Looking for a specific DC CC driver for LEDs

[karimwassef]

well.. how about this.

A fixture only needs one MCU, maybe with a MUX to grab the data from the different LED drivers. So think of it as a single master controller with multiple slave drivers.

This way, the cost per LED is in check, but the data for all LEDs can be captured and analyzed.

In my case, I'm planning on two fixtures. It may even be possible to have the controller be an external board that communicates with each fixture and all the LEDs on it. This way, the cost of the master controller per LED is insignificant.

 

[theatrus]

Signal levels will still dictate that the analog doesn't leave the board. Nothing wrong with one MCU and several drivers on that.

 

[karimwassef]

I was planning on sensing current externally with a shunt. That's $7 per LED, but probably cheaper in bulk.

You're saying the driver's own current sense can be used but since it's not referenced to ground, we need the circuitry to reference that signal down to ground?

Is the current sense average or instantaneous?

 

[karimwassef]

Signal levels will still dictate that the analog doesn't leave the board. Nothing wrong with one MCU and several drivers on that.

So - just so I'm clear. You're saying the analog sensing needs to be at the driver (or it gets swamped out by noise), but then the data can be relayed to the MCU digitally.

 

[theatrus]

The driver already has a shunt (a $0.20 power resistor), just placed on the high side of the switch (so signal levels are at 40-50V, +/- 10s of mV)

Looking at cheaper options than the LCT6101 to act as an amplifier. Once amplified, a low pass filter before the ADC (and maybe a buffer) will give you a great averaged current.

A simple divider, filter and buffer (LMV321 or other popcorn) will handle the Vout signal.

 

[karimwassef]

That should work. I would like to use an Arduino. I'm planning on having one anyway for dimming control, etc... and I'm a much better programmer than I am a circuit designer (or maker). More control to the user community out there too.

 

[theatrus]

So - just so I'm clear. You're saying the analog sensing needs to be at the driver (or it gets swamped out by noise), but then the data can be relayed to the MCU digitally.

ADC is MCU resident since we don't need anything fancy. MCUs for this can be *cheap*. A single PIC16F18323 could manage 1-2 drivers, and that part costs about $0.70.

 

[karimwassef]

Sure. I would like to capture the data over long periods of time - maybe uploading to a network, etc... Just thinking ahead.

 

[theatrus]

The DAC output can also drive the IAdj pin on the converter, so sampling (up to 32 levels of current) is under full control of the MCU. 32 steps might be enough to determine the curve shape.

Or else you set a sweep and sample the curve assuming you can compute outliers in real time (as you only have 256 bytes of RAM ). Looks like min conversion time for one input is 1uS.

Outside of the PIC area, there are a number of Cortex-M0 parts at the $1-2 price point which will easily handle this.

 

[theatrus]

Sure. I would like to capture the data over long periods of time - maybe uploading to a network, etc... Just thinking ahead.

Steady state shouldn't be a problem. If the driver data lines are connected to (whatever you want, RPi, Arduino) you can do continuous sampling.

 

[perkint]

No feedback on the results above?

Too techie or not interesting?

Interesting - yeah

Too techie? Hmm - not quite, but certainly a little deeper than I tend to read thru enough times to completely follow, unless it's something I am looking to implement (or may do soon).

But still interesting

Tim

 

[karimwassef]

Thanks Tim. I hope I made It relatable.

If anyone has any questions, please ask.