Night rider effect with 15 segments

Discussion in 'The Projects Forum' started by Fenris, Aug 6, 2009.

  1. Fenris

    Thread Starter Active Member

    Oct 21, 2007
    288
    2
    Hi Guy's

    I've had a request for help with a project for a Special Weapons Dalek my friend has built. This is a fullsize, sit and drive prop. On the front of the shoulder section there is a vertical louver. There are 15 gaps between the louvers.

    So the plan is to install a night rider effect chaser light. I have drawn up the 4017 circuit from Bills schematic which has 3 4017's connected. I have added a fourth 4017 to it so that there are enough outputs to cover the 15 slots.

    Now the only caveat is. 3 LED's per output so thats 45 LED's in total. Colours will be 5 sections red, 3 sections green either side and 2 sections yellow at each end. I am considering using an darlington array to act as the interface for the higher load.

    The LED's will be super bright types as well. signal diodes will be fitted to the 4017's outputs to protect them from each other and a 1K resistor to the ULN2804. On the LED side I need to sort out the current resistor and how best to arrange each group of LED's (series/parallel) I have read parallel is bad and seen in Bills ebooks LED's arranged in series per channel.

    I am also aware that due to the nature of the LED's that their behaviour will not emulate the fade effect of the filament bulb based circuit. That is unless I fit a capacitor across each group to cause the effect.

    How do you calculate the current resistor value with LED's in series? Oh and it will be running on 12V

    Heres a picture of the front of the shoulders. You can see the vertical slats where this project is to be displayed.

    regards

    Fenris
     
    • s31.jpg
      s31.jpg
      File size:
      175.1 KB
      Views:
      155
  2. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Please post your circuit diagram so that we know what your thinking is so far.

    You DO realize that the ULN2804s have a 10k resistor on the input, correct? That should be all that is necessary.
    Depending on the current, you'll lose approximately 1v to 1.1v on the output of the Darlington. That leaves you with roughly 11v.
    Rlimit >= (Vsupply - totalVfLED) / DesiredCurrent

    Scary lookin'. :eek: ;)
     
  3. Fenris

    Thread Starter Active Member

    Oct 21, 2007
    288
    2
    Hi Sgt

    Heres the schematic for the design. Thanks for the pointers. The particular LED's I am looking at totaled ->

    3x Red for 1 channel = Vf 6.6V
    3x Grn for 1 channel = Vf 9.9V
    3x Ylw for 1 channel = Vf 6.6V all are rated for 50mA max.

    So.

    (11V-6.6V)/40ma
    (11V-6.6V)/0.04 = 110Ω or 120Ω nearest standard value

    (11V-9.9V)/40ma
    (11V-9.9)/0.04 = 27.5Ω or 33Ω nearest standard value

    40mA was chosen to give some headroom.

    Will 1N4148 diodes be exceptable to use for the CR denoted parts?

    My main concern is have I understood Bills circuit well enough to have added the fourth 4017 correctly.

    The only reason I was going to fit the 1K resistor was because of a single 4017+ULN2001 circuit I found on the web but that was going to drive 6off 12V 2.2W filament bulbs. But thats a massive difference isn't it.

    I have shown only the first few connections between the 4017's and the ULN's for clarity. Hopefully this shows I know what I am about :D

    EDIT added the 'what it will look like image'

    Here's another picture and a link to a Local TV appearance with my mate who built it. The Design was done in the 1980's By Alan Marshal for the 30th Anniversary show that never happened. Chris Smith, an extremely talented CGI artist, it's a hobby not a job, added his own style to it and Alan Clarke built it.

    http://www.projectdalek.co.uk/storm/calendar.html

    regards

    Fenris
     
    Last edited: Aug 8, 2009
  4. Fenris

    Thread Starter Active Member

    Oct 21, 2007
    288
    2
    Little bump ;)

    regards

    Fenris
     
  5. Bernard

    AAC Fanatic!

    Aug 7, 2008
    4,170
    395
    Does display start in center and spread outward in both directions, with all LED's lit at full extension, then collapse at same rate untill only center red LED is on??
    I was thinking of using two MM74C95 4-bit right-left shift registers,a NAND gate, a C555, & 7 stage LED driver.First stage drives both L & R of center, so only 7 stages neened?
     
  6. Fenris

    Thread Starter Active Member

    Oct 21, 2007
    288
    2
    Hi Bernard

    No. This will start at one and and only one LED set will be lit at any one time, I guess 'dot mode would describe it well, and this 'dot' will sweep left then right. With a bit of fancy work with the wiring between the LED's and the driver circuit you could set it up to have 2 dots sweeping back and forth either side of center.

    I think to get a bar effect you would need a couple of LM3914 chips to drive the LED's. I have seen a circuit that uses these but they don't use a 555 they use an op-amp to provide a 'sine wave like' voltage to drive the 3914's because they work in proportion to the voltage input.

    regards

    Fenris
     
  7. Bernard

    AAC Fanatic!

    Aug 7, 2008
    4,170
    395
    Been too long since I saw Night Rider, thanks for explination. There have been several Night Rider projects over the last couple of years on AAC
     
  8. Fenris

    Thread Starter Active Member

    Oct 21, 2007
    288
    2
    LED's circuits in general seem to be very popular :D What I am trying to do is simple as far as the 4017 part goes it's just driving the Higher rated and quantity of LED's that's my main concern I think. Though if my maths is right the 40ma current for each group is well under the 500mA that the ULN chip can handle per output.

    I understand that the ULN can also be 'stacked' so that in effect each output could handle 1A so driving filament bulbs is an option to.

    Battlestar galactica - the cylons. Their helmet 'eye' does a knight rider like sweep. At least it did in the 80's version I haven't seen the new version.

    regards

    Fenris
     
  9. RiJoRI

    Well-Known Member

    Aug 15, 2007
    536
    26
    There have been a number of articles in Nuts & Volts about using first an Atmel Butterfly, and then an Arduino Duemillenuove (sp?) board to produce "Cylon Eyes". While the series just uses 7 LEDs, it can readily be expanded to 15 or more LEDs.

    --Rich
     
  10. Wendy

    Moderator

    Mar 24, 2008
    20,765
    2,536
    I'll be putting some thought into the fading, but you're approach might work. How would you feel about putting transistors per LED?

    My thought is something called a capacitance multiplier, where the higher impedance of a transistor driver would do the same thing. It's late, and I don't have time to draw right at the moment, but start with this drawing...

    [​IMG]

    Add a diode (to prevent backdischarging) and a capacitor to ground on the base. I'll go into more detail later. The diodes would also be useful for the special effects, such as back and forth.
     
  11. Fenris

    Thread Starter Active Member

    Oct 21, 2007
    288
    2
    Hi Bill

    I came up with this but I think I have not quite understood you as far as the diode. I get the idea that it prevents emf/disharge being sent where we don't want it. Did you mean inline on the base from the 4017's? If so we are covered as the outputs have 2 each to run the knight rider pattern. The 1st and last output require one though.

    The capacitor will feed the base with a reducing voltage when the signal from the 4017 has moved on. Which will cause the transistor to proportionally control the load voltage so that the LED's fade. Did I get it? :D

    The transistor per LED module has a lot going for it. It will simplify the interface. Each of the 15 LED modules would consist of 3 LED's, the required current resistor and capacitor. Would the 2N7000 mosfet be a candidate again or would it's function be different to the transistor you suggest?

    regards

    Fenris

    PS. How long does a mail shot from the US usually take?
     
  12. jj_alukkas

    Well-Known Member

    Jan 8, 2009
    751
    5
  13. Fenris

    Thread Starter Active Member

    Oct 21, 2007
    288
    2
    Hi JJ thanks for the link. That is where I found the ULN chip idea :) The ULN is probably overkill as it can handle 500mA per output and I am only drawing 40mA per output so the ULN's are probably overkill :D Transistors or mosfets are the better over all solution I think.

    regards

    Fenris
     
  14. jj_alukkas

    Well-Known Member

    Jan 8, 2009
    751
    5
    Aren't you driving the led's from the 4017?? You can drive the LED's simply with an NPN and a resistor, but considering the no of transistors you need per output, the ULN isn't much a bad idea. It would be cleaner with possibilty for large no of led connections. :)

    Anyway where is it that you are facing problems??
     
  15. jj_alukkas

    Well-Known Member

    Jan 8, 2009
    751
    5
  16. Fenris

    Thread Starter Active Member

    Oct 21, 2007
    288
    2
    Thats what Bill is proposing but with a twist. For what this unit is going to be used for it will actually work out better to have a transistor per group. It makes the setup of the unit neater and easier in this case. But the ULN is certainly a viable option for alternate setups/uses of the end product.

