> Zhahai Stewart wrote: > > > Anybody else have any experience with such a stretched out > > multiplexing cycle? I presume it would work (at the 48 KHz > > pulse/500 HZ cycle you suggest to avoid flicker), but be > > fairly dim. > > The human eye is more sensitive to peak brightness levels than > it is to average levels; in my experience a 1/10 duty cycle > looks about half as bright as a 1/1 duty cycle at the same > peak current. Given the highly non-linear response of the eye (closer to log, I believe), I have a hard time knowing what half as bright is. As bright as the same LED at half the current? How about this: What duty cycle at 20 ma tends to approximate the brightness of an identical LED at constant X ma (eg: 5ma)? This should allow an A/B sort of test: set the constant current then increment/decrement the duty cycle to match - or vice versa. I'm certain this experiment has been done many times. Anybody know the results (or have a ready way to test this now?) ==== As for the LED matrix, I think the Maxim 7219 is very interesting, if the price is reasonable. One such chip would handle *everything* for a single 8x8 LED block: 8 anode drivers, 8 cathode drivers, internal clock and multiplexing, 8x8 internal RAM, serial shift register, current limiting (no R's needed even). It even has brightness controls in 16 levels (as well as current set by one resistor). I had seen the chip mentioned (Scott Edwards?) for controlling 7-segment LEDs (up to 8 of them), but didn't know it could also be configured to control LED matrixes (by disabling the internal BCD-to-7-seg decoding). Only problem (other than price and availability) is that the peak current is <40ma, typ 37ma. This is less than half the peak for the LED blocks (80 or 100ma). I could live with that; close enough. Thanks for the pointer, folks! (Anybody got a source for a dozen or so at good prices?) Zhahai @ Zhahai Stewart zhahai@hisys.com @ A Meme Gardener http://rainbow.rmii.com/~hisys/zhahai.html @ Standard Disclaimer YMMV - Your Maya May Vary