Hi all, this is correct, I'm using digital control to make a constant=20 current source. The reason to do all this is that the design is for a=20 specialized application that needs constant-current LEDs with highspeed=20 PWM: 29Khz with 1000 steps. This is more than an order of magnitude=20 better than LED driver chips can do. So it is basically 7 channels of buck constant-voltage dc/dcs with a=20 large cap on them, with their feedback (i.e. pump/don't pump) controlled=20 by a separate highspeed comparator that samples the current when the=20 LEDs are on during the PWM cycle, which can be as short as 100nsec=20 range. So this way I have 7 dynamically-adjusting "power supplies" that=20 are creating the correct voltage to produce the right amount of current=20 thru the LEDs. (180mA or 700ma or 800mA @ approx 44V in this case). The problem appears to be the dc/dc chip is having trouble with the=20 "faked" feedback I give it, which has to be 1.2V to 1.5V, since that is=20 what it expects to see across a current sense resistor when it is used=20 as it is normally designed. It is getting a very rapid 1.2V to 1.5V=20 signal (basically the comparator's low/high output generates 1.2V or=20 1.5V) that wouldn't happen if the chip were used as normal and I believe=20 this is the core of the problem. So I'm looking into ways to clean up the highspeed comparator's=20 "chatter". Digital methods preferred but all considered. Cheers J Robert Dvoracek wrote: > ---------------------------------------- >> Date: Mon, 10 Mar 2014 12:00:00 +0000 Subject: Re: [EE]: pulse/rate >> limiter From: cpmcsweeny@gmail.com To: piclist@mit.edu >> >> On Mon, Mar 10, 2014 at 1:54 AM, Jesse Lackey >> wrote: >> >> I'm not sure if I'm missing something here, but it seems what >> you're trying to produce is a constant current, yet you're trying >> to do this using digital rather than analogue control. The constant >> current controllers I've done (including those with uC control of >> current for dimming, using a PWM output and RC filter to provide a >> control voltage) > > Exactly. He needs to cut out the middleman (comparator) and use the > feedback pin on the buck chip directly. There is already a > comparator inside. Another one will only add to the latency of the > system. The exact implementation depends on the chip you're using. > I would say to look up the datasheet. There should be formulas for > the proper component selection and maybe even example circuits. It > all depends on the operation frequency and the inductance value used > in the output. > --=20 http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .