Light Pipes

ufacturer with a decent web site where you can actually pick what you want in an intelligent way.

m wide. I can't find anyone who provides info on their parts with the dimen sional data in the search. VCC is seriously bad and they don't actually hav e that many parts. I think I could create a selection guide for them that w ould put it all on one page!

I haven't done it yet, but was thinking about 3D printing some clear custom pattern. You can just design what you need with clear filament, then cove r the outside with metallic paint.

These Bivars are nice.

They look nice and create a wide viewing angle. They press into the panel. They would work with the board parallel or perpendicular to the panel.

Filament printers leave voids, and the voids will be dispersive, but if the path of the print head follows the path you want the lighte to take it might work like an optic fibre.

Having the guide cut from a sheet of acrylic might be a better bet but that limits you to two dimensions.

If you must print you may need to use a resin printer to get satisfactory results.

anufacturer with a decent web site where you can actually pick what you wan t in an intelligent way.

mm wide. I can't find anyone who provides info on their parts with the dim ensional data in the search. VCC is seriously bad and they don't actually h ave that many parts. I think I could create a selection guide for them that would put it all on one page!

om pattern. You can just design what you need with clear filament, then co ver the outside with metallic paint.

I think that would produce a fairly poor light pipe. They work by total in ternal reflection and very smooth surfaces. I don't think 3D printing can create surfaces anywhere near smooth enough and the light would be rather s cattered about even with the reflective coating.

But thanks for the idea.

I can get three color LEDs in smaller packages that will work with some of the light pipes I can find, but then I need 30 drivers for them all. The L ED I want to use has a driver chip built in that is loaded serially and the y can all be daisy chained requiring only one pin! But it's larger. Might be able to use one that spaces the pipes at 5.08 mm by skipping every othe r pipe. This still fits on the board if we need 9 indicators. But the spa ce under the pipe is not quite wide enough to center the LED package. Some room is required to prevent interference. I guess the question is how muc h. I've always used 10 mils (0.25 mm), but I know parts float around on th e pads. I guess 10 mils is more of a pad to pad spacing.

I've found two manufacturers who make the chip on board RGB LED. They both use the same protocol. But pumping the data out to all the LEDs over a si ngle pipe is a bit tedious for the CPU having to time 1200 ns bits to 80 nS resolution for some quarter of a millisecond might be a problem when inter rupts need to do something important.

I don't get why the protocol can't be interrupted for a while. They are ti ming the bit high time, but maybe they are using the bit cycle time to cali brate the time reference. No, that doesn't really make sense. Anyway, the re are two bit timing cycles and a reset is no data for 50us. So if the bi t low time were interrupted for a significantly less time than 50 us, it se ems the bit transmission could resume. I suppose a USART or SPI port could be used to push out bit patterns with less involvement from the CPU than b it banging. The timing is based on quarters of the bit time, so an 8 bit b yte shifted out at the correct speed would send 2 data bits. Likely the en tire sequence can be shifted out using DMA.

nufacturer with a decent web site where you can actually pick what you want in an intelligent way.

mm wide. I can't find anyone who provides info on their parts with the dim ensional data in the search. VCC is seriously bad and they don't actually have that many parts. I think I could create a selection guide for them th at would put it all on one page!

guess most calls for light pipes are from boards with their own front pane l attached to the board edge rather than the board being mounted parallel t o the panel. I suppose there is also the small box design with the board i n the bottom again perpendicular to the front panel on the end of the case.

Nice thought, but we can't put any holes in the front panel overlay. It is a sheet of plastic which is part of the outer shell to allow thorough clea ning of the device. But that, like much of this box, is likely to be retho ught by the manufacturer who picks this up. Even though the entire front p anel is one unpenetrated surface, the edges are not sealed, large openings for plumbing, etc.

So the light pipes need to be surface mounted with a length around 14 and 1

manufacturer with a decent web site where you can actually pick what you wa nt in an intelligent way.

stom pattern. You can just design what you need with clear filament, then cover the outside with metallic paint.

