Pixel String Data Extender

[erm213]

Hello,

This is my fist shot at making a board. The intention is to make a transmitter/receiver pair that will allow a RGB controller board (such as the E680, or the PropController) to be far away from the start of the pixel string. My plan is to have it be 24 VDC input to the transmitter. A voltage regulator will output 5 VDC to a 485 transceiver. The 24 VDC, common, and A & B outputs of the 485 chip will be connected to a RJ45 jack.

Cat 5 cabling will connect it to the receiver. The 24 VDC will connect to a voltage regulator to drive the 485 transceiver, and ultimately the pixel string. The data out of the 485 will be connected to the data line of the pixel string. I am initially making it 5 VDC, as thats what I need. It could probably easily be adapted for 12 VDC.

I have drawn up a schematic of what I believe it should look like. I am sure I missed things, and am looking for feedback.

https://lh4.googleusercontent.com/_zoDQMLF3HQ0/TaD-qY8u5qI/AAAAAAAACzM/ZkOIA5OrbGo/s912/Pixel%20String%20Data%20Extender%20Schematic.jpg

Thanks,

Erik

[David_AVD]

A few comments:

Add decoupling capacitors to RS485 chips and regulators
The resistor on the sending end is not required
The resistor on the remote end should be 120 Ohms
The regulator on the remote end will not cope with the current / power dissipation
Won't you need two data paths for most pixel types?

[n1ist]

A few more comments:
- On the transmitter, tie ~RE high
- On the receiver, tie DE and DI low
- You don't show the RJ45 pinout. Make sure A and B are run on a twisted pair (say pins 4 and 5). Parallel the remaining lines for +24 and GND
- On the LM1085-5.0, pin 1 is GND, not ADJ
- For the kind of power draw you are looking at, you will likely want a switching regulator on the receiver

[erm213]

A few comments:

Add decoupling capacitors to RS485 chips and regulators
The resistor on the sending end is not required
The resistor on the remote end should be 120 Ohms
The regulator on the remote end will not cope with the current / power dissipation
Won't you need two data paths for most pixel types?


Thank you for the feedback. I will make the changes. Do the capacitors go between the Vcc and ground correct?

The part on the voltage regulators is wrong. For the transmitting side, I plan on using a LP2950-50LPRE3 (http://www.mouser.com/ProductDetail/Texas-Instruments/LP2950-50LPRE3/?qs=sGAEpiMZZMsGz1a6aV8DcGFffGRbHdVDvAwqdf7%252bPS 8%3d) and a LD1085V50 (http://www.mouser.com/search/ProductDetail.aspx?qs=9iSXU9KXrZlaERTj280krQ%3D%3D ) on the receiving side. Should these be ok, or would you recommend something else?

I am going to be prototyping this on the GECE strings, they have one data line, as they do not use the clock line. I should probably include the clock line. Would that need to go through a RS485 chip as well.

I used ExpressSCH to make the diagram. I installed Eagle, and am going to try to remake the diagram in there. It seems like better software. Hopefully I can make a better and more accurate schematic.

Thanks again,

Erik

[erm213]

A few more comments:
- On the transmitter, tie ~RE high
- On the receiver, tie DE and DI low
- You don't show the RJ45 pinout. Make sure A and B are run on a twisted pair (say pins 4 and 5). Parallel the remaining lines for +24 and GND
- On the LM1085-5.0, pin 1 is GND, not ADJ
- For the kind of power draw you are looking at, you will likely want a switching regulator on the receiver

High means supply it the 5V, and low means ties it to ground, correct?

I will add the RJ45 pin outs. As far as the parallel comment, do you mean make 3 lines +24, and 3 lines GND? Should the +24 be paired with a GND?

I will fix the ADJ in the schematic, I think the Eagle software will work a lot better.

Is this the kind of switching regulator you mean? LM2576TV-5G (http://www.mouser.com/ProductDetail/ON-Semiconductor/LM2576TV-5G/?qs=sGAEpiMZZMu%252byDKIQBKlySmltPskP5EaH0l3Bd5UFd s%3d)

I will post the updated schematic once I get all of the changes incorporate.

Thanks,

Erik

[David_AVD]

Thank you for the feedback. I will make the changes. Do the capacitors go between the Vcc and ground correct?
Correct. Maybe use a 1uF monolithic cap on the 485 chip supply and a 100uF electrolytic cap on the regulator input.


