Data over dc-power-supply-line

A

Andreas

Guest
Hello!

For an application I have to transmit data over a dc power supply line.
The supply voltage is 3 or 5V (don't know yet), max. current about 1A.
The bus length will be shorter than 10m. What solution would you prefer
if you are not allowed to use a ready chip for this? It has to be very
cheap and very small! Footprint available is about 1cm X 2cm. There will
be a very lazy microcontroller (1MHz/1KB ROM/128B RAM) available to control.
Up to now, I would generate a certain frequency clock with the micro,
filter it to get something like a sinus and then couple it onto the
power line with a capacitor. Modulation would be AM. For receiving I
would filter the wanted frequency out of the power line and try to
detect the AM signal.
What do you think?

Andreas
 
Andreas wrote:
Hello!

For an application I have to transmit data over a dc power supply line.
Click to expand...
Phones do this- just maybe you find some ideas in the cumulative
literature from the past 100 years....
 
On Fri, 26 Mar 2004 13:01:53 +0100, Andreas <user@example.net> wrote:

For an application I have to transmit data over a dc power supply line.
The supply voltage is 3 or 5V (don't know yet), max. current about 1A.
The bus length will be shorter than 10m. What solution would you prefer
if you are not allowed to use a ready chip for this? It has to be very
cheap and very small! Footprint available is about 1cm X 2cm. There will
be a very lazy microcontroller (1MHz/1KB ROM/128B RAM) available to control.
Up to now, I would generate a certain frequency clock with the micro,
filter it to get something like a sinus and then couple it onto the
power line with a capacitor. Modulation would be AM. For receiving I
would filter the wanted frequency out of the power line and try to
detect the AM signal.
Click to expand...
Small and cheap? It's good that you didn't say anything about the
expected data rate, wire distance, or about it being reliable. Is
crude also acceptable?

Forget the carrier frequency. We're talking small and cheap here.
Let's assume your +5VDC line is fairly stable. Just modulate the
power supply regulator. Add 0.1VDC to the 5VDC line for a "one" and
subtract 0.1VDC for a "zero". Scramble the data to make sure that the
average DC voltage is +5.0VDC.

At the receive end, use a comparator to extract the data and feed it
to the micro.


--
Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
(831)421-6491 pgr (831)336-2558 home
http://www.LearnByDestroying.com AE6KS
jeffl@comix.santa-cruz.ca.us jeffl@cruzio.com
 
Andreas wrote:
Hello!

For an application I have to transmit data over a dc power supply line.
The supply voltage is 3 or 5V (don't know yet), max. current about 1A.
The bus length will be shorter than 10m. What solution would you prefer
if you are not allowed to use a ready chip for this? It has to be very
cheap and very small! Footprint available is about 1cm X 2cm. There will
be a very lazy microcontroller (1MHz/1KB ROM/128B RAM) available to
control.
Up to now, I would generate a certain frequency clock with the micro,
filter it to get something like a sinus and then couple it onto the
power line with a capacitor. Modulation would be AM. For receiving I
would filter the wanted frequency out of the power line and try to
detect the AM signal.
What do you think?

Andreas
Click to expand...
Hi,

I would put short data pulses on the 5 volt line by going to zero volts
for about 0.1 ms per pulse. All of the units taking power from the line
should have a diode in series with their power feed and a suitable
capacitor to filter power for the circuit.

Data detection is then nothing more than any logic gate connected
directly to the 5 volt line. The pattern of the pulses can then be used
in any number of ways for the data. My favorite is 'ten pulse coding'
see this link...

http://members.cox.net/berniekm/tricks.html

--
Luhan Monat: Luhan Knows at Yahoo dot Com
http://members.cox.net/berniekm
"The Future is not what it used to be."
 
On a sunny day (Fri, 26 Mar 2004 13:01:53 +0100) it happened Andreas
<user@example.net> wrote in <c4163d$pkj$1@news.uni-paderborn.de>:

Hello!

