# Comm over DC Modulation

#### cgha20@yahoo.com

Joined Oct 21, 2009
78
I have a project that requires communication over a DC Bus (19V to 41V). I will use a microcontrollers UART to modulate a signal and have read I can use inductive coupling to couple AC to DC line. And then decouple on other end. Also I have read I will require inductors at DC power supply to isolate AC from DC. I have also read I can use a transformer to couple my ac to DC line. What is the best way of doing this. What circuit can I build/design that does not require a microcontroller to at least test this before hand. I have all lab equipment available: function generator, oscilloscope, dc power supply. I need help!

#### joeyd999

Joined Jun 6, 2011
4,477
You don't mention what your bit rate is, but, assuming it's high, PoE (power over ethernet) is a state-of-the-art solution. Here is how they do it:

#### AnalogKid

Joined Aug 1, 2013
9,352
And I need data!

Let's see... UART ... 3.3V or 5V? data rate? Is the DC wandering around between 19V and 41V, or is it stable somewhere between those extremes? What is the current or power level of the DC bus? What is the distance the data will travel on the bus? Do you have access to the DC power supply regulator circuit, or must you deal with the BUS directly? This is *much* easier if you can turn the output of the DC source up and down through its regulator.

Also, are you thinking of direct modulation at the base data rate, or putting the data on a high frequency FM carrier like home automation systems use?

ak

#### cgha20@yahoo.com

Joined Oct 21, 2009
78
Communication signal can be 3.3V or 5V. Lets choose 5V for sake of this discussion. It will be a relatively slow communication. What I need is a 1 represented by 300kHz and 0 represented by 100kHz. The Voltage will be stable but needs to be anywhere between 19-41V range. It will be stable DC. Most likely at 28V but can be anywhere in between. The 19-41VDC line will be an input for me so I have no control over it. Direct modulation. Basically there will be a master board talking to slaves and receiving response from slaves. The part i am experimenting with is using signal gen to simulate the carrier and then adding ac signal and transmitting on dc bus and then demodulating and separating the ac signal. Havent done something like this before

#### cgha20@yahoo.com

Joined Oct 21, 2009
78
Master to node connection can me at longest 50feet and at shortest 6 feet.

#### AnalogKid

Joined Aug 1, 2013
9,352
Read up on power line communication, X-10, power line intercoms, etc.
https://en.wikipedia.org/wiki/Power-line_communication

A critical parameter is the impedance of the DC bus at the datacom frequencies. It is fairly easy to superimpose 100 kHz information on an AC powerline because AC powerlines do not have decoupling capacitors and no matter how low the output impedance of the source is, it is miles away. A local DC supply might not be so accommodating.

Your two frequencies are very far apart and harmonically related. Neither of these is good for FSK datacomm. A 100 kHz square wave has sinewave energy at 300 kHz equal to 42% of the fundamental square wave amplitude. This is enough to cause false detection. using sine wave carriers is better, but you should pick the two key frequencies such that their harmonics do not overlap.

ak

#### crutschow

Joined Mar 14, 2008
27,710
What is the maximum DC current over the line?

#### cgha20@yahoo.com

Joined Oct 21, 2009
78
Maximum current on line is approximately 5amps

#### Sensacell

Joined Jun 19, 2012
2,890
If your data rate is low (110 baud or less?) you could send the data by keying the power on and off, with a diode and a big filter cap at the receiving end to 'hold up' the output voltage during the 'low' signalling periods.

Alternately, you could also use a dual voltage approach, switching between two voltages- assuming you could regulate the power on the receiving end.

This idea is simple and crude, but workable for low data rates.

#### cgha20@yahoo.com

Joined Oct 21, 2009
78
So I am testing a circuit I found online but I am unable to reconstruct after the comparator to make it look like my original signal. Any ideas? Between C1 and C2 would be where I transmit to DC bus as well as receive from DC bus

#### Attachments

• 377.1 KB Views: 16

#### crutschow

Joined Mar 14, 2008
27,710
Your circuit assumes the DC bus has a high impedance to ground but it's likely to have a low impedance from the decoupling capacitors.
Thus you would need some inductors to provide a high impedance to the AC signal to ground or you need to sense the AC current, not AC voltage.

I don't understand why the comparator output has such slow rise and fall times. The comparator output transition should be very fast. Does it need a pull-up resistor at the output?

#### cgha20@yahoo.com

Joined Oct 21, 2009
78
Where would I place the inductors? How do I select an appropriate value? Also, what would I need to do so that at the output of the comparator I get a match to Channel 1 signal.?

#### crutschow

Joined Mar 14, 2008
27,710
Where would I place the inductors? How do I select an appropriate value? Also, what would I need to do so that at the output of the comparator I get a match to Channel 1 signal.?
The inductors would go in series with the power supply output and the circuit power input.

Your comparator model doesn't seem to be working properly.
Do you have a different model comparator you can use in the simulation?

#### cgha20@yahoo.com

Joined Oct 21, 2009
78
Crutchshow, Ok I see what you mean. The inductors you mention are on the reference drawing. Actually the final drawing will have 100uH inductors. Good catch. What do you think may be causing the problem on comparator? Based on the previous screen captures input to comparator, should my output be the original signal? Also, I think I found the output transition problem. Look at channel 4 scope signal. It looks like there is a transition at correct timing, but it is at a very small level. How can I make the first very attenuated transition be my starting point instead of next rising edge.

#### Attachments

• 237.3 KB Views: 8

#### crutschow

Joined Mar 14, 2008
27,710
As I previously asked, do you have a different comparator model you can use?

#### cgha20@yahoo.com

Joined Oct 21, 2009
78
Here is with a different COMP that seems to be working. Why do you think it does the signal not look like input signal?

#### Attachments

• 389.3 KB Views: 6

#### crutschow

Joined Mar 14, 2008
27,710
Here is with a different COMP that seems to be working. Why do you think it does the signal not look like input signal?
Because your R3C3 filter time-constant is only 1μs.
Why is it so short?
It needs to be longer than the carrier frequency period.

#### cgha20@yahoo.com

Joined Oct 21, 2009
78
OK so I got it to work now. Thanks The last issue to resolve is why my timing is off by 5uS. please take a look at drawings of rise and fall time

#### Attachments

• 309.6 KB Views: 5
• 81.2 KB Views: 4

#### crutschow

Joined Mar 14, 2008
27,710
OK so I got it to work now. Thanks The last issue to resolve is why my timing is off by 5uS. please take a look at drawings of rise and fall time
It takes some time to recognize a signal due to the RC filter. That's unavoidable with such a simple circuit.
How fast a response do you need?

Why do you have C1 and C2 in series?
All that does is make the total series capacitance about 0.9nF.