JNSF DIY lighting project (PCB based)

[chadbag]

So, I could not wait until tomorrow, so I hand soldered one of these together.  I used the LM7805CV 5v regulator so as to not use up one of the better choice ones (smaller etc) since this one works just as well to test the circuit with.  I used 75ohm for the small resistor leading in to the caps and 2.2k for the "dimming" resister leading into the LEDs.    I used one 220uF tantalum (since I have them) and 5x  22uF ceramic caps.  That is a face value of 330uF (plus or minus whatever the tolerance is  -- most of these are 5-10% I think though I have some 20% ones in the stash.  I used bright white (not warm) 1206 LEDs.  I only used 9 LEDs.  I filled it up to where the snap-off pieces start (so the main part is full but none of the extension pieces have LEDs).

 

I did this with the hand soldering iron.  Big pain in the tush.   I did buy a hot plate and already have the solder paste (and a solder air gun) so will do some more test ones this weekend using the solder paste and the hot plate method.  I suspect it will work much better 🙂

 

I will test fit some of the boards into various wagons, EMU, shinkansen, etc this weekend.   Trying to get info to use on the V2 board.  I also plan on mocking up both the V1 and V2 circuits in the breadboard to test the various configs.

 

Here is what it looks like assembled and then lit (using DCC track power into the rectifier bridge).

 

[Martijn Meerts]

Will be interesting to see the brightness when installed in a coach. The one thing that usually bothers me the most with regular interior light strips, is that they’re always too bright. 

[chadbag]

That is what the dimming resistor is for.   You can tune it for the brightness you want.   

[Martijn Meerts]

Yeah, but LEDs have a limit to how much they can be dimmed. Will also be interesting to see how this works with DCC and its dimming functionality, the capacitors might cause issues there. 

 

But, all things worth experimenting with 🙂

[cteno4]

The main limiting resistor can drop the leds as much as your want to to a dim glow. Caps only replace the track current when that is lost.

 

jeff

[Martijn Meerts]

Most LEDs have a minimum brightness at their minimum current, so you won't be able to dim them beyond that really. 

 

Thing with DCC dimming, is that it uses PWM to dim the LEDs, but if you have a capacitor to bridge loss of power, will PWM still work? I should actually be able to test this with me Lenz 0-gauge stuff, they have a large capacitor in them that can bridge power less for several seconds, but then again, those are also custom decoders so they might have additional circuitry for dimming the LEDs.

 

[cteno4]

but Most of this won’t be dcc decoder pwm, but dc or dcc ac rectifier to dc.

 

jeff

[Martijn Meerts]

Not in my case, I eventually want all my lights to be controlled by a decoder 😄

 

I do realise I'm not the standard here of course, especially not for Japanese layouts where DCC isn't exactly the most common thing 😉

 

 

[chadbag]

The anti-flicker duration, using 330uF, is not really sufficient on the test board with 9 LEDs.   I have a hard wired sample circuit using the same circuit, that uses 770uF with 2 LEDs and it has much better duration.  I am going to add some more capacitance to the test board to see what we can come up with.

[chadbag]

I am working on an alternative V2 form -- using the same basic circuit but with no LEDs.  A small board you attach to existing lights, or to an LED strip, to provide anti-flicker (or in the case of the LED strip, to use that as the LEDs).  There will be two physical forms, one more tuned to LED strips, and one with just a small board.  This is so if you have existing lights, or a wagon or cab end car or something where the normal board does not fit, you can put together something using LED strips (for those special cases).

 

This alternative will only be set up for a specific pinout voltage regulator etc., fewer cap spaces, etc. to make it smaller.

 

[chadbag]

Here is a video I made of the V1 circuit mocked up on a breadboard, showing the full 14 LED capacity, using a 2.2K dimming resistor, and various levels of capacitance for the anti-flicker.  At the end I show my prototype board I have made up as well.  More tests need to be done, but I think for myself, I will aim for somewhere between 600 and 1000 uF on a board for normal use.

 

I am now mocking up a V2 circuit on the breadboard.

