2003-09-03 22:37:34.403502+00 by Dan Lyke 10 comments
My lesson of the day: Don't underestimate the power of an optoisolator to smooth your nice square waves down into completely undifferentiateable slopes. Corrolary: Remove the redundant bits and your board'll work better.
[ related topics: Dan's Life Work, productivity and environment Embedded Devices ]
comments in ascending chronological order (reverse):
#Comment Re: Lesson OTD made: 2003-09-03 23:21:44.250053+00 by: td
As Adam Levinthal used to say: Every little picofarad has a nanohenry all its own.
#Comment Re: Lesson OTD made: 2003-09-04 00:12:19.486776+00 by: Dan Lyke
Lesson two: Switching power supplies don't like back EMF (I think, I can't see it on the scope, but when I spin the stepper motor with no power on the windings occasionally the power supply fades to a little less than 1v over the course of a second, then flips back up). Got some honkin' diodes, but still a little concerned about this one.
#Comment Re: Lesson OTD made: 2003-09-04 02:03:49.474908+00 by: Brian
Switching power supplies don't like much of anything besides a nice simple resistive load. Remember that they're measuring the output voltage to decide how much power to throw into the load, usually at 10kHz or faster, so you can confuse them by accepting the current and then throwing it back (say, with a highly inductive load).
Most switchers I've seen won't even turn on unless they see a certain minimum load. Some embedded CPU boards I've got have an ATX power supply connector, but you have to attach a dummy load (usually an old hard drive) to make the common ATX supply willing to turn on.
I say add more capacitors. You can never have too many capacitors :).
-Brian
#Comment Re: Lesson OTD made: 2003-09-04 03:20:44.745161+00 by: Dan Lyke
And diodes. Make sure the current is flowing only the direction it should be.
Although I'm a little concerned about my zener diode and resistor hack to keep surges on Vcc from getting too high, I think they're all well over tolerance, but I'm obviously a newbie at this.
And if, like me, you need a refresher on a henry, it's kg * m2 / s2 / A2.
#Comment Re: Lesson OTD made: 2003-09-04 06:53:57.930266+00 by: jjg
Wow, I totally don't understand a single word of this.
#Comment Re: Lesson OTD made: 2003-09-04 16:35:02.2092+00 by: Dan Lyke
It involves this device (I love the little "no sledge hammers" icon on top of that position encoder...), mostly revolving around how that big motor causes wonky stuff to happen to that little power supply when the circuit board on the left decides to leave the motor unpowered (so that things can hopefully spin fairly freely).
I'm up to 4700uF on the capacitor (costs more than the CPU...) and those paired honkin' diodes (2A was the largest I could find, and I'm fairly sure that the motor draws more than that when half-stepped) on the power supply leads, and I think with a watchdog timer this'll be okay if I can get the noise out of the position encoder signal.
#Comment Re: Lesson OTD made: 2003-09-05 21:39:18.855503+00 by: Dan Lyke
Well, after diddling with this a bit I've determined that this power supply just isn't going to deliver what I need. So I'm poking about for alternatives, though I'll probably just use a full-sized PC power supply in the mean-time (it worked for my previous system).
But I also see that my next step is learning more about that.
#Comment Re: Lesson OTD made: 2003-09-06 00:25:43.643077+00 by: Larry Burton
Perhaps the Accopian website could help you find the appropriate power supply.
#Comment Re: Lesson OTD made: 2003-09-06 16:59:43.861174+00 by: Dan Lyke
Given that I'm driving the larger system off of a $15 power supply, and that we have working prototypes, adding $100 to $200 to the cost of the production system wouldn't go over well with my higher-ups...
They look like something I'd want on my bench, though.
#Comment Re: Lesson OTD made: 2003-09-08 23:40:26.185711+00 by: Dan Lyke
Amazingly, a cheap PC power supply seems to work fairly well. Occasionally when I try to freewheel the motor very quickly it'll shut down, but even lacking the .5 to 1A minimum load I thought I'd have to have on it it's working well enough for me to finish debugging the code while I try to find a permanent solution. Hmmm... Wonder where I can find some 4-5A diodes that have really low resistance...