    I built that dice circuit.............about 27 years ago :D happy times!

    regards

    Fenris
     
  17. Bernard

    AAC Fanatic!

    Aug 7, 2008
    4,170
    395
    The NTE46, darlington, NPN, .5A, gain 10k, TO92 case, should allow a reasonable RC time delay of around 50msec. or greator.
     
  18. Wendy

    Moderator

    Mar 24, 2008
    20,765
    2,536
    Basically you've got the idea, the diodes let the cap charge as fast as the 4017 will do it, and then the cap will discharge though the transistor. You'll need to tweak the LED resistor of course, but that isn't a big deal.

    A Darlington will extend that time dramatically.

    As to the post office, I suspect they are the same all over, I meant to email you the invoice, which has some tracking numbers. I'll dig it up and scan it.
     
  19. Fenris

    Thread Starter Active Member

    Oct 21, 2007
    288
    2
    Hi Bill

    OK then I have that bit understood :) The resistors will be 120Ω and 33Ω to suit the LED's used. 2 off the colours use the 120Ω and the remaining colour uses the 33Ω.

    I take it I have the 4017 circuit with the extra 4017 added correctly? I attempted to extrapolate from your original 3 x 4017 circuit.

    Is there a way of working out the fade time? This is going to be affected by how fast the circuit is going as controlled by the 555 timer circuit. The slower the sweep occurs the more charge will be put into the caps. The fade will be proportional in duration to how much charge the cap was able to get before the cycle moved on. The faster the cycle the less charge the cap will get during the on period of the output it's attached to.

    Is there any need for a base resistor? and what transistor should I select? I am guessing that what ever is it must be able to handle more than 40mA which is the LED load. As you say a darlington type will add to the performance. Bernard has suggested an NTE46. Is this just a general purpose type. It can handle 500mA so in this instance I am well under it's capacity. That would also mean That there is a possibility of expanding the LED quota per group.

    That in mind as a theoretical question only - the setup I have now has 3 LED's per module. their Vf totals 6.6Vf and 9.9Vf depending on the colour. Now I only have 11V available after deductions, If I added one more LED of the same Vf to a group of 9.9Vf LED's then the total would go to 13.2Vf. That module would not work would it because it needs more than the supply available?

    To add more LED's I would have to add an identical trio in parallel to each existing group. For best performance they would then need to be matched to the original group for visual equality. and the total mA drawn with this enlarged group would be 80mA per the expanded group so still well under the example transistors capacity to handle.

    Thanks Bill

    regards

    Dave (understanding a little more everyday :D )
     
    Last edited: Aug 12, 2009
  20. Wendy

    Moderator

    Mar 24, 2008
    20,765
    2,536
    No need for a base resistor. The only problem I forsee is tolerance, between transistor gain variation (huge), capacitor variation (small), and different colored LEDs, which cause a large emitter resistance variation and voltage cutoff variation, the exact time for fade out is going to be all over the map.

    I suspect the way to fight that is to test select capacitors. This is one of those cases I wish I could let you use my junk box. Dan's Small Parts and Kits provided me with an interesting assortment of various electrolytic caps extremely cheap.

    This will likely be a situation when you design a PCB you just have to allow a slot for 2 or 3 capacitors to adjust to the exact value needed, and use the breadboard to pretweak as much as possible. It would probably be practical to measure the time off a test setup, then calculate the capcitance need to bring the time to whatever tolerance you set.

    If this were a commercial design none of this would be very practical, a microcontroller with PWM would be much better and smaller, but for for a one or two circuit knock off it is probably easier than writing code. You could even breadboard this in entirety, then transfer all the parts to their exact locations in the PCB, it would save replacement time on the PCB for parts.

    I'm thinking of a binary selection of capacitors, using 1µF, 2.2µF, 4.7µ, and 10µF you should be able to get pretty close to what you need (or a X10 or X100 variation, depending on how long the fade time needs to be).

    The base concept for extending the number of counts wasn't mine. My only addition is the diode gate instead of a conventional logic gate, and I think Bill Bowden came up with it first. From what I could see it looks good.
     
    Last edited: Aug 12, 2009
Loading...