On Sun, Jan 22, 2006 at 10:30:11AM -0500, Olin Lathrop wrote:
> You want to make them long enough so that the center gets close to the
> exepected temperature for an infinitely long fuse with the same current.  I
> don't know what the length/width ratio needs to be, but I'd knee jerk pick
> around 10:1 for starters.

Good idea, I'll try them around that level.

> Of course if you can deliver the energy very quickly, then shorter shouldn't
> matter much since there won't be time for heat flow away from the heated
> area.  You probably want to slowly charge up a large electrolytic cap (or
> probably a bunch in parallel) and then discharge it as quickly as possible
> accross the fuse.  In that case you want the fuse resistance to roughly
> match the cap ESR.

Sounds about right. Lithium coin cells are only rated for a maximum of
anywhere from 1ma to 20ma discharge from what I've seen in my research.

The shelf-life of the capacitors themselves could be an issue I think.
Electrolytic caps loose their electrolyte... This will be a
room-temperature application, but it's a very, very long time frame.

Still gotta figure out how many joules of energy I need to blow a fuse
first of course. :)

> In any case, making fuses from PCB copper is going to be rather
> unrepeatable.  A "6 mil" track is probably anywhere from 4-8 mils, check the
> specs.  Also most board houses only guarantee the minimum coppper width, not
> the maximum since for most applications more is better.  I think you should
> probably expect a factor of 2-4 to 1 in resistance accross production lots.

Thanks for pointing that one out, according to Advanced Circuits:

Trace Width/Air Gap
The greater of +/- 20% or +/- 0.002" 

This is from their pcb tolerances section.

So I assume if I went with 4mil tracks, they could be 2mil, or 6mil.

So, roughly, the minimum fusing current is something like a third of
the maximum fusing current. 

Here's a question... When they say "impedance controlled traces" do they
mean that they maintain consistant trace thicknesses? (amoung other
things)

> > Testing this would then be a matter of connecting each fuse up to a
> > low-impedance power source and and firing. I think a DSO measuring the
> > voltage rise across a shunt resistor should get me my current, and
> > integrating the current waveform and voltage should get me roughly the
> > power used, I think...
> 
> You're going to have a tough enough time blowing them without a shunt
> resistor in there.  For testing take a large cap and measure the before and
> after voltage.  From that you can calculate the energy used in blowing the
> fuse.  Since that will be how the real circuit will likely do it, it will be

Good idea. So I would charge the cap up to maximum, disconect it from
the charger, blow my fuse, and measure how much more current is needed
to charge it again?

> a relevant measure of circuit power drain to blow one fuse.  Also make sure
> you connect the cap to the fuse with a transistor or FET, not a mechanical
> switch.  The characteristics of a switch will otherwise get in the way.  You
> may also find that the transistor makes a better fuse than the copper trace.

A very real issue... One problem I forsee, is that I need to have around
32 to 128 of these transistors in the final array for the x and y. They
will need to be very low inductance, but also be physically small enough
to fit on the perifery of the array.

I've got plenty of beefy FET transistors with really low ESR's, but
they're big!

> How about a bunch of the smallest and cheapest surface mount diodes?  I
> think they can probably take less abuse than a 6 mil copper trace.

Good idea. I'm not familiar with surface mount diodes, know any part
numbers I should buy to try experimenting with?

-- 
pete@petertodd.ca http://www.petertodd.ca
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist