JNSF DIY lighting project (PCB based)

[cteno4]

Yes charging to 14v gives you almost 3x the saved energy than being charged to 5v. Also for a lot of the discharge time the 14v cap side will provide you will full 5v to the led side whereas with the cap on the 5v side it’s going to only be in the decent voltage range for the leds for a bit of that time.

 

jeff

[chadbag]

2 minutes ago, cteno4 said:

Yes charging to 14v gives you almost 3x the saved energy than being charged to 5v. Also for a lot of the discharge time the 14v cap side will provide you will full 5v to the led side whereas with the cap on the 5v side it’s going to only be in the decent voltage range for the leds for a bit of that time.

 

 

Yes, that was the goal of doing the V2 -- get the full 5V for longer and have more stored energy.  But it seems to last for a lot longer than 3x. I am not complaining.  I'll use the V1 boards as they work, and if anyone wants to experiment with them they can have some for cost, but for most folks I'd wait for the V2 boards.

 

[chadbag]

I am now going to  populate a V1 board to test in-wagon using the hot plate method and the smaller, lie flat voltage regulators.

 

Wish me luck.

 

[gavino200]

1 hour ago, chadbag said:

 

You can just send me 1 for testing of 3.3V.  Can you post a pic so I see what form factor they are in?

 

 

Here it is. If you send me your address, I'll put one in the mail tomorrow. Though you might want to check your tracking number first if you have one. I think I ordered mine at essentially the same time as you, and mine just came today.

 

[chadbag]

OK, I'll go look up that order and see where it is at.  I got some LEDs today but nothing else.

[chadbag]

I put a V1 board together tonight with 13 LEDs, breaking off the last segment so it would fit in my Tomix Hikari Rail Star 700.  

 

I used 7x 100uF, 1x 33uF, 1x 22uF (need to use up the smaller value ones) for a nominal 755 uF.  I don't remember the tolerance of these parts but this is probably +- 10-20%.  I used a 75 ohm inrush protection resistor before the caps and a 2,2k ohm dimming resistor.  I used 1206 warm white LEDs.  I used my low temp solder paste and the hot plate method.  I think my hot plate got too hot but it worked.  I bought one of those infrared thermometers to test it with but I think it was still heating or something as I checked it afterwards and it was much higher, even though I had let it warm up for a bit.  The hot plate is also pretty uneven.

 

I did the voltage regulator separately by hand as I wanted to check some stuff and make sure there was no short (it has a ground plate sticking up on the top that should be bent up to make sure it does not short with one of the other voltage regulator spots.

 

One LED got bumped and was crooked.  I used the hot air rework station to loosen it up and put it in its place.   I tried the board fed from DCC track power and it worked first time.

 

This shows the parts on the board with the paste.  Not yet heated

 

This shows the board after being heated and the parts are now soldered on, but the board is not cleaned up yet.   You can see the crooked LED.

 

You get all these little solder balls that form around all the components and need to be removed.  You can see some around the LED D14.  I use my xacto knife blade point to scrape the balls away.

 

Worked first time.  Yay!

 

This shows the "production" assembled board on the top (with the last segment already broken off) and the hand assembled test board on the bottom.

 

I took one of the passenger units from the Tomix Hikari Rail Star 700 and installed the board and a LaisDCC function decoder.  I used a bit of double sided tape to attach it to the wagon ceiling.   Here are some pics of the car with the board (in a lighted room and a dark room).   In the shots with 2 lighted wagons the left unit uses an LED strip and a hand wired test V2 circuit (I forget but around 300-400 uF IIRC) and the one on the right uses the V1 board (755 uF).   The LEDs every 10mm make for good light coverage.  I've used no diffuser paper or anything on either one.  They don't look as bright in real life as compared to the photos and I think both look pretty good.

 

 

I've run it around in a 3 unit train (cab with headlight, motor car (with the LED strip proto hard wired V2 lights) and the 3rd car with the V1 board.   My track has never been cleaned and you see flickering with the headllight but though I saw quick momentary dimming of the lights slightly, once in a while, the lights never flickered.   The V1 showed a little more partial dimming than did the proto V2, which I don't remember seeing dim at all, or barely.  But even the V1 board was acceptable -- a slight dimming is much better than on/off flicker.

 

 

[chadbag]

The only "problem" with the V2 board is it requires caps with higher voltage ratings.  But I've found that the caps of a given capacity with the 16V or 25V rating are often less expensive than the same size (physical and uF rating) with the 6.3V or 10V rating.   I plan on adding a 12V zener diode to the circuit so we can use 16V caps safely as the small ceramics are hard to find in 100uF at anything better than 16V.  I think I found some 25V at one place and they were pretty expensive -- need to go look that order up.  But the 16V ones are easier to find and about 4-6 cents each.   5-6 of them on a V2 board is probably about right.   (The V1 board also requires more capacitance to be relatively acceptable so you need more on the V1).

