I've used carbon piles to test very high current power supplies.
One needs to verify software. Set 800 amps at 5v and see what is
flowing. Set 1000 amps and see if the pile is smoking.

We actually smoked our pile once - man that was a fast race to
disconnect and start to float some co2 around it to cut off the air
and turn off the fire if any.

It was a bit more crunchy the next time we used it, inside it was toast.

We had 2000 Watts across it and it was rated at 1500. Was to be used
at max of 1000 by ourselves.

We never used them to ramp up/down since ours was a mechanical
'resistor'.

We had static voltages in hundreds of amps at fixed voltages variable I
fixed E.

We had IEEE 488 voltage programming with high current ability. The
boxes protected themselves by trimming the voltage back to limit the
current.


I would think you could use that but not by turning knobs in real time.
Set and switch & see. Set and switch & see. You might be watching
results of the 'machine' and not of the battery. Once you find
something specific, try to re-create or verify with another load.
Which then takes out the servo effect of the pile electronics and
heating... of electronics and pile. Hot pile, wrong results.

Martin

On 3/19/2016 8:33 AM, Jim Wilkins wrote:
Yesterday I bought the HF 500A carbon pile battery tester 91129. The
instructions are aimed at a one time Go/NoGo garage test of customer
batteries, for 15 seconds at 1/2 of the CCA rating.
http://www.harborfreight.com/500-amp...ter-91129.html
The adjustment is unstable below 30A and drifts upward as the carbon
heats up. It isn't the good 5A or 10A discharge test load I had hoped
for, but it does what it's supposed to well enough.

It seems to me that the current draw at the pass/fail voltage step for
the ambient temperature would be a more sensitive indication of
battery condition -if- you record it to compare to later. There is a
large change in current for a small change in voltage, and the
pass/fail voltage point should be the same for any starting battery
size or condition.

Has anyone else used a variable load tester this way, or have a better
idea?

--jsw