LEDBrick Project - DIY pendant w/ pucks

[theatrus]

If anyone is interested in boards ahead of any docs, I've shared them up on OSHPark:

https://oshpark.com/shared_projects/7uwwfspd

Docs coming soon.

 

[shah1]

This is really nice build, tagging along to see progresss

 

[miroslavign]

LM3414 and driver board

LM3414 and driver board

First of all, thanks for sharing a design idea,

I was wondering,
it would be nice to also share measurements of the driver board.
On the original TI page:
http://www.ti.com/product/lm3414

a couple of users not praising this driver. How is the chip
performing on your board? I am having some doubts about
this current-mirror tech without true sense resistor and feedback.

If everything is ok, I would use this chip in conjunction with
http://www.microchip.com/paramchart...6&mid=11&lang=en&pageid=79&redirects=digipots
to digitally set max current per channel.

Anyway, before I found out this chip on your design, I looked at
lm3405/06, but again no digital max current set.

But, there is one really interesting chip - LT3476, not from TI. Quad based,
1.5A, up to 36Vdc. The driver is using REAL sense resistor, only setting
sense voltage.
So, for the sake of trying out different solutions,
it might be interesting for me to try this quad design ?

Let us first wait and see your measurements. Specifically set current
to measured current in different Vin. And also, power efficiency.

 

[theatrus]

First of all, thanks for sharing a design idea,

I was wondering,
it would be nice to also share measurements of the driver board.
On the original TI page:
http://www.ti.com/product/lm3414

a couple of users not praising this driver. How is the chip
performing on your board? I am having some doubts about
this current-mirror tech without true sense resistor and feedback.

If everything is ok, I would use this chip in conjunction with
http://www.microchip.com/paramchart...6&mid=11&lang=en&pageid=79&redirects=digipots
to digitally set max current per channel.

Anyway, before I found out this chip on your design, I looked at
lm3405/06, but again no digital max current set.

But, there is one really interesting chip - LT3476, not from TI. Quad based,
1.5A, up to 36Vdc. The driver is using REAL sense resistor, only setting
sense voltage.
So, for the sake of trying out different solutions,
it might be interesting for me to try this quad design ?

Let us first wait and see your measurements. Specifically set current
to measured current in different Vin. And also, power efficiency.

I haven't done a true cycle by cycle test of peak currents, but in terms of average current it is tracking very well in relation to Vin and the set resistor.

I need to do a full evaluation of efficiency at different operating points (as they are all running off of a 24V input, but have dramatically different Vfw on each channel), but rough measurements put this layout and design at > 90%.

 

[theatrus]

Finally, a complete driver + emitter build (so many other concurrent projects :mad2

All channels set to a 500mA limit. No external PWM.

Minimum input voltage to the driver is about 18V. Set to 20V for testing. The LM3414 driver board is still cool, but I haven't evaluated long term with all channels running.

 

[R_Mc]

Very nice! I've been waiting for a reveal. What is the footprint of the led board? Do you think the less could be grouped more tightly?

 

[R_Mc]

Also when will you start selling ; )

 

[theatrus]

Very nice! I've been waiting for a reveal. What is the footprint of the led board? Do you think the less could be grouped more tightly?

Yes, they could, with some thermal limits. The footprint of the board is identical to an 80mm fan. The heatsink is oversize a bit, and maybe a future rev would shrink those dimensions.

 

[theatrus]

Update:

PWM board. Based on the nRF51822, which offers a Cortex-M0 and a Bluetooth LE 4.1 radio+stack. Its lacking in the hardware PWM land but, minus some radio interrupt glitches, is fast enough to bang out a reasonable software PWM. (The not yet released successor, the nRF52xx, will do full hardware PWM but thats EOY).

Wire free, except for power!

Current WIP CAD board images:

Messy schematics: http://theatr.us/images/ledbrick/pwm/schem.pdf

As you can tell, this is about 1/6th the area of the driver, and could actually fit on the current footprint of the driver board for a future revision.

 

[theatrus]

For those asking about the efficiency of the LM3414HV driver, I did some in/out measurements today:

Test string: 6 (older) XP-G Cool White LEDs, warmed up. Cathode line has a 0.33ohm (0.33159 measured) resistor. Driver board has 4.7uF X7R ceramic output filter installed. Measured ripple at 304kHz is about 25mVRMS (there is a high frequency component ringing above 20MHz)

Vin: 24.002V, 0.396A, 9.505W
Vf: 18.105V, Istring: 0.498A, 9.0171W, 94.8%

Vin: 19.004V, 0.488A, 9.274W
Vf: 18.1299V, 0.4931A, 8.9398W, 96.5%

Vin: 26V, ..., 94.7%

Could be a bit higher, but as a whole it's good.

 

[ReefRescue]

tagging along

 

[ReefRescue]

very nice built.

 

[theatrus]

I played around with the mechanical bits while I wait for the PWM boards to arrive:

And it even hangs (mostly) balanced straight

 

[perkint]

Really does look pretty smart

Tim

 

[theatrus]

To make sure I'm in the right ballpark on light I tested a single LEDBrick (sans cyan channel, all channels at 500mA) against a set of 250W SE MH bulbs in a standard reflector, ARO Eballast. All are at 18inches, Apogee sensor is 6 inches off my garage floor. I did not correct for the -12% error from blue LEDs.

XM250 10k : PAR 147
1x LEDBrick 7 chan: PAR 112
Hamilton 14k: PAR 88

I'm planning one per 12", so that's 2 boards for every MH.

 

[R_Mc]

Looks great! . This is Diy done right

 

[theatrus]

First cut of the PWM board is ordered.

I made some small adjustments to it, including using a PCA9685 (whats another $2 to the BOM cost, and the PWM is jitter free) and supporting a 4 wire fan (there is also a FET to support a three wire fan with power control, if thats your kinda fun)

I'll be updating the GitHub shortly, and the OSHPark board is here: https://oshpark.com/shared_projects/5Te0s49n

 

[theatrus]

The family is growing:

 

[marco.sv]

Looks awesome!

 

[theatrus]

Spent a lot of quality time crimping, and have four functional units, all sans Bluetooth PWM controller board.

Here are two running, just because.

I'm experimenting with some shrouds on the splash guard. In this case, I applied some mylar tape to two vertical guards. This is all hand-snapped acrylic so excuse the imprecise edges:

If nothing else, it helps cut down on accidentally looking straight at the diodes from the side