LEDBrick Project - DIY pendant w/ pucks

[theatrus]

Putting those parameters into the spreadsheet:

(last row)

Looks like your inductor will be just fine, but will run a bit hot (due to high coil resistance) .

 

[Maryland Guppy]

My only miss right now is the pull down resistor.
If PWM were to fail I want no driver output.

 

[theatrus]

Current 4 channel board:

The PWM breakout header isn't wired into place yet

1.5x2 inches

 

[theatrus]

Her is where I am at.

Opted for the TSOT23-5 package.

I picked the 100UH inductor based on a datasheet from Richtek.

Will try to remember where and post link.

For RSense @ .15 1% to run 3 watt LED's @ 667mA



See anything flawed in my layout???

Layout looks just fine to me and doable with a home etch board.

 

[Maryland Guppy]

Thanks for the support, it has been 30+ years since etching a board.

Still the issue is dropping out the driver if PWM failure.
Any thoughts???

I use a StormX for many tanks right now.
How many individual drivers can one PWM signal run???

Revisions tonight to remove less copper from the board.

 

[Foggman]

i just got my lights in today, damn they are bright. now i just need to get proper heat sinks and drivers, i have 3 ELN-60-48D but i dont think they are enough to run 2-16 up boards properly. The math sayes they are ok as long as i figuard them right what do you think?

 

[theatrus]

i just got my lights in today, damn they are bright. now i just need to get proper heat sinks and drivers, i have 3 ELN-60-48D but i dont think they are enough to run 2-16 up boards properly. The math sayes they are ok as long as i figuard them right what do you think?

If you were to series together each board, thats about 50V (using 3.1V per LED), so with three drivers between them you should be fine (10, 8, and 14 LEDs in series). You'll need to combine two of the channels since you only have three drivers (which is only ~ 43V). It may involve a less slightly wire loop to loop it back

I'd start at lower powers on the driver (300-500mA) until you can verify the heatsink performance, and since its a ton of PAR.

 

[theatrus]

Thanks for the support, it has been 30+ years since etching a board.

Still the issue is dropping out the driver if PWM failure.
Any thoughts???

I use a StormX for many tanks right now.
How many individual drivers can one PWM signal run???

Revisions tonight to remove less copper from the board.

The StormX uses a PCA9685 PWM controller doesn't it? That things has a ton of drive power, and is relatively low frequency to boot so you could chain a "bunch" (highly technical term) of PWM inputs to one output channel on the StormX. The average Arduino output is also fairly high current so it can totem-pole a reasonable number of logic inputs. I'll have to refer to the RT8471 datasheet to see if the IAdj pin is specified for capacitance or other parameters, but I wouldn't worry even with 8+ fanout.

Its been probably a good 15 years myself since I tried to deal with a home etch board. All I remember is be wary of acid traps, which would imply less 90 degree corners but my memory is hazy

Also a detail picture of the inductor/IC/diode/capacitor/resistor section might help spot any errors.

 

[Foggman]

ok i combined ch 3 and 4. what is the proper way to test the amprage uesed using a normal digital multimeter?

 

[oreo57]

If you were to series together each board, thats about 50V (using 3.1V per LED), so with three drivers between them you should be fine (10, 8, and 14 LEDs in series). You'll need to combine two of the channels since you only have three drivers (which is only ~ 43V). It may involve a less slightly wire loop to loop it back

I'd start at lower powers on the driver (300-500mA) until you can verify the heatsink performance, and since its a ton of PAR.

I believe the ELN-48 goes into constant current mode at 1.3A w/ a 3% "boost possible via internal pot..
43.2 to 52.8v..

http://www.meanwell.com/mw_search/ELN-60/ELN-60-SPEC.PDF

Woudn't the drivers be limited to that current??

 

[theatrus]

ok i combined ch 3 and 4. what is the proper way to test the amprage uesed using a normal digital multimeter?

Start with the 10A range, and put it in series between one leg of the driver and the LED. Either side (anode or cathode) is fine. The resolution might be limited in the 10A range on your meter, but it will help you get in range (in case you put the driver into > 1A current output). You can swap to the lower current range to fine tune if you need to - just make sure to not swap "live" - unplug the driver before doing any wiring changes.

