News:

LATEST RELEASE:  FPP 6.1 - Download from here - https://github.com/FalconChristmas/fpp/releases/tag/6.1

+-+-

+-User

Welcome, Guest.
Please login or register.
 
 
 
Forgot your password?

+-Site Stats

Members
Total Members: 15290
Latest: eatumup
New This Month: 36
New This Week: 5
New Today: 4
Stats
Total Posts: 125007
Total Topics: 15226
Most Online Today: 89
Most Online Ever: 7634
(January 21, 2020, 02:14:03 AM)
Users Online
Members: 5
Guests: 60
Total: 65

Feature request for micro boards

Started by lrhorer, January 03, 2017, 03:21:35 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

lrhorer

While almost all of us are using 12V pixels, there are still quite a few strings and other devices that use 5V.  Indeed, Ray Wu sells some very nice pixel strings that are only available in 5V.  This one is an example:  https://www.aliexpress.com/store/product/Addressable-DC5V-WS2812B-IP68-pixel-node-12mm-open-hole-epoxy-resin-filled-50pcs-a-string/701799_32337071529.html?spm=2114.12010608.0.0.vm3M4r

I find it far superior to the ones we most often use in many ways.  In any case, my suggestion would be to create a couple of micro boards with built-in 5VDC buck converters.  I think there might be some interest for these out there.

One would be a uSC with a 5V output to the string rather than a 12V.  Obviously, it would have to be a bit more expensive, but stand-alone 5V, 3A peak buck converters can be had for well under $1, so I would not expect designing one into a uAmp board would add much cost.

Another idea I have would be a uAmp with built-in 5V converter plus an output for data.  This board would sport three input pins / pads, three output pins / pads, and 4 pins / pads for the pixel string.  Three of the pixel string pins / pads would feed the pixel inputs, while the fourth would accept a return data line from the end of the string.  Three jumpers or switches on the board could select which data line feeds the input of the uAmp, which one feeds the pixel data port, and which one feeds the output data port.  This way, the uAmp could be placed either before or after the pixel string, whichever is more appropriate.  Incidentally, the user could also choose to split the uAmp output and feed the same data to the pixel data port and the output data port.  Of course ground would pass through on all three ports, and 12V would pass between the input and output port, with 5V feeding the pixel string port.  See the attached drawing.


lrhorer


corey872

Thanks for the input.  I will take a look into it, but I always try to be cautious about adding extra cost to a board when a majority of people would not see a benefit.  We also have to keep a close eye on economy of scale.  I see the one dollar 5V/3A regulators on ebay, but just the first 3-4 listings I see have a combined 'units sold' total over 10,000.  Considering all vendors and all points of sale the build quantity could easily be many times that 10K figure.

As far as the actual boards, the micro series boards already pass along what ever voltage is sent to them - and they are tolerant of 5V input.  So if you want to run 5V pixels, simply send 5V to the board, it will work fine and pass the 5V onto the pixel.  Putting a 5V converter on every board would only benefit those using 5V pixels, but chose to use a 12V main power supply then step down to 5V at every prop.  A 3A supply would further limit use to ~50 or 60 nodes where uSC is capable of 150 and uAmp is essentially unlimited.

I'm not sure I follow adding all the pins/pads to the uAmp...

If you want "Pixel Data Out", then that is essentially just bypassing the uAmp, so not sure why you'd pay for a board and install, just to bypass it? If you were looking to 'daisy chain' or 'branch out', that is already somewhat covered in the manual where multiple uAmps can be added to one 'Data In' line and generate separate (and strong) 'Data Out' signals. ie = run 3, 4, 5...node strings from one 'Data Input'

"... uAmp could be placed either before or after the pixel string, whichever is more appropriate." - The data only flows one way through a string,
so having uAmp 'after' the string would require a long data lead to go back to the 'input' of the string.  An equivalent function is 'reverse channel order' in the set-up utility.  This allows the 'data input' end of a string to be either the highest or lowest channel number of the string so the channel layout can be in a logical order regardless of which way the string faces.

"pixel data return" - We have no real 'return' data.  3 bytes of data are consumed at each pixel and the remainder is propagated down the string.  The last node takes its 3 bytes and that is that.  The only way to have data at the end of the string would be to program more nodes in the set-up
than exist in the string.

