Searching \ for '[EE] help for luminaria project' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: techref.massmind.org/techref/index.htm?key=help+luminaria+project
Search entire site for: 'help for luminaria project'.

Exact match. Not showing close matches.
PICList Thread
'[EE] help for luminaria project'
2006\03\03@035222 by Gökhan SEVER

picon face
hello,

   i'm planning to design a luminaria system which powered with lots of
leds. In this decorative

lighting system there will be 800 artifical stones (each stones will have 5
rgb leds in it.) which will

be buried to the roads. The system will divided to 20 pieces that every
branch will have 40 stones.

There will be a main control unit to control rgb led's colors. The main
controller is only responsible

to send proper pwm signals to the leds. So there will be only 3 cables from
main controller to the leds,

the other 2 power signals will be supplied from the seperate power
supplies'. Thats is the basic scheme

of my project.


     The problem is, how can i control the whole system if i put artificial
stones 1meter apart from each

others. That makes at least 800 meters. I think, i need to place an
amplifier for amplifying the pwm

signals to approciate levels. But how?


All of the comments and suggestions are very appreciated.

Gökhan SEVER

2006\03\03@075119 by olin piclist

face picon face
"Gökhan SEVER" wrote:
> There will be a main control unit to control rgb led's colors. The main
> controller is only responsible
> to send proper pwm signals to the leds. So there will be only 3 cables
> from main controller to the leds,

This seems like a bad idea.  The PWM signals are essentially analog and
should be generated locally.  Use a single communication signal and transmit
the desired red, green, and blue intensity digitally, then convert that to
PWM locally as needed.  For good noise immunity, you should use a
differential pair to transmit the digital intensity information.  If you
stick to 0-5V differential, you can use off the shelf RS-485 driver and
receiver chips.

To keep the signal from degrading as it pass along a chain of stones, have
each stone receive each byte, then re-transmit it.  That should allow you to
push the signal indefinitely far at the cost of about 1mS latency per stone
if using 9600 baud.

Delivering power over long distances without excessively large and expensive
cables is more of a challenge.  I'd probably buss around 48V and put a small
buck regulator in each stone.  The cost saving on cable alone will probably
pay for the switchers.


******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products

2006\03\03@082952 by Gökhan SEVER

picon face
{Quote hidden}

We have similar solution in our company but if i apply this scenerio to new
system, this cost 800 p12f675s and 800 rs-485 driver ic plus nearly 20
repeater excluding power stage.

So, i'm looking for possibilities to change  the  existing system to lower
the ic costs.

Besides Mr. Lathrop,  can i adress each stone separately with your suggested
method to control each one individually?

2006\03\03@084824 by olin piclist

face picon face
"Gökhan SEVER" wrote:
> Besides Mr. Lathrop,  can i adress each stone separately with your
> suggested method to control each one individually?

You can make the protocol anything you want.

It wasn't clear from your original post if this was a requirement.  Are you
running 3 separate RGB PWM signals to each stone from a central controller
with each stone possibly displaying a different brightness?  If so, then a
digital daisy chain approach should save a lot of money on copper cabling.
How is the power distributed now?  What voltage/current does each stone get
now?


******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products

2006\03\03@105258 by Gökhan SEVER

picon face
it seems like the file is rejected, so ive uploaded the file to rapidshare

rapidshare.de/files/14589914/luminarie_project.jpg.html

2006\03\03@152237 by Robert Ammerman

picon face
The OP:
> There will be a main control unit to control rgb led's colors. The main
> controller is only responsible
> to send proper pwm signals to the leds. So there will be only 3 cables
> from main controller to the leds,

Olin:
--This seems like a bad idea.  The PWM signals are essentially analog and
--should be generated locally. Use a single communication signal and
transmit
--the desired red, green, and blue intensity digitally, then convert that to
--PWM locally as needed. For good noise immunity, you should use a
--differential pair to transmit the digital intensity information.  If you
--stick to 0-5V differential, you can use off the shelf RS-485 driver and
--receiver chips.

I pretty much agree with this except that I would strongly consider sending
the control signal via a simple current loop, with series connected
optoisolators at each board to receive the signal. This eliminates the need
to pass the signal on from stone to stone. If necessary, the signal could be
regenerated at any given stone with a simple circuit (a switchable current
source driven by the output of the optoisolator) and powered by the main bus
power to continue from there. To avoid the 'christmas tree light' problem in
event of finding location of a break in the loop, the stones should display
some type of error signal when they are not seeing a control input, but are
seeing power.

--Delivering power over long distances without excessively large and
expensive
--cables is more of a challenge.  I'd probably buss around 48V and put a
small
--buck regulator in each stone.  The cost saving on cable alone will
probably
--pay for the switchers.

Absolutely! Or even set the stones up to always pass the same current,
regardless of the state of the LED's and then just wire them up in series
(just kidding!).