I'd also suggest cranking the max current potentiometer in the supply to minimum output or the lowest setting and then raise from there. And don't run the LEDs off a heatsink for more than 1 second tops at any current above ~50mA - there isn't much thermal mass in the star and it will toast pretty fast

(Most of the pictures which clearly show the LEDs were taken at about 5-10mA of drive current, or else the camera quickly saturates when looking right at them)

 

[oreo57]

The StormX uses a PCA9685 PWM controller doesn't it? That things has a ton of drive power, and is relatively low frequency to boot so you could chain a "bunch" (highly technical term) of PWM inputs to one output channel on the StormX. The average Arduino output is also fairly high current so it can totem-pole a reasonable number of logic inputs. I'll have to refer to the RT8471 datasheet to see if the IAdj pin is specified for capacitance or other parameters, but I wouldn't worry even with 8+ fanout.

Its been probably a good 15 years myself since I tried to deal with a home etch board. All I remember is be wary of acid traps, which would imply less 90 degree corners but my memory is hazy

Also a detail picture of the inductor/IC/diode/capacitor/resistor section might help spot any errors.

on LDD drivers the draw is typically measured in microamps (1mA equals 1000 micro amps) meaning the draw is so low that you'll almost never be limited by the number of drivers per channel if using LDD drivers.

and last I read the pwm channels ( I believe there is some dependency on which "bus"???) are limited to 20mA, collectively 40.. Don't quote me.

Q: Anyone know the specific amount of amps that can be controlled per channel?
A:yes.. coralux does and they put it on page 1 of the manual

Well I was close:

Max PWM current per pin :

40mA (Storm),
10mA (Storm X)

http://coralux.net/wp-content/uploads/wpsc/downloadables/Storm-Controller-Manual.pdf

 

[theatrus]

Mucking around with the driver board again.

I figured to add an optional on-board APEX 0-10V control input set using the PIC16F18323 (<$1 micro) as the interface. You don't have to populate it on the board if you don't want.

And then I realized RJ45s are about the size of a whole channel in this layout :eep:

Since connectors are expensive, how much would you hate having to solder the 0-10V wires to the board? Saves trying to cram 2 RJ45s for a direct wire APEX on to the board.

 

[theatrus]

and last I read the pwm channels ( I believe there is some dependency on which "bus"???) are limited to 20mA, collectively 40.. Don't quote me.





Well I was close:


http://coralux.net/wp-content/uploads/wpsc/downloadables/Storm-Controller-Manual.pdf

Looks like the RT8471 doesn't specify anything about input capacitance of the input pin. This would have to measured in circuit to figure out how much drive you'd need. I doubt its too much.

 

[theatrus]

Current status:

Two resistor networks (10k, 6.8k)
http://www.digikey.com/product-deta.../PIC16F18323-I-JQ/PIC16F18323-I-JQ-ND/5401250
Pads for a http://www.tag-connect.com/
http://www.digikey.com/product-deta...002MBV50HTSA1/IFX20002MBV50HTSA1CT-ND/4692356

Small incremental cost add for I2C control and full analog dimming - I realize the Adj pin also supports analog voltages, but I opted for an intermediary to allow the full range, not 2-10V. You can always divide 10V to the PWM input pins sans any of this circuitry.

 

[theatrus]

Discovered a few glitches. Most importantly, my diode was on the wrong side of the switch node (oops).

I updated some parts, brought in more models, rounded off the sharp corners and cleaned up any goofy traces:

 

[theatrus]

Its also a tiny board (1.6x2.2 inches). Looks huge on a monitor

A 1:1 print vs a normal credit card sized card:

 

[Foggman]

looks good
i got my lights mounted to proper heatsinks, and adjusted properly. doing a heat test now to make sure they wont get to hot if all is good they will be on the tank tomorrow and shall see how they handle a 28 inch deep tank

 

[theatrus]

looks good
i got my lights mounted to proper heatsinks, and adjusted properly. doing a heat test now to make sure they wont get to hot if all is good they will be on the tank tomorrow and shall see how they handle a 28 inch deep tank

Awesome!

I've been running a 16 up board continuously on a big heatsink with fan at 700mA for about 1.5 weeks now as part of a stress test. Hoping to keep that going as a non-accelerated (well, 2x accelerated since its 24/7) benchmark, even though its a lousy sample size.

Let me know if you see any issues, and also what drive currents you are running.

 

[theatrus]

For the above driver, I put the output files, schematics, etc under the CERN Open Hardware License on GitHub. I can try various exports from my CAD package (sadly Eagle is not one of those I can export to), it may be useful or be more of a mess.

https://github.com/blueacro/acroiq_drivers