The part on the voltage regulators is wrong. For the transmitting side, I plan on using a LP2950-50LPRE3 (http://www.mouser.com/ProductDetail/Texas-Instruments/LP2950-50LPRE3/?qs=sGAEpiMZZMsGz1a6aV8DcGFffGRbHdVDvAwqdf7%252bPS 8%3d) and a LD1085V50 (http://www.mouser.com/search/ProductDetail.aspx?qs=9iSXU9KXrZlaERTj280krQ%3D%3D ) on the transmitting side. Should these be ok, or would you recommend something else?
The transmitter side is probably fine. The problem with the regulator on the right is that it will be dropping 19V @ whatever current the pixels draw. That's 19W of pure heat (and 19W power wasted) for just 1A of output! As n1ist says, use a switching regulator to get the high efficiency (and low heat) you need.


I am going to be prototyping this on the GECE strings, they have one data line, as they do not use the clock line. I should probably include the clock line. Would that need to go through a RS485 chip as well.
You'd need as many differential data lines (with 485 chips) as you have data signals to send.

As for the pinout, I would use:

D1+
D1-
D2+
+24
+24
D2-
GND
GND

This allows for 2 differential signals, correct pairing and doubling up on the +24 and GND wires. It's also compatible with DMX over CAT5 wiring and PoE with respect to the power.

[erm213]

Correct. Maybe use a 1uF monolithic cap on the 485 chip supply and a 100uF electrolytic cap on the regulator input.


The transmitter side is probably fine. The problem with the regulator on the right is that it will be dropping 19V @ whatever current the pixels draw. That's 19W of pure heat (and 19W power wasted) for just 1A of output! As n1ist says, use a switching regulator to get the high efficiency (and low heat) you need.


You'd need as many differential data lines (with 485 chips) as you have data signals to send.

As for the pinout, I would use:

D1+
D1-
D2+
+24
+24
D2-
GND
GND

This allows for 2 differential signals, correct pairing and doubling up on the +24 and GND wires. It's also compatible with DMX over CAT5 wiring and PoE with respect to the power.

I made the new diagram last night in the Eagle software. It's much better. I think I had made all of the changes you suggested, but it was late. I will check again tonight, and add the second data line into the schematic.

I found the LM2576TV-5G (http://www.mouser.com/ProductDetail/ON-Semiconductor/LM2576TV-5G/?qs=sGAEpiMZZMu%252byDKIQBKlySmltPskP5EaH0l3Bd5UFd s%3d) and plan on using that on the receiving side.

Thanks,

Erik

[erm213]

Ok,

I have updated the schematic. Here is the image:

https://lh4.googleusercontent.com/_zoDQMLF3HQ0/TaD-qY8u5qI/AAAAAAAACzM/ZkOIA5OrbGo/s912/Pixel%20String%20Data%20Extender%20Schematic.jpg

I have made a BOM (http://www.mouser.com:80/ProjectManager/ProjectDetail.aspx?AccessID=5e059c4617) from Mouser as well.

Would it be helpful to have the sch file from EAGLE?

Thanks,

Erik

[P. Short]

At first glance it appears that you are transposing the A and B on the data signals between the transmitter and receiver. Is that your intent?

[erm213]

At first glance it appears that you are transposing the A and B on the data signals between the transmitter and receiver. Is that your intent?

Oops, that is not my intention, and I double checked on the data sheet, I do not want to be doing that. Good catch.

Thanks,

Erik

[erm213]

I have updated the schematic and the uploaded the revised image.

Erik

[P. Short]

It would be a good idea to add a capacitor to the input of the LP2950 regulator. The LP2950 does not seem to require the input capacitor, but some potential substitute regulators (such as the MC78L05) will need it.

[RPM]

Personally, I would drop the LP2950 in favor of a standard TO220 case 7805.
Reason being is if you are going to use an input voltage of 24VDC, some of those RS485 IC variants can draw a bit of current and even as shown, if you have a differential voltage of 5V with a 120ohm termination you are going to draw about 40mA per transmitter. The regulator will be getting pretty warm at close to max current output with 24V input.

Just my 2 cents

[erm213]

Personally, I would drop the LP2950 in favor of a standard TO220 case 7805.
Reason being is if you are going to use an input voltage of 24VDC, some of those RS485 IC variants can draw a bit of current and even as shown, if you have a differential voltage of 5V with a 120ohm termination you are going to draw about 40mA per transmitter. The regulator will be getting pretty warm at close to max current output with 24V input.