Interesting thoughts. They have some laser cut acrylic parts they use now. It seems like a PITA to laser cut custom light pipes. They would need to heat polish the cut sides too. Not sure how to mount them. I suppose we could use through hole fasteners, those things they call bolts! The ready made units have compliant press fit pins. I don't think we could make thos e. Maybe a square pin could be left and a die used to cut a thread on it.

manufacturer with a decent web site where you can actually pick what you w ant in an intelligent way.

stom pattern. You can just design what you need with clear filament, then c over the outside with metallic paint.

nternal reflection and very smooth surfaces. I don't think 3D printing can create surfaces anywhere near smooth enough and the light would be rather s cattered about even with reflective coating.

You can smooth out the surface with wet sanding, then buff is out with alco hol based solution. I have seen video of finished product that look like m oulded parts, rather than 3D printed. With external metallic paint, they a re good mirrors.

manufacturer with a decent web site where you can actually pick what you w ant in an intelligent way.

stom pattern. You can just design what you need with clear filament, then cover the outside with metallic paint.

internal reflection and very smooth surfaces. I don't think 3D printing ca n create surfaces anywhere near smooth enough and the light would be rather scattered about even with the reflective coating.

f the light pipes I can find, but then I need 30 drivers for them all. The LED I want to use has a driver chip built in that is loaded serially and t hey can all be daisy chained requiring only one pin! But it's larger. Mig ht be able to use one that spaces the pipes at 5.08 mm by skipping every ot her pipe. This still fits on the board if we need 9 indicators. But the s pace under the pipe is not quite wide enough to center the LED package. So me room is required to prevent interference. I guess the question is how m uch. I've always used 10 mils (0.25 mm), but I know parts float around on the pads. I guess 10 mils is more of a pad to pad spacing.

th use the same protocol. But pumping the data out to all the LEDs over a single pipe is a bit tedious for the CPU having to time 1200 ns bits to 80 nS resolution for some quarter of a millisecond might be a problem when int errupts need to do something important.

timing the bit high time, but maybe they are using the bit cycle time to ca librate the time reference. No, that doesn't really make sense. Anyway, t here are two bit timing cycles and a reset is no data for 50us. So if the bit low time were interrupted for a significantly less time than 50 us, it seems the bit transmission could resume. I suppose a USART or SPI port cou ld be used to push out bit patterns with less involvement from the CPU than bit banging. The timing is based on quarters of the bit time, so an 8 bit byte shifted out at the correct speed would send 2 data bits. Likely the entire sequence can be shifted out using DMA.

-get-to-know-them/ ?

pe manufacturer with a decent web site where you can actually pick what you want in an intelligent way.

is 5 mm wide. I can't find anyone who provides info on their parts with the dimensional data in the search. VCC is seriously bad and they don't actual ly have that many parts. I think I could create a selection guide for them that would put it all on one page!

custom pattern. You can just design what you need with clear filament, then cover the outside with metallic paint.

internal reflection and very smooth surfaces. I don't think 3D printing ca n create surfaces anywhere near smooth enough and the light would be rather scattered about even with reflective coating.

cohol based solution. I have seen video of finished product that look like moulded parts, rather than 3D printed. With external metallic paint, they are good mirrors.

I guess you could do that, but we need a commercial product. They want thi s to be built in large quantities.

pe manufacturer with a decent web site where you can actually pick what you want in an intelligent way.

is 5 mm wide. I can't find anyone who provides info on their parts with the dimensional data in the search. VCC is seriously bad and they don't actual ly have that many parts. I think I could create a selection guide for them that would put it all on one page!

custom pattern. You can just design what you need with clear filament, the n cover the outside with metallic paint.

l internal reflection and very smooth surfaces. I don't think 3D printing can create surfaces anywhere near smooth enough and the light would be rath er scattered about even with the reflective coating.

of the light pipes I can find, but then I need 30 drivers for them all. T he LED I want to use has a driver chip built in that is loaded serially and they can all be daisy chained requiring only one pin! But it's larger. M ight be able to use one that spaces the pipes at 5.08 mm by skipping every other pipe. This still fits on the board if we need 9 indicators. But the space under the pipe is not quite wide enough to center the LED package. Some room is required to prevent interference. I guess the question is how much. I've always used 10 mils (0.25 mm), but I know parts float around o n the pads. I guess 10 mils is more of a pad to pad spacing.

both use the same protocol. But pumping the data out to all the LEDs over a single pipe is a bit tedious for the CPU having to time 1200 ns bits to 8

e timing the bit high time, but maybe they are using the bit cycle time to calibrate the time reference. No, that doesn't really make sense. Anyway, there are two bit timing cycles and a reset is no data for 50us. So if th e bit low time were interrupted for a significantly less time than 50 us, i t seems the bit transmission could resume. I suppose a USART or SPI port c ould be used to push out bit patterns with less involvement from the CPU th an bit banging. The timing is based on quarters of the bit time, so an 8 b it byte shifted out at the correct speed would send 2 data bits. Likely th e entire sequence can be shifted out using DMA.

ou-get-to-know-them/ ?