Corey

lrhorer

Think of the topology for any display that has geographically diverse groups of elements.  One example would be a set of snowfall / meteor tubes.  Another might be a bunch of snowflakes.  As it happens, I have both.  I can think of others.  Sending 5V to the entire set of tubes is not the best notion.  The current may be fairly high - easily 8A or more - and the headroom not very great.  A string of ten 5V tubes spaced 2m apart is liable not to work without power insertion.  OTOH, use 12V, and the display works just fine at only 3.3 amps or so from the main supply.  Each group will have an input data stream going into the first pixel of the display element, and a return data stream coming out of the last element back to the same physical location where the input to the display element resides.  This stream needs to be passed on to the next element.  The 5V power terminates at the last pixel of the element, while 12V is carried on to the next display element.  In some cases, it is appropriate to put the uAmp on the input to the element, prior to the first pixel.  In others, it may be better to put the uAmp  on the output of the element, after the last pixel of the element.  The board layout I propose would allow the show designer to select either without having to have any unique elements or unique wiring.  Of course, as I mentioned, it also allows the designer to create multiple identically addressed elements wherever he wants, once again without any unique or different wiring.  Set the jumpers accordingly, and the next element will be identical to the one at hand.  I'll try to do a couple of additional illustrations tonight or tomorrow night that may make it clearer.

lrhorer

#4
OK, here we go.  Take a look at the following tables:



Both assume a string of 5 Volt display elements, each drawing 10.6 Watts (22 pixels at .48 Watts each).  To power each, we will use 3 pairs of a 4 pair Cat5e cable, each 2 meters long.  The minimum input to a 5V element is going to be around 4V.  Less than that, and it will be noticeably dim.  The top table shows the input voltage to each element for a given number of elements.  Using a 5V main supply, as you can see, we can only deploy a single display element unless we use power injection.  (If you think about it, power injection at each element is actually what I am suggesting.  One can think of the 5V Buck converter as a power injector.)  We could overcome this by employing a larger conductor for the power, of course, but to deploy 6 of these display units, we would need to use 12 AWG wire, with a power supply capable of delivering more than 12.7 Amps continuously.

Compare that to the bottom table, where we use a 12V main supply with a 5V buck converter at each display element.  Since each buck converter can deliver 5V as long as its input is higher than 6V, we can easily deploy 6 elements using our old friend Cat5, and we can probably squeak in a 7th whose input voltage will be about 4.2V.  All of this with a 12V, 6.5A power supply.  The Cat5 is a lot cheaper than 12 AWG copper, and it solders easily into the uAmp and uSC modules.  Not only is 12 AWG wire heavier and more expensive than Cat5, it also has to be spliced into smaller wires at the module junctions.  That's a lot of extra time and effort, as well as expense.

lrhorer

#5
In the last post I talked about the advantages of incorporating a 5V converter with each uAmp.  Note I am not suggesting you do this with every uAmp.  Rather I am suggesting a different, more expensive model to go with the micro modules.  Unless, that is, the hobby can produce enough uAmps with 5V converter to make the cost difference between the two less than $1 - which I really don't expect.  The other mod I suggested probably doesn't make too much sense on a uAmp unless it has the buck converter embedded, as well.  It would make it a little bit more convenient, perhaps, but not much.  Look at the following drawing:



In some cases, it is required to place the uAmp before the display element.  In others, after the display element.  Of course, if there is no buck converter, it doesn't make a whole lot of difference.  If one wants generic cabling and generic display elements, one can simply connectorize the uAmps themselves with an input and output connector and plug them in before or after the element as needed.  It does mean a few more widgets to box up, but the uAmps are small.  If one has a uAmp and Buck converter with the wiring mod I suggested, however, one can attach the same module permantly to each display element and employ completely identical elements and identical cables (except possibly the length) for every piece of the show.  Rearranging a show won't require any soldering or modifications to cables or elements at all.  Just pick the right length of pre-made cables, plug together the elements in the right order, and set the uAmp jumpers to where the uAmp needs to sit.

lrhorer

Oh, and just some food for thought:  If we were able to hike the main voltage to 24V, then we could easily extend the number of display elements to 10, or maybe even push it a bit to 11 or 12.  Of course, at 10 or 11 elements of 22 pixels each, we are running up against the address limits of the controllers, unless some of the elements are identically addressed.

lrhorer

Any further thoughts, questions, or comments on my request?

corey872

I'm afraid this particular use case would be a small subset of 5V users, and 5V users themselves seem to be a small subset of all users.  So I feel that adding cost to every board which would only be a benefit to a small group of users would result in fewer boards bought overall... if there even is a uSC co-op for 2017.  We only sold ~125 boards in 2016.  At levels much lower than that it's not really feasible to do the co-op.