 

 

[gavino200]

Great work.

 

I have a question about the physical location and orientation of the voltage regulator on the LED strip. Can it be soldered in a flat orientation? It looks like this is sticking up quite a bit. Am I missing something?

[chadbag]

5 minutes ago, gavino200 said:

Great work.

 

I have a question about the physical location and orientation of the voltage regulator on the LED strip. Can it be soldered in a flat orientation? It looks like this is sticking up quite a bit. Am I missing something?

 

No, you are not missing something.  I designed the board to take 3 different form factors of voltage regulators, two of which are small and lie flat.  I plan on using the middle size ones (which are much smaller than the ones used in the board/circuits I have shown so far) as they are available for pennies a piece from Aliexpress in 5v and 3.3v varieties (I have a few hundred of each ordered).  I ordered 20 of the 5V version on eBay from a US seller for test purposes.  You can see them below.  

 

I am using the larger voltage regulators in my tests since they are test pieces and so the size does not matter and so I use up my supply.  You can bend them to lie parallel to the board (see the second pic below) but you block an LED.  They can't touch the board as the rectifier etc lie there. If you dremel off the top of the heat sink you can let it stick up perpendicular to the board and it should fit in those wagons that have a work room or something on the end without lots of windows where the chip wouldn't be noticed (we are not pulling any serious current through and I have touched the heat sunk on a sample circuit and it was cold so I don't thin we need that bit of a heat sunk).  

 

But in general I would use the voltage regulators of the medium size.  I plan on soldering up another board tonight using one.  The smallest form factor voltage regulator I made compatible are not that cheap from what I have seen so I don't plan on using them much myself unless I can find a cheap source.

 

[chadbag]

Here is a video of the V2 circuit mocked up on the breadboard.  The LEDs are brighter using the same dimming resistor value -- probably because of the side of the voltage regulator it is on.  I am going to play with the inrush current limiting resistor after the voltage regulator to see if I can get that to dim more like the V1 levels, which I think are pretty good.

 

Again, 14 LEDs were used, and the 2,2K resistor is the same.  This time, half bright white and half warm white.  Their specs should be very close to the same.   I ran out of bright white (or at least my stash is missing).

 

Here, the lights stay on a longer using the same level of capacitance.   The V2 circuit should require less capacitance to be acceptable.

 

 

 

[gavino200]

V2 looks like a win. What exactly is the difference between V1 and V2?

 

Are you going to run tests with 3.3V regulators?

[chadbag]

42 minutes ago, gavino200 said:

V2 looks like a win. What exactly is the difference between V1 and V2?

 

Are you going to run tests with 3.3V regulators?

 

The main difference between V1 and V2 is that

 

on V1 we have rectifier --> voltage regulator --> caps --> dimming resistor --> LEDs  (in simplified form).  

 

In V2 we have  rectifier --> caps --> dimming resistor --> voltage regulator --> (2nd dimming resistor?) --> LEDs

 

I am playing around with the second dimming resistor (current V2 has a 100 ohm in-rush protection resistor there -- playing with replacing that with a higher value resistor to act as both dimming and inrush protection).  The reason the main dimming resistor is on the Cap side is that it works together with the Caps to create the RC time constant circuit which is what gives us the capacitor capability to feed the LEDs over time.

 

I am not an electronics guru but I came up with V2 after playing with V1 some(in my test hard wired one) and realized that the capacitors were only charging to the 5V regulated power level and discharging from that.  I figured if we put the capacitors on the other side of the voltage regulator, we'd charge the capacitors to the full 9-13V or so from your DC/DCC/decoder output and have that whole voltage to discharge into the LEDs, with the voltage regulator converting that to our 5V, so we get a steadier, less drop off on the LEDs.  

 

I did not think the difference would be as great as it is.  See the comparison video below (I only used 470uF and 680uF in the comparison video -- did not try the 1000uF since that is a stretch to achieve on the V1 board anyway).