Interesting. These are a different manufacturer, Worldsemi vs. Kingbright and Lumex who seem to be interchangable. The WS2812B from Worldsemi has sl ightly different timing specs with tighter tolerances, but for what this gu y was showing is that the tolerances are just data sheet numbers.

The other two brands are sold through distribution and are much lower in pr ice when dealing with quantities.

You seem to always find pertinent and interesting links. Is it your google

-foo or have you worked with this before?

A:

Raise the neopixels closer to the panel. (use a daughter board suported by the connectors)

B:

LP5030 needs two pins (fewer if you already have an IIC bus)

Or cut hooks that go through holes or slots in the PCB.

O=acrylic or polycarbonate |###| |###| | | | | OOOOOOOOOOOOOOOOOOOOOOOOOOO OO###| O O###| O |###|NNNNNNN O |###|NNNNNNN O |###|NNNNNNN-_ OOOOOOOOOOOOOOOOOOOO |###|NNNNNNN \ OOOOOOOOOOOOOOOOOOOO |###|NNNNNNN \ OOOOOOOOOOOOOOOOOOO |###|NNNNNNN |OOOOOOOOOOOOOOOOOOO |###|NNNNNNN |OOOOOOOOOOOOOOOOOOO |###|NNNNNNN / OOOOOOOOOOOOOOOOOOO |###|NNNNNNN _/ OOOOOOOOOOOOOOOOOOOO |###|NNNNNNN- OOOOOOOOOOOOOOOOOOOO |###|NNNNNNN O |###|NNNNNNN O O###| O OO###| O OOOOOOOOOOOOOOOOOOOOOOOOOOO | | | | |###| NOT TO SCALE |###|

pipe manufacturer with a decent web site where you can actually pick what y ou want in an intelligent way.

e is 5 mm wide. I can't find anyone who provides info on their parts with t he dimensional data in the search. VCC is seriously bad and they don't actu ally have that many parts. I think I could create a selection guide for the m that would put it all on one page!

r custom pattern. You can just design what you need with clear filament, t hen cover the outside with metallic paint.

tal internal reflection and very smooth surfaces. I don't think 3D printin g can create surfaces anywhere near smooth enough and the light would be ra ther scattered about even with the reflective coating.

me of the light pipes I can find, but then I need 30 drivers for them all. The LED I want to use has a driver chip built in that is loaded serially a nd they can all be daisy chained requiring only one pin! But it's larger. Might be able to use one that spaces the pipes at 5.08 mm by skipping ever y other pipe. This still fits on the board if we need 9 indicators. But t he space under the pipe is not quite wide enough to center the LED package. Some room is required to prevent interference. I guess the question is h ow much. I've always used 10 mils (0.25 mm), but I know parts float around on the pads. I guess 10 mils is more of a pad to pad spacing.

y both use the same protocol. But pumping the data out to all the LEDs ove r a single pipe is a bit tedious for the CPU having to time 1200 ns bits to 80 nS resolution for some quarter of a millisecond might be a problem when interrupts need to do something important.

are timing the bit high time, but maybe they are using the bit cycle time t o calibrate the time reference. No, that doesn't really make sense. Anywa y, there are two bit timing cycles and a reset is no data for 50us. So if the bit low time were interrupted for a significantly less time than 50 us, it seems the bit transmission could resume. I suppose a USART or SPI port could be used to push out bit patterns with less involvement from the CPU than bit banging. The timing is based on quarters of the bit time, so an 8 bit byte shifted out at the correct speed would send 2 data bits. Likely the entire sequence can be shifted out using DMA.