Also, note that having even minimum local intelligence at each stone allows
for much more interesting displays than if the entire chain of stones is
simply driven by the same analog signal.

Sounds like a couple of 10F's at each stone: 1 for an Olin switcher, 1 for
the actual control.

=========================

After thinking about this a little bit, how about this idea:

We wire everything up with a single pair of wires driven via an H-bridge
from the main controller to the stones as follows:

We would set up a signal that repeated at a given 'baud interval', perhaps 1
KHz. This would allow us to send 1 Kbaud on the link. Using the pulse width
modulation scheme outlined below, we would be able to send four bits for
each baud interval, so that we would then have 4 Kbits per second of data.

During the first part of the baud interval the H-bridge would drive in the
forward direction some multiple of 5% of the full interval. The length of
this signal can be decoded by the stone as follows:

10%    - Special 'start bit' marker to indicate the beginning of a command
frame, or an idle condition on the link.

15% - 0000
20% - 0001
25% - 0010
etc.
90% - 1111

For the remaining portion of the baud interval the H-bridge drives in the
reverse direction.

Then, at each stone we simply full wave rectify and filter (it won't take
much filtering) the signal to provide unregulated power for the stone at,
for example 48V.

An optoisolator back to back with a diode and in series with a resistor
connects directly to the input signal ahead of the bridge rectifier to
provide a logic-level signal to the PIC.

--------

Finally, with a little bit of care, it should be pretty easy to have the
same microcontroller serve as both the buck switcher and the actual
application processor! This could be done as follows:

1: Set up a simple R-Zener-cap regulator driven off the 48V power capable of
powering only the PIC at a relatively low voltage (say 4.0V or so). Connect
this to the PICs Vdd via a diode and also to an analog input or comparator
input pin on the PIC via a voltage divider.

2: Set up a series PFET driven by the filtered 48V line that is held in
cutoff by an R until the PIC is up and running. The output of the series FET
is then filtered to provide the normal operating Vdd of the circuit, which
is supplied to the PIC Vdd via a diode, but which can  be directly connected
to the LED anodes (no need to have the voltage drop of this diode in the LED
circuit).

3: The PIC can then use the voltage on the analog input, referenced to its
Vdd to determine when it needs to get a 'burst of power' and turn on the
FET. The FET is driven via a gate capacitor for two reasons: (a) to level
shift from the PIC output to the 48V world, and (b) to serve as a
'keep-the-smoke-in' circuit by preventing the FET from staying on if the PIC
output gets stuck. A diode from the FET gate to the 48V line dumps the
charge in the C when the input to the cap goes from low-to-high. A resistor
from the FET gate to the 48V line keeps the FET in cutoff and serves as the
R in the RC timing for turning off the FET if the PIC goes south.

4: Assuming each PIC had some sort of ID or serial number in it, they could
even out the load on the 48V bus by drawing their power during a 'timeslot'
within the baud interval that was a function of their ID. This would greatly
reduce the need to filter the 48V bus power at each stone.

---------

So, requirements for the PIC:

1: Analog in or comparator input pin.

2: Three PINs to drive RGB leds.

3: One input to receive the control signal.

4: One output to drive the regulator FET.

6: I/Os altogether. It won't fit in a 10F, but a 12F can do it!

-----------

A rough parts list for the stone:

Input bridge rectifier
48V bus filter C

2 Vdd steering diodes
12F PIC
Bypass C for PIC power

Zener
R for R-Zener reg
2 Rs for analog input voltage divider
Vzener filter C

Switcher PFET
C for switcher FET gate drive
diode to dump C storage when drive turned off
R to hold switcher FET in cutoff
Vreg filter C

3 Series Rs for LEDs
3 FETs/BJTs to drive LEDs
LEDs

Optoisolator
Back-to-back diode for optoisolator
Series R for optoisolator
Pullup R for output of optoisolator (or use pullup in PIC)

Not too bad at all!

Bob Ammerman
RAm Systems


2006\03\03@153746 by Robert Ammerman

picon face
> We have similar solution in our company but if i apply this scenerio to
> new
> system, this cost 800 p12f675s and 800 rs-485 driver ic plus nearly 20
> repeater excluding power stage.


If you read the design at the end of my previous post on this, you will note
that the costs for each stone are quite minimal. They should be more than
made up for by the fact that you only have to run one pair of relatively
small guage wire to support a whole chain of stones.

The PIC code to handle the simultaneous decoding of the input, control of
the buck regulator and generation of the PWM would be, to say the least,
interesting. However, I don't think that it would be difficult in any real
sense. It would just require carefully written isochronous code.

I have already developed a similar system of sending the data signal using
pulse width modulation of the power and it works perfectly (central clock
system, slave clocks receive power, current time and bell control commands
over one pair).

Also, of course, the individual stones could be addressable in this scheme,
permitting very attractive displays indeed.

Bob Ammerman
RAm Systems


More... (looser matching)
- Last day of these posts
- In 2006 , 2007 only
- Today
- New search...