 

 

 

Edited March 29, 2019 by chadbag

[chadbag]

16 minutes ago, gavino200 said:

Are you going to run tests with 3.3V regulators?

 

I don't have any 3.3V regulators right now.  I have a ton on order and expect the lights to be similar to the 5V version.  Maybe adjust some resistors or something.  I am not going to hold up production of a test run of V1 boards, once ready, waiting for the 3,3V regulators to get here from the slow boat from China.

[gavino200]

That's a very impressive test.

 

Both my voltage regulators arrived today. I could send you a few if I can work out which is which. They look identical.

[chadbag]

12 minutes ago, gavino200 said:

That's a very impressive test.

 

Both my voltage regulators arrived today. I could send you a few if I can work out which is which. They look identical.

 

They probably have some minor variance in markings.  Can you post a picture with them?

 

[gavino200]

4 minutes ago, chadbag said:

 

They probably have some minor variance in markings.  Can you post a picture with them?

 

 

I worked it out. Just had to go get my loupes. How many do you need? 

[chadbag]

So I did a couple tests on the V2 with resistor values.   I'll need to do more.  Using 2x 2.2k (one on the cap side and one as an inrush limiter on the LEDs) provided a similar dimnes to the 2,2k on the V1.  I also tried 1K instead of 100 ohm (leaving the 2.2k on the cap side) and also swapping the 100 ohm and the 2.2k ohm.

 

Definitely need more testing for optimal values.

 

[cteno4]

27 minutes ago, chadbag said:

I am not an electronics guru but I came up with V2 after playing with V1 some(in my test hard wired one) and realized that the capacitors were only charging to the 5V regulated power level and discharging from that.  I figured if we put the capacitors on the other side of the voltage regulator, we'd charge the capacitors to the full 9-13V or so from your DC/DCC/decoder output and have that whole voltage to discharge into the LEDs, with the voltage regulator converting that to our 5V, so we get a steadier, less drop off on the LEDs.  

 

I did not think the difference would be as great as it is.  See the comparison video below (I only used 470uF and 68-uF in the comparison video -- did not try the 1000uF since that is a stretch to achieve on the V1 board anyway).

 

Yep that right having the caps charge to higher voltage will provide more energy to keep the led lit. The only reason to put the cap after the voltage regulator was to use lower voltage caps. Also the caps after the voltage regulator are only charged to the 5 or 3.3v and will drop below effective led voltage faster. 

 

My guess is the best shoud be using the 3.3v regulator.

 

jeff

 

 

[chadbag]

1 minute ago, gavino200 said:

 

I worked it out. Just had to go get my loupes. How many do you need? 

 

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?

 

[gavino200]

2 minutes ago, chadbag said:

 provided a similar dimnes to the 2,2k on the V1.  I also tried 1K instead of 100 ohm (leaving the 2.2k on the cap side) and also swapping the 100 ohm and the 2.2k ohm.

 

 

I'm not too surprised by this. The relation between brightness and resistance is totally nonlinear. You see minimal change through a huge range of values, followed by an inflexion point where you start to get a big change in brightness with minimal change in resistance. I've experienced this whenever I've used potentiometers to control LEDs.

[chadbag]

1 minute ago, cteno4 said:

 

Yep that right having the caps charge to higher voltage will provide more energy to keep the led lit. The only reason to put the cap after the voltage regulator was to use lower voltage caps. Also the caps after the voltage regulator are only charged to the 5 or 3.3v and will drop below effective led voltage faster. 

 

Yeah, but I did not expect the difference I saw.  I expected a difference but it is a heck of a lot longer.  I also wanted to get a long flat response since the dying voltage of the caps would be converted to the lower 5V (or 3,3V) so as it moved from 12V down in its decay, which would show steady voltage until it dropped to belowe what the voltage regulator can do (the RC constant is to about 63% so 12V would go down to about 7.5V)

 

BTW I was using DCC track power in the tests, which is around 14V coming off the rectifier.

 

1 minute ago, cteno4 said:

My guess is the best shoud be using the 3.3v regulator.

 

I think you are probably right.