On Fri, 23 Sep 2005 22:42:14 +0000 (UTC), Christopher Tidy
wrote:

Geoff M wrote:
Use a water pump, and a valve to adjust the flow. For a proper job, mount
the pump on bearings so it can rotate, restrained by a spring balance,
which is the basis for a Heenan and Froude water brake engine dyno.
Geoff

Thanks for the suggestion. I don't have a water pump, but I've been
offered a 2 hp hydraulic pump/motor unit free. I don't see a problem
with using this, do you?

Hydraulic pump should work well. Most hydraulic pumps have some
"drag" even with no hydraulic load -- which is OK because most loads
you'll have on your converter will also have some minimum drag.

You probably know that the fluid will get hot with extended testing.
Your "brake" power is being converted to heat. No problem for short
tests or with plenty of fluid.

The hydraulic approach is considerably easier than a prony brake if
you want any torque measurement. The primary downside is that
hydraulics can be messy.

Car brakes, disc or drum, tend to chatter. You don't notice a little
chatter when decelerating the mass of a car, but you sure do in
something like this! Torque data has a lot of noise in it. (Been
there, done that.)

If you have a pressure gage, you can record output power as pressure *
flowrate. Flowrate will be pretty constant (displacement of pump *
speed) because induction motors only vary a few percent in speed from
noload to full load.
---

Jerry Martes once built an eddycurrent dyno for studying RPC's. It
worked extremely well, with torque easily adjustable from near zero to
about 5 HP worth at 1725 RPM. It used a thick aluminum disc and four
electromagnets made from alternator field coils. Torque was measured
by means of a torque arm and straingage loadcell, speed with an
inductive pickup. Both were recorded on a P.C. datalogger.

I designed some elex for it: a high resolution (about 200 parts per
million) speed sensor and a phasemeter that converted phase difference
between two voltages (or a voltage and a current) to an analog
voltage. I don't recall if we ever actually built either. The speed
sensor (really time per rev) was only good for average speed; it
measured counts during 1 rev from a 1 MHz clock.