As I've mentioned before, I'm working on an application
to monitor a bunch of computer and communications systems
in a data center environment. Most of the sensor designs
have been straightforward, but there's one that sort of
eludes me -- I have some ideas but I'm not 100% sure that
I'm not missing something more straightforward -- and I'd
appreciate y'all's thoughts on this.

I have 35 full-height (40U) 19" machine racks, each
of which has, at this point, a single single-phase,
120VAC, 20A dedicated circuit that is delivered through a
Pulizzi Z-line power controller.  (the 3302, picture
at http://www.pulizzi.com/p_ip3300.html, and data sheet
at http://www.pulizzi.com/ipc3xxx.html).  Each 20A circuit
is terminated in a single L5-30R twist-lock recepticle in
the raised floor plenum near the rack, and the Pulizzi
controller has an L5-30P plug on the end of a long,
rubber-jacketed supply cable; all power into the rack
has to come out of the Pulizzi.

What I'd most like to do is monitor the current draw from
each 20A circuit. At a minimum, this probably means interposing
some sensing device between the Pulizzi L5-30P and
the L5-30R in the floor so that I can separate out the
hot conductor and measure the current on it. I very much
do not want to have to open the Pulizzi and modify it;
any sensing strategy needs to leave the controller -- including
the power cable -- intact.

It would be icing on the cake if I could also, at low cost,
have an additional eight sensors to monitor the current
on each leg of the fan-out, but that information would
be much less useful on a routine basis than the
aggregate total.

I don't really need a lot of resolution on this, the 256
counts on an 8-bit ADC are more than sufficient; really
if I even had 1A resolution it would be OK, 0.1A
resolution would be great. What I expect to do is
routinely poll every few minutes and trigger an alarm if the
load exceeds 18A or some similar threshold. I'd also
like to be able to do a continuous poll on request to
generate a strip-chart output on demand if we want to,
say, watch the effect of powering up a new or balky
system.

The board I'm working on uses a PIC16C76, and there's
plenty of ADC ports availble for this, so all I really
need to do is to get that AC current turned into a voltage
input for the PIC. The obvious solution, to me, is to use
some sort of current transducer. Right now, I've got
some of the PC-50 units described here:

http://fwbell.com/Catalogs/Sensors/Current_Sensors/General_Information/4-20_mA_d
c_Output/PC_PCS/pc_pcs.html

However, the biggest drawback to those, IMHO, is the
need for the 24V power supply to drive the current loop
(thanks to everyone for helping me figure out how to
use these). Virtually everything else I'm doing will
run off just 5V.  There are probably other good transducers
that don't need all those volts, and some which may not
require any power supply at all. I'd love to hear any
suggestions along these lines. I'd also love to hear about
other ingenious ways to accomplish the same goal. I am
concerned a little about the linearity of the output,
and the ability to straightforwardly and accurately
(although not particularly precisely) translate the sense
voltage into an amperage reading. The solution also has
to be able to operate 24x7, ad infinitum, and should also
have a decently small response lag and conversion time,
so that I can get at least 1 or two samples per second.

You get the idea, sorry I'm so long winded about this
kind of thing.

Thanks,
--Bob

--
============================================================
Bob Drzyzgula                             It's not a problem
bob@drzyzgula.org                until something bad happens
============================================================