Just my 2 cents

I will change to the 7805. The 2950 was chosen by searching in Mouser for the specs. I am changing the schematic to use the 7805.

Also, I just noticed I should have .1 uf caps on the RS485 ICs on the receiving side as well.

I will update the schematic and upload the latest image when I am done.

Thanks,

Erik

[RPM]

I will change to the 7805. The 2950 was chosen by searching in Mouser for the specs. I am changing the schematic to use the 7805.

Also, I just noticed I should have .1 uf caps on the RS485 ICs on the receiving side as well.

I will update the schematic and upload the latest image when I am done.

Thanks,

Erik

With a 7805 you will need at least a 0.33uF capacitor on the input side, so since your diagram shows a .1uF you are OK in that respect.

Have you worked out any cost estimates to build one of these yet?

[erm213]

With a 7805 you will need at least a 0.33uF capacitor on the input side, so since your diagram shows a .1uF you are OK in that respect.

Wait, isn't .33uF more than .1uf? Should I change it to .33uF?



Have you worked out any cost estimates to build one of these yet?

I need to update my BOM, and recalculate. It should be $20, or less for a transmitter/receiver pair (based on 1 BOM), plus the cost of the board. Based on the needed components, the board should be small. Hopefully it will be an easy on to etch. I have not looked at PCB costs, as I have not started laying them out yet.

Once I get the schematic solidified, I am going to order a couple BOM's and breadboard everything to prove it out.

Thanks,

Erik

[budude]

Wait, isn't .33uF more than .1uf? Should I change it to .33uF?

I'm guessing he was thinking 0.033uF - but the spec sheet does say 0.33uF so you are correct - you should change it...

[RPM]

I'm guessing he was thinking 0.033uF - but the spec sheet does say 0.33uF so you are correct - you should change it...

Yup... my bad. Actually I was thinking 1uF not 0.33uF.

[erm213]

I have update the BOM I made at Mouser. It can be found at http://www.mouser.com:80/ProjectManager/ProjectDetail.aspx?AccessID=5e059c4617

Currently, it costs $15.98. That included the connectors, of which almost $1 could be saved by soldering the wires to the board. I added a heat sink for both regulators. Not too bad.

I would appreciate if somebody could review the BOM for any errors, omissions, or changes.

Thanks,

Erik

[n8huntsman]

I wish I was as smart as you guys. It really amazes me when I read this stuff even though I don't really understand it. So what exactly will this thing do? I get the part about having the controller far away from the start of the strings but what's with the 24v stuff? How would that be used?

[erm213]

I wish I was as smart as you guys. It really amazes me when I read this stuff even though I don't really understand it. So what exactly will this thing do? I get the part about having the controller far away from the start of the strings but what's with the 24v stuff? How would that be used?

There were two basic ideas that could have gone with. The first, the controllwer could have just extended the distance from the controller (provide the data lines), and no power. Then you would have had to have a seperate power supply at start of each pixel string. It would make this controller much simpler and cheaper, but it would require a lot more planning on how to route the power.

The second option, the one I am pursuing, it to supply power along with the data. The idea is that you supply power to the controller board, say the E680, and it passes the power and the data down the line to the pixel string. I have a few strings of 5 VDC pixel strings (the GE Color Effects light), and then 12 VDC is the other common voltage of pixel strings. I plan on making this controller support both 5 and 12 VDC, but right now I am concentrating on 5 VDC (I am hoping it should be a simple as replacing the receiver voltage regulator that outputs 12 VDC instead of 5 VDC, and any support components, caps, etc).

The reason for 24 VDC is there will be a voltage drop when running the power down the extended lines. After the voltage drop, there would not be enough power to run the strings. The solution was to run higher voltage, and convert it down when it is close to the string. Here is a rough diagram:



24 VDC Input --> E680 Controller --> Transmistter --> Receiver --> Pixel String
24 VDC 24 VDC 24 VDC --> 5 VDC


I hope that explains it better.

Erik

[intwoit2002]

This is probably not the best place for this but here goes. I got some 2801 strings and then looked and the wires are not labeled. I got this info from Jim St. John and passing it on.


There are never any guarantees, color coding can and does change from batch to batch. Having said that, the last batch of 2801s I received were red, white, green, and blue as follows:

RED +5V
WHT GND
GRN CLK
BLU DATA

If in doubt, hooking up a single pixel is a good way to double-check. The important thing is to have +5V and ground correct. If you swap clock and data, it won't work but it won't hurt anything.

Another check, depending on how well you can see the internal wire connections of the pixels, both power and ground will have their 'in' and 'out' wires connected to the same point, whereas data and clock will have separate connection points in the pixel for in and out.

You might also post this on the forum, as others may have the same pixels you have.

[RPM]

This is probably not the best place for this but here goes. I got some 2801 strings and then looked and the wires are not labeled. I got this info from Jim St. John and passing it on.


There are never any guarantees, color coding can and does change from batch to batch. Having said that, the last batch of 2801s I received were red, white, green, and blue as follows:

RED +5V
WHT GND
GRN CLK
BLU DATA

If in doubt, hooking up a single pixel is a good way to double-check. The important thing is to have +5V and ground correct. If you swap clock and data, it won't work but it won't hurt anything.

Another check, depending on how well you can see the internal wire connections of the pixels, both power and ground will have their 'in' and 'out' wires connected to the same point, whereas data and clock will have separate connection points in the pixel for in and out.

You might also post this on the forum, as others may have the same pixels you have.

You are correct in that there are variants to the color coding. I have several strings of 2801's and some of them use the above color code, and some use this

RED +5V
BLUE DATA
WHITE CLOCK
BLACK GND

The IN and OUT side to the wires are not marked, but from my observations the IN has always been on the side where the IC is soldered to the pixel PC board.

If in doubt, check with the pixel supplier for the hookup data sheet. If you get them from Ray Wu he will email you the hookup diagram if you ask him.

[Entropy]

The transmitter side is probably fine. The problem with the regulator on the right is that it will be dropping 19V @ whatever current the pixels draw. That's 19W of pure heat (and 19W power wasted) for just 1A of output! As n1ist says, use a switching regulator to get the high efficiency (and low heat) you need.


It's moot at this point since a switcher is already in here, but just for future reference - In addition to significantly reduced efficiency, a linear regulator negates most of the advantages of power transmission at 24v, since Iin = Iout. With the switcher, Iin = Iout*(5/24)/(regulator efficiency) - e.g. significantly less than Iout.

[erm213]

Thanks for the help everyone. I am back to working on this (life got in the way for a bit). I put in a mouser order to prototype a few of these. I added a few extra parts, and different parts to play with. I will post progress, and pictures as I actually get something working. I need to update the schematic I have posted.

Thanks,

Erik

[CaptKirk]

Is the idea behind using 24V in to the tranmitter so that you have a high enough voltage at the other end of the ethernet cable to be able to regulate down to 12V eventually? In my experience, 24V power supplies are rarer than 12V (or 5V). Would this work with something like a 19V power supply (I have access to some 120W bricks that I guess would result in higher amps of 5V out of the switcher)?

If you decide to do a 12V out one, it looks like you would need at least a low current 5V regulator to get 5V for the logic from the 12V maybe?

THX, Kirk

[erm213]

Yes, the 24 VDC was chosen to account for a voltage drop along the length of the extension. The voltage regulators will work with up to 24 VDC, so 19 VDC power supply would work, albeit not as far. How much shorter? I don't know. I chose 5 VDC out, as that was my immediate need. I have several 5 VDC strings, and and a flexible LED strip to power with them. A 12 VDC out could be accomplished. The RS485 chips still need about 5 VDC, so another 7805 would be plenty for that, and there is a 12 VDC out of the LM2576. I'm sure there is something I left out.

As far as the power supply goes, I am going to use this:

http://www.circuitspecialists.com/prod.itml/icOid/9590

Erik

[CaptKirk]

Cool- I have a few of these (sorry easiest link I could find to show it):

http://www.batteryfast.com/laptop-ac-adapter/liteon/liteon-19V-6.3A-120w-4pin-round.php

Hopefully they would work.

The idea would be to get maybe 20-40 ft so the E680 could be under cover. If these would work then this gets really interesting...

[erm213]

You should be fine with that distance. According to this calculator :

http://www.netkrom.com/voltage_loss_over_cat5_calculator.html

If you are pulling the max 3A @ 12 VDC, and a 150ft cable, you would need 17.7 volts. Most regulators want the voltage to be higher, but given your distance, I think you should be fine. This is just a quick back of the napkin calculation, so it may be off.

Erik

[CaptKirk]

Thanks for the info. I could always just give it a try once this is a reality! Kirk

[erm213]

I got the transmitter side brwad boarded up. So far, so good!

Erik