On Thu, Sep 08, 2005 at 03:01:34PM -0400, Timothy Miller wrote:
> On 9/8/05, Jack Carroll <[EMAIL PROTECTED]> wrote:
> > On Thu, Sep 08, 2005 at 11:52:51AM -0400, Timothy Miller wrote:
> > >
> > > (1) We need a test power supply circuit built and tested.
> > >
> > > We have a schematic for the power supply circuit designed. We need to
> > > build one on copper clad or something like that and test. Primarily,
> > > we need to see how much noise is in the output. If someone could
> > > build that for us, we can tell you how to measure it, or you can ship
> > > it to us for testing.
> >
> > I've got a lot on my plate right now, and no scope at home, but I
> > could help a little with this one. If you shoot me the schematic PDF, parts
> > list, and input and output electrical specs, I can do a pretty realistic
> > calculation of the output ripple and trash. If you're using parts I don't
> > have data sheets for, I might need to ask for copies of those too, if
> > they're not easily available from the web. I can also suggest PCB design
> > instructions to make sure the board as laid out achieves the predicted
> > ripple level. I've done that several times in the last couple of years.
> > What's the board layer arrangement, GND and Vcc in the middle, and
> > two signal layers each top and bottom?
> >
> > Another thing I could whack out pretty quick is a standard set of
> > PCB fab drawing notes. I think I remember most of what I wrote at the last
> > company.
>
> Ok, since I don't grok this stuff so well, I'm afriad I won't make a
> good middle-man, but Howard seems to prefer a physical circuit,
> because there are issues with how the layout can affect things. What
> he would do is us copper-clad and build something very much like the
> circuit for the final PCB and then test it on a scope.
>
> But simulating it in SPICE, if you can accurately model the PCB
> traces, might work just as well.
>
> In any event, if your concern is cost, just send me a bill for the
> parts and other materials you need.
No, my concern is time, and not having the right test equipment
here. But I've designed a _lot_ of switching power supplies, and written
the instructions to the PCB designer, and checked the resulting layouts
until all requirements were met, and hit all goals for ripple and transients
without building any breadboards. Is 1.5 millivolts peak-to-peak at 1.5 MHz
quiet enough for you? If not, I can improve on that, at the cost of
increased load transients.
Actually, I haven't bothered to simulate track inductance in the
past. I just write very paranoid layout rules, and keep the tracks out of
the common-mode paths that add to filter cap impedance. I also pay a lot
of attention to capacitor ESR and ESL, magnetic field leakage, and the
details of component grounding. SPICE can be useful in this kind of work,
but I'd resort to it more for transient analysis than for simple ripple
prediction. That's easy enough to figure with a calculator, because we only
need to know if it's below a limit.
So if Howard would like to send the rough design and the
requirements over here, I'll go over it and hunt for problems. You may
still want to breadboard it afterward for your peace of mind, but I can
improve the odds of it passing on the first try. I can offer some
suggestions about the test setup, too, so no trash gets into the scope input
that isn't really there on the power supply terminals.
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)