For an application I have to transmit data over a dc power supply line.
The supply voltage is 3 or 5V (don't know yet), max. current about 1A.
The bus length will be shorter than 10m. What solution would you prefer
if you are not allowed to use a ready chip for this? It has to be very
cheap and very small! Footprint available is about 1cm X 2cm. There will
be a very lazy microcontroller (1MHz/1KB ROM/128B RAM) available to control.
Up to now, I would generate a certain frequency clock with the micro,
filter it to get something like a sinus and then couple it onto the
power line with a capacitor. Modulation would be AM. For receiving I
would filter the wanted frequency out of the power line and try to
detect the AM signal.
What do you think?
AM sucks, and FM is easier made with a micro.
Click to expand...
JP
Andreas
Click to expand...
 
Hi,

thanks for the replies.
Easiest seems to use simple pulses, but don't you think the bus line
will work as an antenna for all the frequencies contained in the signal
edges (fourier)? This is not a hobby project and I have to fulfill
different EMI regulations.

Best Regards,

Andreas
 
I forgot to tell, that bitrate is about 9600 bit/s, bus length about 10
meters, no bus cable shield possible!
 
On a sunny day (Fri, 26 Mar 2004 22:18:56 +0100) it happened Andreas
<andi@andreas-bellgardt.com> wrote in <c426j7$8bq$06$4@news.t-online.com>:

Hi,

thanks for the replies.
Easiest seems to use simple pulses, but don't you think the bus line
will work as an antenna for all the frequencies contained in the signal
edges (fourier)? This is not a hobby project and I have to fulfill
different EMI regulations.

Best Regards,

Andreas
If you use FM (MFM), a simple low pass or pi filter can get rid
Click to expand...
of harmonics.
Some packet oriented system, HDLC, synchr. communication.
It can be very clean.
JP
 
I read in sci.electronics.design that Andreas <andi@andreas-
bellgardt.com> wrote (in <c426j7$8bq$06$4@news.t-online.com>) about
'Data over dc-power-supply-line', on Fri, 26 Mar 2004:

Easiest seems to use simple pulses, but don't you think the bus line
will work as an antenna for all the frequencies contained in the signal
edges (fourier)? This is not a hobby project and I have to fulfill
different EMI regulations.
Click to expand...
Control the rise and fall-times to be as long as your system will
tolerate. In principle, there's nothing wrong with near-sinusoidal
pulses if you can slice them in the receiver. You can reduce the
emissions substantially by driving the + and - supply lines
differentially. Of course, you are a bit restricted on modulation if
your receiver needs 5 V and you can send only 5 V on average. But if you
sent 8 V on average (and it doesn't matter if the data is sometimes such
that the average is not exactly 8 V), you could modulate each line by
+/- 0.5 V and still have 7 V minimum for a 5 V regulator to provide your
receiver supply. You can capacitively couple to the supply lines to get
the pulses without the standing d.c., so an ordinary single-rail op-amp
with differential inputs will recover your data.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk
 
I read in sci.electronics.design that Andreas <andi@andreas-
bellgardt.com> wrote (in <c426pn$dfq$07$2@news.t-online.com>) about
'Data over dc-power-supply-line', on Fri, 26 Mar 2004:
I forgot to tell, that bitrate is about 9600 bit/s, bus length about 10
meters, no bus cable shield possible!
Click to expand...
Use a twisted pair, driven differentially, if possible.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk
 
I think there is some confusion as to what you are trying to do. I think you
are saying you want to transmit data on a wire that is carrying 1A of DC
current. Some questions come up:
1. What keeps the DC filter capacitors from shorting out the data signal?
Maybe series transformers that can pass 1 amp?
2. Your carrier is now noise on the DC. How big a carrier can you send?
3 You mention "power line". Is that still the DC power line or 50/60 Hz
power line?

You need to AC couple your signal. A very important consideration is
baseline wander, which you get with an unmodulated signal if you are not
transmitting an equal number of 1s and 0s. Of the simple approaches that
people have done is to either
1.Transmit pulses at twice the data rate and encode, for instance, a 1 as a
10 and a 0 as a 01.
2. Send no pulse for a 0, and alternate positive or negative pulses for a 1.
For instance 10111001 would be sent as +0-+-00+. This is very robust, and
used in T1 and E1 systems all over the world.
3. Original modems all used frequency shift or phase modulation.

Look up Manchester Encoding, and Kansas City Standard. You will need
filtering to meet regulatory standards.

Tam

Just in case this opens up a new can of worms, I won't be around here for a
few days.

