op-amp? Which type number op-amp are you using?

No, I haven't. I am using a TI op-amp, with part number OP07CP.

- J.

PS: Thanks for your continous help!

 

[poener]

Hi,

I finally solved my problem.

Apparently, what has happening was this:

The ground of the voltage generator and the ground of the usb hub are
the same. Since I had changed the wiring on the node output to give to
the amplifier a negative input (because this is an inverting
amplifier), I was introducing voltage in the ground, which was being
derived to the other nodes.

The solution was very simple: I uncrossed the wires between the node's
output and the inverter's input and crossed them between the inverter's
output and the dimmer's input.

Now everything is working as it should.

Thanks for the help!

 

[jasen]

Hello,

Another thing is that if I have 5 nodes (with its 5 amplifiers and 5
dimmers) and I send the command to set one of them at 5 V, what I get
is approx. 1 V in each (Voltage/Number of nodes), so the problem seems
to lie on how the experiment has been built...

yeah... it sounds like an isolation problem, try running one of the nodes
from a separate supply and see if that helps.

I'm guessing you'll want to power each node individually in the end anyway...

Bye.
Jasen