Variable LED current source on SPI

[karri]

Hi,
some months ago I published this project:
http://doitwireless.com/2014/09/18/wireless-led-driver-with-pca9634/
There is PCA9634, which is PWM led controller.

Now for a new project I am looking for another LED driver - for some LED current source for multiple LEDs, which works on SPI or some another similar fast bus. I need change 32 RGB LED intensity (a few steps) in less than 1ms. PWM in useless for me, because of noise, which could cause (there is hi-impedance measurement on the board).

Edit: Expected LED current about 20mA for each LED.

Have you someone any idea, which devide could I use for this project?
Thank you

[Ichan]

The current requirement is not mentioned, for arrays of low power rgb leds WS2811 is very attractive, or WS2812 which is WS2811 on 5050 style rgb led.

-ichan

[solutions]

PWM is useless to you because you are building a PAM communications link.

If you just come out and tell us what you are actually building, you'll get a much higher quality answer than such as the one from the rogue that posted before my post

[Gallymimu]

As solutions said, some more detail, specifically on your high impedance measurement would be nice.  There isn't really a reason why PWM shouldn't work if you design your circuits properly.  It sounds like your switching currents should be pretty low so noise should be easy to deal with.  Softening your switch times can also help.

That said, you could use a bunch of these to do it:
http://www.ti.com/product/tlc5922

"The TLC5922 is a 16-channel constant-current sink driver. Each channel has an On/Off state and a 128-step adjustable constant-current sink (dot correction). The dot correction adjusts the brightness variations between LED, LED channels, and other LED drivers. Both dot correction and On/Off state are accessible via a serial data interface. A single external resistor sets the maximum current of all 16 channels."

[Ichan]

If you just come out and tell us what you are actually building, you'll get a much higher quality answer than such as the one from the rogue that posted before my post

  Glad to see you post.

To the topic, consider this if you are going to make it in good quantity, i use some from them.

-ichan

[karri]

Thank you all for tips.
I have no complete idea about contruction, because I cooperate only on LED driving part. This device measure some biological signals (EKG I think) on the head with high-impedace sensors. LEDs are only for eye stimulation. This is second generation of device. Previous generation used PWM LED drivers and there was some false alarms from PWM LED driver on sensors output. So, the idea is, use the another method of RGB led driving.

O.

[Gallymimu]

I've built these kind of systems before.  You are right that non-PWM will be easier, but if designed properly it should work with PWM as well.  You will have similar issues with logic noise associated with the SPI.  It's really a matter of having proper ground planes, appropriate current return paths and keeping the switching currents from feeding through the sensitive circuitry.