-you-get-to-know-them/ ?

t and Lumex who seem to be interchangable. The WS2812B from Worldsemi has slightly different timing specs with tighter tolerances, but for what this guy was showing is that the tolerances are just data sheet numbers.

price when dealing with quantities.

le-foo or have you worked with this before?

mostly google-foo, I looked at using them some time ago but never got aroun d to it

e:

t pipe manufacturer with a decent web site where you can actually pick what you want in an intelligent way.

age is 5 mm wide. I can't find anyone who provides info on their parts with the dimensional data in the search. VCC is seriously bad and they don't ac tually have that many parts. I think I could create a selection guide for t hem that would put it all on one page!

ear custom pattern. You can just design what you need with clear filament, then cover the outside with metallic paint.

total internal reflection and very smooth surfaces. I don't think 3D print ing can create surfaces anywhere near smooth enough and the light would be rather scattered about even with the reflective coating.

some of the light pipes I can find, but then I need 30 drivers for them all . The LED I want to use has a driver chip built in that is loaded serially and they can all be daisy chained requiring only one pin! But it's larger . Might be able to use one that spaces the pipes at 5.08 mm by skipping ev ery other pipe. This still fits on the board if we need 9 indicators. But the space under the pipe is not quite wide enough to center the LED packag e. Some room is required to prevent interference. I guess the question is how much. I've always used 10 mils (0.25 mm), but I know parts float arou nd on the pads. I guess 10 mils is more of a pad to pad spacing.

hey both use the same protocol. But pumping the data out to all the LEDs o ver a single pipe is a bit tedious for the CPU having to time 1200 ns bits to 80 nS resolution for some quarter of a millisecond might be a problem wh en interrupts need to do something important.

y are timing the bit high time, but maybe they are using the bit cycle time to calibrate the time reference. No, that doesn't really make sense. Any way, there are two bit timing cycles and a reset is no data for 50us. So i f the bit low time were interrupted for a significantly less time than 50 u s, it seems the bit transmission could resume. I suppose a USART or SPI po rt could be used to push out bit patterns with less involvement from the CP U than bit banging. The timing is based on quarters of the bit time, so an 8 bit byte shifted out at the correct speed would send 2 data bits. Likel y the entire sequence can be shifted out using DMA.

ce-you-get-to-know-them/ ?

ght and Lumex who seem to be interchangable. The WS2812B from Worldsemi ha s slightly different timing specs with tighter tolerances, but for what thi s guy was showing is that the tolerances are just data sheet numbers.

n price when dealing with quantities.

ogle-foo or have you worked with this before?

und

I'd really like to use them for this project. Needing just one line to dri ve 30 LEDs is a big advantage without using extra chips. I just need to pi pe the light to the front panel a half inch away.

I can use this light pipe, but the chip will need to be enough off center t o allow for manufacturing tolerances and soldering movement. At some point that will impact the optical connection to the light pipe.

Anyone know what to expect for component misalignment? The part is 5x5 mm with six pads on two sides, not QFN, more like a J lead.

te:

ote:

ght pipe manufacturer with a decent web site where you can actually pick wh at you want in an intelligent way.

ckage is 5 mm wide. I can't find anyone who provides info on their parts wi th the dimensional data in the search. VCC is seriously bad and they don't actually have that many parts. I think I could create a selection guide for them that would put it all on one page!

clear custom pattern. You can just design what you need with clear filamen t, then cover the outside with metallic paint.

y total internal reflection and very smooth surfaces. I don't think 3D pri nting can create surfaces anywhere near smooth enough and the light would b e rather scattered about even with the reflective coating.

h some of the light pipes I can find, but then I need 30 drivers for them a ll. The LED I want to use has a driver chip built in that is loaded serial ly and they can all be daisy chained requiring only one pin! But it's larg er. Might be able to use one that spaces the pipes at 5.08 mm by skipping every other pipe. This still fits on the board if we need 9 indicators. B ut the space under the pipe is not quite wide enough to center the LED pack age. Some room is required to prevent interference. I guess the question is how much. I've always used 10 mils (0.25 mm), but I know parts float ar ound on the pads. I guess 10 mils is more of a pad to pad spacing.

