Recent changes to Panel

Panel modified by Bernhard Trummer

Bernhard Trummer — Mon, 01 Dec 2014 20:14:07 -0000

--- v4
+++ v5
@@ -1,6 +1,4 @@
-## Hardware
-
-### The panel
+## The panel

 Before starting to cut and drill the plexi glass for the panel, I suggest to get a case first, where the panel is supposed to be finally mounted.
 This is to know exactly how big the plexi glass has to be cut to fit into the case.
@@ -11,7 +9,7 @@
 For drilling, I used a 5mm wood drill in a drill press machine.
 I definitely recommend to use a drill press to get the holes straight, which won't be possible by using a standard pistol-grip drill.

-#### Mounting the LEDs
+### Mounting the LEDs

 I decided to do multiplexing with the six segments of the snowflake.
 So for each segment I soldered the LED anodes together.
@@ -22,71 +20,3 @@
 Doing this kind of wiring is a hell lot of work.
 Since I like soldering stuff, it was quite fun for me.
 But I have to admit that I'll do this differently (simpler) the next time. :-)
-
-### The LED drivers
-
-A small comparison of the three different constant current sink LED driver chips I stumbled upon:
-
-
-    
-        
-            
-            
-            
-            
-        
-    
-    
-        
-            
-            
-            
-            
-        
-        
-            
-            
-            
-            
-        
-        
-            
-            
-            
-            
-        
-        
-            
-            
-            
-            
-        
-        
-            
-            
-            
-            
-        
-    
- MBI5026 TLC5940NT WS2803D
Outputs 16 16 18
Latch input pin yes yes no
Cascadable yes yes yes
PWM no 12 bit, no built-in oscillator 8 bit
Dot correction no 6 bit + EEPROM no
-
-MBI5026:
-Works simple and nice like a 74595.
-But because I want to use PWM and do not want to do this in software, I decided to not use this chip.
-
-TLC5940NT:
-This was my second choice. It has some nice features and an impressive datasheet, but lacks some core functionality what I'd expect from a "PWM capable" driver. Because you have to provide the clock signal for the internal counter and even have to reset it continuously, the software to drive it becomes quite complicated.
-Further, after building two prototype driver circuits (using 6 chips each), I managed to burn 4 chips in total. Probably due to some strange latch-up effects, even when using a 100nF and 2.2uF capacitor per chip.
-
-WS2803D:
-Finally I stumbled upon this one. Compared to the TLC it's much better for my needs, because:
-
-   * 5 chips provide exactly the number of outputs I need. :-)
-   * A much simpler and cleaner circuit board due to a different PIN layout compared to the TLC.
-   * No "self-destruction" in the initial prototype. The chips just work stable and without flaws.
-
-The only minor disadvantage is a missing input pin to latch the clocked-in data. The WS will do an automatic latch after 600 microseconds of inactivity on the clock pin. So this piece of timing needs to be done right in the software. Further, this delay also limits the maximum frequency for multiplexing (respective "frame rate"). However, in the end it turned out that I even can use an additional millisecond delay between switching the 6 segments during multiplexing and still be "flicker-free".
-
-### Multiplexing
-
-TODO

Hardware modified by Bernhard Trummer

Bernhard Trummer — Sat, 29 Nov 2014 14:08:23 -0000

--- v3
+++ v4
@@ -10,6 +10,18 @@

 For drilling, I used a 5mm wood drill in a drill press machine.
 I definitely recommend to use a drill press to get the holes straight, which won't be possible by using a standard pistol-grip drill.
+
+#### Mounting the LEDs
+
+I decided to do multiplexing with the six segments of the snowflake.
+So for each segment I soldered the LED anodes together.
+For the cathodes, I used wire pieces to make rings.
+Finally, I soldered six thick wires for the anodes and 30 thin wires for the cathode rings
+in order to connect the cathodes to the driver circuit and the anodes to the multiplexing circuit.
+
+Doing this kind of wiring is a hell lot of work.
+Since I like soldering stuff, it was quite fun for me.
+But I have to admit that I'll do this differently (simpler) the next time. :-)

 ### The LED drivers

Discussion for Hardware page

Bernhard Trummer — Sat, 29 Nov 2014 13:33:24 -0000

drill marks on the plexi glass

Hardware modified by Bernhard Trummer

Bernhard Trummer — Fri, 28 Nov 2014 22:38:51 -0000

--- v2
+++ v3
@@ -60,18 +60,20 @@

 MBI5026:
 Works simple and nice like a 74595.
-But because I want to use PWM and do not want to do this in software, I decided not to use this chip.
+But because I want to use PWM and do not want to do this in software, I decided to not use this chip.

