I'm working on a project where I need 54 channels of PWM (min 8-bit resolution per channel, 10 or 12 preferred) controlling common-anode LED arrays at 1.5-3A per channel. The LED assemblies have built in current limiting resistors, so I don't need constant current outputs. All I really need is a device with plain logic-level PWM outputs that I can use to control a bank of FETs. Trouble is, any device I've found that has more than one or two PWM outputs has constant current outputs, and I'd need an extra inversion step before I could drive an N-channel FET, which I'd rather avoid.

Any suggestions for something that would work better for my needs?

Thanks

 

Is it not possible to invert the PWM logic before? What circuitry are you using?

If so, you could use the constant current output (ON = "0" (sink)), resistor from Vcc to the gate of the n-FET, gate from the n-FET to the driver output.

 

Yes, I could do it that way, but it would mean that the default state of the outputs would be on at full. That's not likely to be a problem for this application, but it'd be a bit inelegant to have this giant LED array flash to full at power on before the mcu gets things under control. I suppose I could do what you suggest, but instead of pullups to Vcc I could tie the pullups to a buffer that defaults to gnd (holding the FETs off) until switched to Vcc by the controller.

But that said, another forum pointed me to the LT8500 (48 x 12 bit PWM logic outputs, SPI interface). It would actually be more convenient to find something with 16 channels due to the way this project will be modularized, but unless I find a 16ch LT8500 equivalent, I'll probably just go with that.

Thanks.

 

How about this one: http://www.nxp.com/documents/data_sheet/PCA9685.pdf

16ch 12bit.
It sources 10mA. However, output voltage is 5V, I don't know if one can drive a n-FET with this.

cheers

 

TLC5940 16ch 12-bit

 

I've used the PCA9635 and the TLC59401 drivers for low power common anode display. In my estimation, the I2C PCA driver sinking a low gate charge, logic level p-fet on the high side would be least amount of hassle. Say like the DMP3056LSS if it doesn't get too toasty.

EDIT: I understand this is supposed to be a common anode display, but at those currents it must be custom, or a commercial product is being hacked. Any links to the led array?

 

Wow, blast from the past. Funny timing that I got an email notification from this thread as I'm actually working on the project again finally, a year after initially asking the question. A lot has changed with the project overall, but I'm still using the LT8500--just got that part working today, in fact. It's quite a nice part--phase shifting, synchronous or asynchronous updating, self-test mode, and output enable/disable all built-in and available through the SPI interface. Downside is it has an absolutely asinine pinout. Since some of the arrays that I'll be driving have gotten bigger, I'll be using the LT8500 to control dedicated FET drivers to reduce switching losses on some chunkier n-channel FETs.

 

EDIT: I understand this is supposed to be a common anode display, but at those currents it must be custom, or a commercial product is being hacked. Any links to the led array?

Neither custom nor hacked. It's the plain old flexible RGB LED tape you can get from any number of vendors these days. Just a loooooot of it. I'm also overspec'ing the system somewhat to allow for other uses later on, like low voltage incandescent lamps. And it's easy enough to make the FET drivers optional, so the PCB can be populated for higher power or lower cost.

 

O.K. now I get it, for a minute there I thought it might be high power leds. The LT8500 is certainly an interesting part. If you get it all put together show us a pic, or video link for all to see, that would be impressive.