They both use the same protocol. But pumping the data out to all the LEDs over a single pipe is a bit tedious for the CPU having to time 1200 ns bit s to 80 nS resolution for some quarter of a millisecond might be a problem when interrupts need to do something important.

hey are timing the bit high time, but maybe they are using the bit cycle ti me to calibrate the time reference. No, that doesn't really make sense. A nyway, there are two bit timing cycles and a reset is no data for 50us. So if the bit low time were interrupted for a significantly less time than 50 us, it seems the bit transmission could resume. I suppose a USART or SPI port could be used to push out bit patterns with less involvement from the CPU than bit banging. The timing is based on quarters of the bit time, so an 8 bit byte shifted out at the correct speed would send 2 data bits. Lik ely the entire sequence can be shifted out using DMA.

right and Lumex who seem to be interchangable. The WS2812B from Worldsemi has slightly different timing specs with tighter tolerances, but for what t his guy was showing is that the tolerances are just data sheet numbers.

in price when dealing with quantities.

google-foo or have you worked with this before?

round

rive 30 LEDs is a big advantage without using extra chips. I just need to pipe the light to the front panel a half inch away.

to allow for manufacturing tolerances and soldering movement. At some poi nt that will impact the optical connection to the light pipe.

m with six pads on two sides, not QFN, more like a J lead.

It is looking less likely I will need the 5x5 LED with the built in driver. I was mistaken and most of the LED positions only need a single color wit h one needing two and two others needing three colors. So that is only 13 I/Os. Since we are adding an FPGA to handle the alarm conditions we will h ave enough I/Os I'm pretty sure. I might need to roll an ADC converter to directly measure analog inputs. That or add a multiple input ADC.

I was surprised to learn they use musical notes for audible alarms. That will be fun to do in the FPGA.

ufacturer with a decent web site where you can actually pick what you want in an intelligent way.

m wide. I can't find anyone who provides info on their parts with the dimen sional data in the search. VCC is seriously bad and they don't actually hav e that many parts. I think I could create a selection guide for them that w ould put it all on one page!

uess most calls for light pipes are from boards with their own front panel attached to the board edge rather than the board being mounted parallel to the panel. I suppose there is also the small box design with the board in t he bottom again perpendicular to the front panel on the end of the case.

Element14 (Newark in the US?) have a subset of their range available.

Importing from EU to US if you see soemthing not available locally may be a bit time consuming ...

What's strange is that light pipes cost more, sometimes outrageously more, than LEDs.

anufacturer with a decent web site where you can actually pick what you wan t in an intelligent way.

mm wide. I can't find anyone who provides info on their parts with the dim ensional data in the search. VCC is seriously bad and they don't actually h ave that many parts. I think I could create a selection guide for them that would put it all on one page!

guess most calls for light pipes are from boards with their own front pane l attached to the board edge rather than the board being mounted parallel t o the panel. I suppose there is also the small box design with the board in the bottom again perpendicular to the front panel on the end of the case.

s/

a bit time consuming ...

Those are the parts we are looking at, but they don't have room for the 5x5 .5 mm serially configured RGB LEDs I was looking at using. We have shifted gears yet again and will be using an FPGA to control the alarms so we have a bunch more pins now and regular tricolor LEDs will fit under them fine.

We do need to enlarge the board by 1 mm though! lol. They are too close t o the LCD modules. I'd like to see the board grow by a bit more, but there 's a flow and oxygen sensor in the way. Maybe we can get 2 mm, lol.

manufacturer with a decent web site where you can actually pick what you wa nt in an intelligent way.

I guess most calls for light pipes are from boards with their own front pan el attached to the board edge rather than the board being mounted parallel to the panel. I suppose there is also the small box design with the board i n the bottom again perpendicular to the front panel on the end of the case.

es/

e a bit time consuming ...

Even weirder, is that they seem to cost nearly the same if they are 1 LED o r up to 10 ganged LED light pipes. I guess it's the fact that they are usi ng the machine and they can pretty much make as many 10 unit pipes per minu te as they can 1 unit pipes.

I have no idea how these things are made optically correct... mostly. I se e some units where a flat side is bowed in dramatically from the shrinking of the plastic as it what? cools? Hardens? I'm mot even sure if they mol d them with heat or make the plastic in the mold. I seem to recall the pro cess of making acrylic is very simple and can be done in a beaker, then you break the beaker to free the plastic.