 

 

[gavino200]

8 hours ago, chadbag said:

The only "problem" with the V2 board is it requires caps with higher voltage ratings.

 

Is it a 100uF 25V cap that you're placing upstream? Still fits no problem? 

 

It might be an idea to try it in an old fashioned passenger coach. They can be less forgiving with vertical height obstacles than shinkansen cars. It's often the "room dividers" and walls that cause problems.

 

 

[gavino200]

9 hours ago, chadbag said:

I put a V1 board together tonight with 13 LEDs, breaking off the last segment so it would fit in my Tomix Hikari Rail Star 700.  

 

 

 

 

 

 

 

 

 

 

Outstanding !!!

[chadbag]

37 minutes ago, gavino200 said:

 

Is it a 100uF 25V cap that you're placing upstream? Still fits no problem? 

 

It is not a size issue but an availability issue on the 100 uF in our size (or something similar like 1210).   For the most part, 16V is the highest voltage rating on small ceramic smd caps.   I think I did find a small number of 25V parts in that range.  

 

37 minutes ago, gavino200 said:

 

It might be an idea to try it in an old fashioned passenger coach. They can be less forgiving with vertical height obstacles than shinkansen cars. It's often the "room dividers" and walls that cause problems.

 

 

 

Yeah. I did the Shinkansen 700 this time as it was what I was otherwise working on anyway.  I don’t have a lot of Japanese passenger coaches and my Euro ones have other issues (like power pickup) to solve.  

 

I think any wagon made to take a factory board of some sort can take ours.  They are pretty thin when completed.   I am thinking about using a thinner PCB even though there is an additional engineering charge as the shipping cost goes down which can make up for it in  quantity.  Though the current 0.8mm is pretty thin.

 

I want to make the V2 board less wide.  

[chadbag]

Put together another V1 board.  This time using 9 0805 warm white LEDs.  Same capacitance and everything as the other board, but since this is to be in a cab car of the Shinkansen, it has to be shorter.  I thought I'd try the 0805 sized LEDs.   The process of assembling the boards is not as bad as hand soldering stuff but still takes a while.  I am sure with practice I can do it faster.  I'll be doing another one tonight for the other end cab car.

 

Here is a pic.  I have not yet broken off the end I don't need.  This one worked the first time as well.  The second pic shows how thin this is with these components.

 

Edited March 31, 2019 by chadbag

[chadbag]

Today I made up two more V1 boards for the two cab end cars of my Tomix Shinkansen 700 Hikari Rail Star.  The passenger compartment on the end cabs is smaller than a full car so I used the 9 LED version of the board, snipping off all the optional LED spaces.  It is late so I will actually install them another day, but both work.  The second one did not work on first test.  

 

I always take a meter and check for shorts across each component (except the chips).  On the 2.2K resistor I was not getting anything, so I removed it, checked the resistor, and then resoldered.  Then testing with the meter I got the right value across it, so maybe there was a bad solder joint.  But it still didn't work.  So I started measuring voltages and right away on the inputs to the rectifier I did not measure any voltage, so I removed that chip.  There was a layer of oxidation on each pad so when the heat melted the solder paste, the legs must not have been pressed well against the pads and it must have oxidized the solder paste (and the carrier in the paste) and created bad joints.  I scraped each pad, put more paste on, and used the air gun reworking station to solder it in place.  It worked then.   I will have to change the "foot print" in the PCB layout software to have bigger pads on that form factor for "hand soldering".  The "footprint" I used was the one used for normal automated soldering (while all the resistor, cap, and LED ones were designated for "hand soldering" and had enlarged pads.   Larger pads help with this way of soldering SMDs by hand (or hot plate).

 

I measured the time it took to do the last one.   50 min (not including trouble shooting it not working the first time).  This is from putting the blank board down on my work table to testing it on the track with alligator clips.   Some of that time was taking pictures, and some of it was because I was not organized and had to find parts in my stash, and some is I got distracted by the forum (just a couple minutes worth) and checking email.  So maybe it should have been 42-45 min.  I need to do it faster and it will get faster as I do this more.  It took about 4 mins to put the paste on.  About 30 mins to place the components (including find them, open, etc and I lost an LED and had to go get one more).  And about 15 mins to heat/solder, inspect, get rid of all the solder balls etc, rework a few LEDs that slid a little and a capacitor that got misplaced, and then testing.   The time to trouble shoot and redo the resistor and chip I did not count as usually they boards work 🙂

 

Here are some pics.  The first shows it on the hot plate.  The second shows it after the hot plate, before reworking any components that moved out of place, and before cleaning up all the little solder balls and other things.  The last shows both boards I did today.   All values across both boards are the same even if some of the components look different (caps in particular).  I used different sized caps on the end though they had the same value -- I grabbed different versions of the same valued cap from my stash.