"Andreas" <user@example.net> wrote in message
news:c4163d$pkj$1@news.uni-paderborn.de...
Hello!

For an application I have to transmit data over a dc power supply line.
The supply voltage is 3 or 5V (don't know yet), max. current about 1A.
The bus length will be shorter than 10m. What solution would you prefer
if you are not allowed to use a ready chip for this? It has to be very
cheap and very small! Footprint available is about 1cm X 2cm. There will
be a very lazy microcontroller (1MHz/1KB ROM/128B RAM) available to
control.
Up to now, I would generate a certain frequency clock with the micro,
filter it to get something like a sinus and then couple it onto the
power line with a capacitor. Modulation would be AM. For receiving I
would filter the wanted frequency out of the power line and try to
detect the AM signal.
What do you think?

Andreas
Click to expand...
 
"Andreas" <user@example.net> wrote in message
news:c4163d$pkj$1@news.uni-paderborn.de...
| Hello!
|
| For an application I have to transmit data over a dc power supply
line.
| The supply voltage is 3 or 5V (don't know yet), max. current about 1A.
| The bus length will be shorter than 10m. What solution would you
prefer
| if you are not allowed to use a ready chip for this? It has to be very
| cheap and very small! Footprint available is about 1cm X 2cm. There
will
| be a very lazy microcontroller (1MHz/1KB ROM/128B RAM) available to
control.
| Up to now, I would generate a certain frequency clock with the micro,
| filter it to get something like a sinus and then couple it onto the
| power line with a capacitor. Modulation would be AM. For receiving I
| would filter the wanted frequency out of the power line and try to
| detect the AM signal.
| What do you think?
|
| Andreas

Ahaaaaaaaaa!

Take your 3V/5V supply, stick a small resistor in its timing capacitor
ground.

Wap a DC voltage from your processor through a divider to this resistor
to modulate the operating frequency of the supply by a little bit.

Lock a PLL at the other end and extract the ones and noughts.

Nice

DNA
 
Andreas wrote:
Hello!

For an application I have to transmit data over a dc power supply line.
The supply voltage is 3 or 5V (don't know yet), max. current about 1A.
The bus length will be shorter than 10m. What solution would you prefer
if you are not allowed to use a ready chip for this? It has to be very
cheap and very small! Footprint available is about 1cm X 2cm. There will
be a very lazy microcontroller (1MHz/1KB ROM/128B RAM) available to
control.
Click to expand...
One method for slow data rates and one- way transmission,is to make the
"DC" line AC- that is, reverse the polarity to distinguish between 1s
and 0s. Simple bridge rectifier at the receive end to recover the DC,
monitor the state of one line relative to your rectified ground to get
the bit state. You can either monitor states or edges.

To transmit back, modulate the current drawn with a resistor, and sense
it at the PSU end.

Paul Burke
 
"Paul Burke" <paul@scazon.com> wrote in message
news:c48jdd$2do5uf$1@ID-128611.news.uni-berlin.de...
Andreas wrote:
Hello!

For an application I have to transmit data over a dc power supply line.
The supply voltage is 3 or 5V (don't know yet), max. current about 1A.
The bus length will be shorter than 10m. What solution would you prefer
if you are not allowed to use a ready chip for this? It has to be very
cheap and very small! Footprint available is about 1cm X 2cm. There will
be a very lazy microcontroller (1MHz/1KB ROM/128B RAM) available to
control.

One method for slow data rates and one- way transmission,is to make the
"DC" line AC- that is, reverse the polarity to distinguish between 1s
and 0s. Simple bridge rectifier at the receive end to recover the DC,
monitor the state of one line relative to your rectified ground to get
the bit state. You can either monitor states or edges.

To transmit back, modulate the current drawn with a resistor, and sense
it at the PSU end.

Paul Burke
Click to expand...
nice. why slow though?