What I can do is offer the ebay converter with the co-op.   It obviously won't be any cheaper, but if anyone wants to do a 'one stop shop' they could get converters and boards all in one purchase.
Corey

zwiller

I am currently testing the LM1804 5V and I think they might be an option.  http://www.ebay.com/itm/280952879925?_trksid=p2060353.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT  TI data: http://www.ti.com/lit/ds/symlink/lm1084.pdf

I bought extras and if you guys want a few to fool with give me PM.  I'd like to get some other feedback before going ape with them.  I plan to use these to run a few 5v things like komblinkin strobes.  Rated up to 25V too so you could try some interesting things... 
Sam

lrhorer

No, that's a linear supply.  Definitely not a good solution.  The point of going with a higher transport voltage is to reduce the current and thus the line loss in the power leads.  Using a linear supply would not do so, at all.  Indeed, they increase the current draw by a tiny amount.

lrhorer

#11
Quote from: corey872 on January 31, 2017, 07:21:54 PM
I'm afraid this particular use case would be a small subset of 5V users, and 5V users themselves seem to be a small subset of all users.  So I feel that adding cost to every board which would only be a benefit to a small group of users would result in fewer boards bought overall... if there even is a uSC co-op for 2017.  We only sold ~125 boards in 2016.
I see your point concerning the uSC.  I surely hope there is a 2017 co-op.  I need some, but this is about the uAmp, not the uSC.  It looks to me like you sold well over 600 of the uAmps.

What about if you designed the uAmp PC board with 3 solder pads allowing the user to jumper the 12V port if they do not need a buck converter, but can attach one if they do?  Make the pad spacing the same as a pin header so one can use jumper headers to attach the supply or bridge around it.  That should only increase the size of the board by a small amount and the cost only a few cents (for the jumper pins).   It then would be a universal board allowing management of the uAmp location (before or after the pixel string) power insertion (from input / output, from external supply, or from a buck converter powered from the input / output) and even allowing for duplicate strings, all on one universal board attached in an identical fashion to each string, whether it be for a long linear string, a meteor tube, one star in a chain of stars, or whatever.  You could even make the jumper pins optional, allowing the user to employ solder bridges rather than pin jumpers to keep the cost almost the same, and also allowing the user to customize any element and heat-shrink the uAmp, if they so desire.

Heck, if the jumpers are arranged the way I show them in the diagram, it even allows the user to bypass the uAmp section altogether.  In a geographically diverse display, some elements will need uAmps, and some not.  The proposed board would allow the user to make every single string element in  the display precisely identical, irrespective of where it will be installed in the display or what the power needs will be.  The display inventory is then vastly reduced, since each element is identical and can be placed anywhere in the display.  Any unique requirements for power or data conditioning can be handled by changing the jumpers or inserting a supply.  Changing the number and location of any element from year to year is a simple matter, requiring only moving the pins, rather than a lot of soldering and re-wiring.

lrhorer

Here is a diagram with the proposed changes:



zwiller

Quote from: lrhorer on February 03, 2017, 03:14:46 AM
No, that's a linear supply.  Definitely not a good solution.  The point of going with a higher transport voltage is to reduce the current and thus the line loss in the power leads.  Using a linear supply would not do so, at all.  Indeed, they increase the current draw by a tiny amount.

I should have been clearer, I am suggesting to use 12v transport but step down to 5v at prop.  That's what I am doing.  Certainly a less elegant solution, but the komblinkin is designed for TA200 and plenty of room to do it there. 
Sam

lrhorer

Yes, I know.  A series pass (e.g. linear) regulator gains one very little in that scenario.  The only advantage is starting with 12 - 24 Volts, one has more headroom before the voltage drops to the minimum usable level - about 6V.  If the string pulls, say, 1A at 5V, then with a linear regulator like the LM1084, one is still pulling 1A from the man supply at 12V or more.  The efficiency is barely 40%, and the voltage drop on the power lines is still the same as if one had a 5V main supply.  With a buck converter, however, the efficiency can be greater than 90%.   If the same string is pulling 1A at 5V, then the buck converter will only draw about 0.46A - less than half the current using the LM1084.  Take a look at the charts above.  Although a linear regulator will allow more strings than the 5V main supply, the number is less than half of that allowed by using buck converters.

Support FPP

+- Recent Topics

K16a-b FPP setup by dkulp
Today at 06:15:18 AM

OrangePi 3 LTS wifi by ctmal
September 25, 2022, 05:27:37 PM

Pi Top on sale by CeP
September 25, 2022, 04:26:29 PM

no Lights by Poporacer
September 24, 2022, 06:47:12 PM

exceding 24000 starting channels by Poporacer
September 24, 2022, 05:53:25 PM

No signs of life, help by JonD
September 24, 2022, 02:44:30 PM

Sequencing timing... by Atljim965
September 24, 2022, 10:22:25 AM

OLED Screen Not Coming On by jja0607
September 23, 2022, 06:34:51 PM

How to connect a button (via GPIO) to a K16A-B by pierlux
September 23, 2022, 05:44:49 PM

Background Image (House) Not Visible In XLights by dkulp
September 23, 2022, 12:41:52 PM

Powered by EzPortal
Powered by SMFPacks Menu Editor Mod