 

[chadbag]

So, quick electronics for dummies type question:

 

On the V1 board I wired all the gnd ends together IN THE ORDER THEY ARE IN ON THE SCHEMATIC.  So it was probably needlessly complicated.  I assume that as long as all the GND ends are connected together, it really doesn't matter which order they are in (unlike the POS side of the circuit).   If you look at the sample V2 circuit I posted a while back in this thread, the POS is basically the top, it goes through the last LED, and the GND / NEG is the return leg.  Do I need to put those in order or can I just make sure they are all connected together?   I think the latter is correct but wanted to make sure.

 

Right now I have the V2 layed out, nice and compactly (and upgraded to 21 cap positions so you can use smaller capacity caps and still get a decent total amount -- just leave empty the ones you don't need).  However, I did lay out the GND ends in order, which complicates it.  I think  can rewire the GND end to just make sure they are all connected.

 

[Sheffie]

The wires and connections should have zero effective resistance. Therefore, it shouldn't matter which wires you connect to which, as long as all points that need to be connected to GND are in fact connected together.

[chadbag]

14 hours ago, Sheffie said:

The wires and connections should have zero effective resistance. Therefore, it shouldn't matter which wires you connect to which, as long as all points that need to be connected to GND are in fact connected together.

 

Thanks, I figured that was the case, which is why I asked, but I wanted to make sure since there are lots of non-intuitive things for non-SME in electronics. 🙂

[chadbag]

So I installed one of the V1 boards I did for the 700 cab cars.  They use 0805 LEDs instead of 1206.  With the same resistor values, they seem to not be as bright or cast as much light (which may be expected, given they are smaller).  I may play around with the resistor values for 0805 LEDs.

 

In the pic, the leftmost has the handwired circuit (V2) using LED strip with 1206.  The middle car is a V1 board with 1206.  The right cab car is a V1 board with 0805 LEDs.  All are "warm white" supposedly.  It is hard to tell but the 0805 looks "whiter" than the other two.   May be a function of the smaller LEDs casting less light combined with the internal passenger compartment insert being made of blue plastic, as it definitely was from the "warm white" LED stash.

 

[cteno4]

In the photo the cab car looks more realistic and the right lighting, but digital camera will make a different look thanthe eye!

 

cheers

 

jeff

[Martijn Meerts]

I prefer the cab car as well, the other 2 seem too bright. Obviously, camera vs eye will be a bit different, but both the amount of light and the colour seems better in the cab car.

 

[chadbag]

In real life they are both not as bright as in the pic.  

 

Stand alone, both of them look good.  Next to each other one looks off.   

[cteno4]

That’s the issue with lighting as what looks good in the camera might not look good to the eye. I use to be amazed when the lighting guy would light models and it never looked great looking at the model and then wow looking thru the camera and even more wow on the film and neat additional effects depending on what film used. That’s changed with digital some.

 

jeff

[cteno4]

Moderator note: the discussion of the dcc version was split off on its own thread

jeff

[chadbag]

So I put the other board in the other end cab car on the 700.  I replaced the 2.2k with a 1.2k to get the 0805 LEDs to be somewhat brighter to equal the 1206 LEDs with a 2.2k dimming resistor.

 

The cab car on the left is the 0805 warm white LEDs with 1.2k dimming resistor.  The next car is a hand wired prototype V2 with 2.2K warm white 1206 LEDs, then the next car has the V1 board with the 1206 LEDs, warm white, and 2.2K.  Then the end car with the 0805 warm white LEDs with the 2.2K.

 

These look brighter in the photo than in real life.

 

[chadbag]

Tonight I built a V1 board using 1206 warm white, like the one I already installed in a 700 wagon, but this time I used the 3.3V version of the regulator.  With the 3.3V, the 2.2k dimming resistor is too much and it is very faint.  To approximately match the 5v regulator with the 2.2k dimming resistor I went with a 1.2k resistor. On the 3.3V board  I tried 1.8k first but it looked a bit dim still when compared to the existing 5V board so I went to the 1.2k.  I have not installed it in a wagon yet but with room lights off, putting it next to the wagon I was able to do a rough comparison.    I'll install it tomorrow night and double check.

 

No issues with the 3.3v regulator.

 

Instead of an MB6F (or MB6S) rectifier, I used an MB1S rectifier.  That makes no difference but I have 20 of them so wanted to try one to make sure.  The MB1S is the same form factor as the MB6S or MB6F (the F model is low profile -- same specs) and is rated for 100V while the MB6S or F is rated at 600V.

 

[cteno4]

Cool. Yep 1.7v less to drop with the dimming resistor. Caps should last longer as well if on the outside of the voltage regulator.

 

same voltage drop on the rectifier with the two different limit voltages.

 

jeff 

[Ochanomizu]

Hello Mr chadbag,

 

How long does it take you to build one board?