EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: scrat on February 23, 2012, 06:31:47 pm
-
Hi all, after a long time I'm back to the forum, just to ask for advice :-[.
A friend asked my though about a bidirectional communication link at about 2Mbit/s over a short (5 meters) 24 Vdc power line (up to 10A load) that he needs to make at work.
The number of pieces they will use is certainly low (1-10).
Since I think this could be quite a common specification, I looked around for standard implementations of DC powerline communication, but to my surprise I didn't find one. Many of the most common solutions are limited to the 100kbit/s range.
A commercial solution I found seems suitable: a 5Mbit/s LinkSprite module (http://www.linkspritedirect.com/product_info.php?products_id=4 (http://www.linkspritedirect.com/product_info.php?products_id=4)). This is quite expensive (50$/piece), and another problem in this case could be the fact this is designed for AC power lines, where there aren't any large parallel capacitors as terminations, as happens in a DC line instead.
I also found this forum page http://e2e.ti.com/support/interface/industrial_interface/f/142/t/75687.aspx (http://e2e.ti.com/support/interface/industrial_interface/f/142/t/75687.aspx), where the advice (in the schematic) is to use RS-485 (encoded or modulated) and differential mode chokes at the two line ends. This looks like a custom solution that could be quite long to set up and hardly reusable in the future (for example, with a longer line).
What would you say to my friend? Are there any standard ways to do the job?
Many thanks,
Sandro
-
I know there are powerline based ethernet modules to jump a wired network along home wiring. I don't remember the speed, but I'm almost 100% sure it is over 2 mbit. Not sure if that would help in this situation.
-
Thanks for your replies...
A well-designed highpass ?
And heavy filtering for the DC .
Well, the solution in the TI forum link I posted is on that way, but they use transformers at both DC sides. Simply filtering the DC will result in shunting the "HF" signal too. This also requires some kind of modulation or encoding, so which one would you choose?
I know there are powerline based ethernet modules to jump a wired network along home wiring. I don't remember the speed, but I'm almost 100% sure it is over 2 mbit. Not sure if that would help in this situation.
Yes, in fact, AFAIK the LinkSprite I mentiond (5Mbit/s) should be one of those, or very similar (it's HomePlug complaint, and HomePlug should one of the most adopted standards in this field).
It seems too much for a 5meters link on a fixed line configuration (much simpler than a power grid)...
There must be a common solution to this quite common need!
-
Interleave the signal on top of the DC voltage at the transmitter then parallel decouple the DC and one highpass for the AC , then another output low pass for the DC .
-
Interleave the signal on top of the DC voltage at the transmitter then parallel decouple the DC and one highpass for the AC , then another output low pass for the DC .
I think I've understood what you mean. High- and low- pass filtering could be done by using inductors, since the DC line has to pass 10A, so resistors are not an option at all. Moreover, a differential signalling would be more robust. In fact, this is the approach posted in the TI's forum.
The problem is still in the choice for the modulation (imagine a digital serial data source). I wanted to check for past experiences and possible common solutions. If there is no standard, there would be complete freedom.
-
Use 2 PLT units, which can do 10Mb/s over the mains,and just get some 16 SWG wire and wind an air cored coil for both lines at each end of the link. Then put another set of 10A rated chokes in, before the capacitors of the power supply and load. Put the PLT signal in and out on the wire side of the air cored chokes, and they will give a reasonable result. You are basically using a RF link, but are using the same wire as both transmitting and receiving antenna. You might want to shield them as well as twist the cores together, with a spacer between the twisted pair wires and the shield, to reduce capacitance.
-
A peculiar idea.. With PLT do you mean the LAN modules that go on AC powerlines?
-
Yes.
-
5 metres is a quite short distance.
Are there mechanical reasons why you can't run a dedicated cable alongside the DC line ?
-
5 metres is a quite short distance.
Are there mechanical reasons why you can't run a dedicated cable alongside the DC line ?
You can get some nice combined heavy DC + light shielded twisted-pair cables, used for DMX lighting applications.
http://www.directcablesystems.com/site/product_details.php?section_id=5&sub_section_id=17&category_id=25 (http://www.directcablesystems.com/site/product_details.php?section_id=5&sub_section_id=17&category_id=25)
-
@SeanB: I think a LAN module would be quite expensive and would require too much overhead on the two microcontroller units on both sides. However, using PLC (or PLT, as you call them) modules was one of the solutions I proposed.
Here http://e2e.ti.com/support/interface/industrial_interface/f/142/t/75687.aspx (http://e2e.ti.com/support/interface/industrial_interface/f/142/t/75687.aspx), they propose using differential mode chokes for decoupling, and this should work in this case, too.
@vk6zgo: That's right. I'm asking for a friend, and this is the question I asked him at first. There are no apparent reasons to complicate in such a way the things, but I must admit the challenge seems to be worth some effort before leaving it alone. A good reason for developing something on a two-wire line would be the possibility to re-use it and offer some new products in the future, as the company where he works is very small and would need to grow.
@mikeselectricstuff: Thanks for the info, I think that if they offer single pieces of cable (not the whole coil) at a reasonable price this kind of offer will make my friend change his mind.