I have done a prototype 10Mbps 60W data link - 60W @ 48Vdc, with 10Mbps FSK
sitting on top, driven/received with RS485 transceivers. The 48Vdc supply
feeds the damped LC input filter of a smps (driving a series inductor
followed by a shunt C). To transmit data, I placed a similar LC filter at
the other end, so that the +48Vdc wire runs between 2 series L's (125uH,
NZ$0.40 each). I then capacitively coupled my 485 chips to this, with
appropriate clamping etc. It worked surprisingly well, and allowed me to
change from 2x twisted pair to 1x coax cable, greatly reducing radiated
emissions (lots of cable....). This cct was never put into production
though, as the idea occurred too late in the design cycle, and is not
backward compatible with existing product *sigh* The real shame is I had
drawn up a 750W version that I wanted to try.....
 
"Andreas" <user@example.net> wrote in message
news:c4bvve$eg9$1@news.uni-paderborn.de...
Hello,

and thanks for all the replies. I've tried something like the following
(please view in a fixed-width font). C1 is in the power supply and very
big >1000ľF. With C2, C3 I couple the data signal onto the dc power
wire. With V1, V2 I tried to prevent the DC filter capacitors from
shorting out the data signal. This works well without a load or if the
load only takes only ~1mA. If the load needs more current the load
voltage goes up and down because the diodes sometimes conduct and
sometimes not. C4 is not able to filter this and I cannot choose a
bigger one.

Then I tried inductors 220ľF instead of V1, V2. But C1 of the voltage
supply seems to be too big, that doesn't work. And again, I cannot
choose a bigger inductor.



Data I/O Data I/O
| |
=== C2 === C3
V1 | | V2
-----+----->|----+------------------------+------->|-------+--------
| |
=== C1 === C4 Load
| |
-----+-----------------------------------------------------+--------


Then tried something like this:

http://pdfserv.maxim-ic.com/en/an/onewirebus.pdf
(See figure 1)

But I have the same problems: The power supply cap. C1 is too big. The
transistor isn't able to pull the dc power line low enough for about 10us.

I need bidirectional communication. There is one big power supply (C1,
V1, Data I/O) and about 30 stations (30x C4, V2, Load).

Any comments?

And thanks again for all the hints!

Andreas
Click to expand...

If you modulate your Data I/O using two different frequency pure sine waves
(one for 1 and the other for 0), then, for sufficiently high frequencies
the inductors would present a very high impedance to your data signals,
no matter what the value of C1.
But you'll have to modulate and demodulate your Data I/O at either end
to achieve this.

Euthymios Kappos
 
"Andreas" <user@example.net> wrote in message
news:c4bvve$eg9$1@news.uni-paderborn.de...
: Hello,
:
: and thanks for all the replies. I've tried something like the
following
: (please view in a fixed-width font). C1 is in the power supply
and very
: big >1000ľF. With C2, C3 I couple the data signal onto the dc
power
: wire. With V1, V2 I tried to prevent the DC filter capacitors
from
: shorting out the data signal. This works well without a load or
if the
: load only takes only ~1mA. If the load needs more current the
load
: voltage goes up and down because the diodes sometimes conduct
and
: sometimes not. C4 is not able to filter this and I cannot choose
a
: bigger one.
:
: Then I tried inductors 220ľF instead of V1, V2. But C1 of the
voltage
: supply seems to be too big, that doesn't work. And again, I
cannot
: choose a bigger inductor.
:
:
:
: Data I/O Data I/O
: | |
: === C2 === C3
: V1 | | V2
: -----+----->|----+------------------------+------->|-------+----
----
: | |
: === C1 ===
C4 Load
: | |
: -----+-----------------------------------------------------+----
----
:
:
: Then tried something like this:
:
: http://pdfserv.maxim-ic.com/en/an/onewirebus.pdf
: (See figure 1)
:
: But I have the same problems: The power supply cap. C1 is too
big. The
: transistor isn't able to pull the dc power line low enough for
about 10us.
:
: I need bidirectional communication. There is one big power
supply (C1,
: V1, Data I/O) and about 30 stations (30x C4, V2, Load).
: Any comments?
: And thanks again for all the hints!
: Andreas

I just modeled this.... What is your baud rate?
At 220uH and with .1 coupling caps.
Using a Schmidt input gate, 115kB will work.
Lower Baud rates will not be as good.
Maybe down to 56kB..
But I wouldn't recommend it for so many loads.

I used 2200 mFd for all caps in the power side.
Clamp the gate input with a 1N4148 or similar.

I've used this scheme before at a lower baud rate, but I had only
4 loads and larger inductors. (1mH).
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top