 TLC5940NT:
 This was my second choice. It has some nice features and an impressive datasheet, but lacks some core functionality what I'd expect from a "PWM capable" driver. Because you have to provide the clock signal for the internal counter and even have to reset it continuously, the software to drive it becomes quite complicated.
-Further, after building two prototype driver circuits (using 6 chips), I managed to burn 4 chips in total. Probably due to some strange latch-up effects, even when using a 100nF and 2.2uF capacitor per chip.
+Further, after building two prototype driver circuits (using 6 chips each), I managed to burn 4 chips in total. Probably due to some strange latch-up effects, even when using a 100nF and 2.2uF capacitor per chip.

 WS2803D:
 Finally I stumbled upon this one. Compared to the TLC it's much better for my needs, because:
+
    * 5 chips provide exactly the number of outputs I need. :-)
    * A much simpler and cleaner circuit board due to a different PIN layout compared to the TLC.
    * No "self-destruction" in the initial prototype. The chips just work stable and without flaws.
-The only minor disadvantage is a missing input pin to latch the clocked-in data. The WS will do an automatic latch after 600 us of inactivity on the clock pin.
+
+The only minor disadvantage is a missing input pin to latch the clocked-in data. The WS will do an automatic latch after 600 microseconds of inactivity on the clock pin. So this piece of timing needs to be done right in the software. Further, this delay also limits the maximum frequency for multiplexing (respective "frame rate"). However, in the end it turned out that I even can use an additional millisecond delay between switching the 6 segments during multiplexing and still be "flicker-free".

 ### Multiplexing

Hardware modified by Bernhard Trummer

Bernhard Trummer — Fri, 28 Nov 2014 22:29:20 -0000

--- v1
+++ v2
@@ -2,7 +2,14 @@

 ### The panel

-TODO
+Before starting to cut and drill the plexi glass for the panel, I suggest to get a case first, where the panel is supposed to be finally mounted.
+This is to know exactly how big the plexi glass has to be cut to fit into the case.
+I didn't do this and had to cut a bit off the plexi glass after the LEDs were already mounted and wired. It's doable but not funny, I can tell. ;-)
+
+After cutting, I used an old-school ruler, a set square, and a pencil to draw the lines and markings for drilling the holes for the LEDs.
+
+For drilling, I used a 5mm wood drill in a drill press machine.
+I definitely recommend to use a drill press to get the holes straight, which won't be possible by using a standard pistol-grip drill.

 ### The LED drivers

Hardware modified by Bernhard Trummer

Bernhard Trummer — Fri, 21 Nov 2014 22:59:39 -0000

Hardware

The panel

TODO

The LED drivers

A small comparison of the three different constant current sink LED driver chips I stumbled upon:

	MBI5026	TLC5940NT	WS2803D
Outputs	16	16	18
Latch input pin	yes	yes	no
Cascadable	yes	yes	yes
PWM	no	12 bit, no built-in oscillator	8 bit
Dot correction	no	6 bit + EEPROM	no

MBI5026:
Works simple and nice like a 74595.
But because I want to use PWM and do not want to do this in software, I decided not to use this chip.

TLC5940NT:
This was my second choice. It has some nice features and an impressive datasheet, but lacks some core functionality what I'd expect from a "PWM capable" driver. Because you have to provide the clock signal for the internal counter and even have to reset it continuously, the software to drive it becomes quite complicated.
Further, after building two prototype driver circuits (using 6 chips), I managed to burn 4 chips in total. Probably due to some strange latch-up effects, even when using a 100nF and 2.2uF capacitor per chip.

WS2803D:
Finally I stumbled upon this one. Compared to the TLC it's much better for my needs, because:
* 5 chips provide exactly the number of outputs I need. :-)
* A much simpler and cleaner circuit board due to a different PIN layout compared to the TLC.
* No "self-destruction" in the initial prototype. The chips just work stable and without flaws.
The only minor disadvantage is a missing input pin to latch the clocked-in data. The WS will do an automatic latch after 600 us of inactivity on the clock pin.

Multiplexing

TODO