markaren1:
basically it depends on the capacitance that each load presents. As a rough guide < 200pF total, 2 resistors will probably be fine.
So my next question then is: how can I find out what the capacitance is?
Ribbon cable has one of the higher capacitance’s at 46pF per metre, so this may be a useful guide.
Look at the Data sheet of the LED drivers - Cin Should be specified for pins.
Any capacitors being driven with data or clock on the LED boards ?
-Mark
Each driver has a 0.1uF cap, there are 24 drivers per string, four strings total. So calculating that means 9.6uF for the whole setup on the strings.
So there is 0.1uF on the data and clock input of each LED driver ?
Can you print the schematic for one LED driver here ??
Oh, on the DATA/CLK there is none. Those are power decoupling caps. The only thing that’s on the DATA/CLK line on each string is a 4.7K pull-down at the end on the string. Without that pull-downs it doesn’t work.
What is the longest length of wiring between controller and LED driver ?
There won’t be any ribbon cables used anywhere in the final design. They will get connected to the controller with, at the most, a 3" piece of wire, six of them. Two heavy gauge ones for each of VCC and GND and two thinner ones for DATA/CLK. Imagine a square, two diagonal corners will be VCC and the other two are GND. In the middle there’s a connection for the signal lines.
Doesn’t sound like a very capacitive load to me. 1 x 100R driving all 4 data and 1 x 100R driving all 4 clocks should be OK
-mark
Ok, test rig seems to work with that setup. Easier to put them on the controller board than to squeeze them in on the LED strips.
The thing I can’t figure out is that the factory made strip that I have looks very different as far as the number of components goes. This is what the schematic looks like from the datasheet:
I don’t need RL so I’m not adding those (and the datasheet also mentions this, at voltages less than 5.5V, and in this configuration, the drivers is fixed at 18mA.) I also don’t need the section in the blue box because I’m driving the string at 2.8V-4.2V - stuff in the blue box is for voltages higher than 5.5V. So that leaves the capacitors. There are two of them on the first IC and only one on the second (and presumably one on all the following ones in the chain.) And based on the symbols used, I can only assume one of those caps is a polarized one. But the datasheet doesn’t have any mention of this anywhere.
But, here’s the fun part, the physical string has way more resistors and caps on every section. Here’s a pic:
After a bit of continuity testing, this is what the schematic looks like:
What’s the 510 Ohm resistor for? Current control? And I know D1 is for driving the strip at higher voltage, so I omitted it from my design. Then there’s the dual caps. The datasheet only shows one. And lastly the 2x 27 Ohm series resistors on the DOUT and CLK_OUT lines. Datasheet says they should be 33 Ohm … Funny thing is, when I tested it with this configuration, with series resistors, things just didn’t work right. The string would just go haywire. However, adding pull-downs at the very end of the string solved all the problems.
Soooo, I don’t know what to believe anymore. I do know that my custom strings are working the way they’re designed: with no current resistors on the LED, with only one cap near the VCC/GND pair of pins, and no series resistors, but instead I have 4.7K pull-downs.
It's to keep the zener, being used as a voltage regulator, reversed biased but not too deeply. So yes, current control. If you're eliminating the zener, it can go too.kirash4:
What’s the 510 Ohm resistor for? Current control? And I know D1 is for driving the strip at higher voltage, so I omitted it from my design.
http://en.wikipedia.org/wiki/Zener_diode#Uses
One's on the Vcc to the IC itself, the other on the higher voltage Vcc. The latter smooths out spikes on that main supply and acts as a current resevoir to mitigate any local brownout when the local LEDs turn on.kirash4:
Then there’s the dual caps. The datasheet only shows one.
Taming line reflections is something of an art. Also the lines may be different from what the IC vendor expected (ie - different IC to IC spacing on this strip, trace widths, etc, etc).kirash4:
And lastly the 2x 27 Ohm series resistors on the DOUT and CLK_OUT lines. Datasheet says they should be 33 Ohm … Funny thing is, when I tested it with this configuration, with series resistors, things just didn’t work right. The string would just go haywire. However, adding pull-downs at the very end of the string solved all the problems.
Mee_n_Mac:
One’s on the Vcc to the IC itself, the other on the higher voltage Vcc. The latter smooths out spikes on that main supply and acts as a current resevoir to mitigate any local brownout when the local LEDs turn on.
When it’s driven at 5V, everything is the same voltage, there’s no higher voltage.
Mee_n_Mac:
Taming line reflections is something of an art. Also the lines may be different from what the IC vendor expected (ie - different IC to IC spacing on this strip, trace widths, etc, etc).
Meh, I’ll stick to the design I have now as that works just fine.
In your design, correct-a-mundo !kirash4:
Mee_n_Mac:
One’s on the Vcc to the IC itself, the other on the higher voltage Vcc. The latter smooths out spikes on that main supply and acts as a current resevoir to mitigate any local brownout when the local LEDs turn on.When it’s driven at 5V, everything is the same voltage, there’s no higher voltage.
As well you should.kirash4:
Mee_n_Mac:
Taming line reflections is something of an art. Also the lines may be different from what the IC vendor expected (ie - different IC to IC spacing on this strip, trace widths, etc, etc).Meh, I’ll stick to the design I have now as that works just fine.
As well you should.[/quote]There is a certain level of questioning myself, am I doing this right, am I forgetting anything, is there a reason why it's done this way over that way, etc., etc. But, I followed the datasheet, so I'm sticking with it. Despite factory made strips being different.kirash4:
Meh, I’ll stick to the design I have now as that works just fine.
Besides, both the factory made strips as well as my custom ones exhibit the same “issue” so I’m no longer thinking it’s my design … it happens on both.
Sounds like the chips are getting powered up through the ESD protection diodes on the data and clock lines, so they are actually running when Vcc is off. The LEDs, however, need Vcc, so they stay off until Vcc is turned on.
/mike
Um, what diodes?n1ist:
Sounds like the chips are getting powered up through the ESD protection diodes on the data and clock lines, so they are actually running when Vcc is off. The LEDs, however, need Vcc, so they stay off until Vcc is turned on./mike
markaren1:
stacked PCB on the main board, carrying the FET switches ?
Well, I suppose this is one way to work the problem … 19mm x 50mm. I can’t say that this will be the final layout but it’s a start. Biggest piece missing is a barrel connector however I left it out on purpose because I can’t figure out how to make it work with the tube that this is going to fit in. I can’t put it at the end because the tube cap is being used as an attachment point, so it’ll have to go in the wall but … it’s a headache. I may scrap it all together (which also means I can swap out the charging IC for one that doesn’t use DC, just USB power.)
this is impressive stuff, you have been rather busy
which design package ?
-mark
It's all done in EagleCAD, then exported to SketchUp. It does a pretty good job once I have all the models in place. It also allows me to see what works, where something gets too tight to deal with, where holes need to be made in the tube wall, etc., etc. Right now, this is about 2mm smaller than the inner diameter of the tube, so either I make them a bit wider, or get a 3D sleeve printed that this sits in. I like the latter idea as that will also anchor it in place. But that is a future step, after I get these made and tested ... in case I screwed up somewhere (which is rather likely.)markaren1:
this is impressive stuff, you have been rather busywhich design package ?
Many chips have internal diodes from each pin to Vcc and GND in order to shunt external ESD to ground instead of blowing up the chip. Normally they are reverse biased so they don’t get in the way. On the other hand, if power is off but the I/O pin is driven high by an external source, it will forward-bias the diode to Vcc and if the draw is low enough, can power up the chip.
/mike