[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

...puters... Was Re: Measurement Errors was ( More Coupling...)



Original poster: "davep by way of Terry Fritz <twftesla-at-uswest-dot-net>" <davep-at-quik-dot-com>

Tesla list wrote:
 
> Original poster: "John H. Couture by way of Terry Fritz
<twftesla-at-uswest-dot-net>" <couturejh-at-worldnet.att-dot-net>
 
> Terry -

> I am pleased to see that coilers on the Tesla List are taking more interest
> in TC testing. However, it should be noted that comparing test measurements
> with computer results can be misleading unless the "error problem" is cleary
> understood.

	Indeed.  Entire books have been writ on how to make measurments.
 
> 1. All real world TC measurements have a certain algebraic (+/-) error. This
> error can be reduced but not eliminated by making several careful
> measurements.

	And understanding the equipment.
	_especially_ when dealing with, uhmmmm, irregular waveforms
	as in Tesla Coil power systems...
 
> 2. If a computer is working properly the calcs and outputs are always
> correct for the inputs used. It is the inputs that are in error, not the
> outputs. 

	If, brother, IF.....

	I've been in the 'puter business for 30 years.  I'd be real
	relucatant to make such a statement.  (I dealt in the
	_real_ world.  30 tons of molten steel in a ladle overhead,
	under computer control....)
  
>This means all inputs involved must be carefully measured.

> For example if your measurement (oper freq) is 5% too high and the computer
> output is 5% too low, the difference is 10% but they can be both considered
> correct if the error is stated. The true value is somewhere between the
> measured value and the computer value. The computer inputs obviously have a
> total algebraic error that equals a low 5%, very confusing??

	Indeed.

	Its Very Difficult (some would say impossible) to get all the
	errors out of the program.  I am NOT criticizing anyones's code,
	but its REAL difficult to get it right.  In General, the rule is:
	when predictions disagree with data, then the predictions
	are in need of checking.   The flip side of that is that
	sometimes the measurements are not what they seem.  Both ends
	need to be checked, carefully.

> It is obvious that coilers should use accuracy estimates with great caution.

> Comments welcomed.

	As above.  Also, beware that some 'errors' are not algebraic.
	_ESPECIALLY_ when fed to a 'puter program.  A 5% error in an
	input to a program can yield a 10% or 100% (or 1%....) error
	in the output.

	Example:
	This regresses to something i did not get a chance to comment on
	some weeks back:  the 5% (or whatever) accuracy spec of a meter.
	Sounds simple.  5%.  One needs to read the fine print (IF the
	manufacturer has given the details...).  Typically, meter specs
	are given as % _of_ _full_ _scale_.  So on a 1V scale, a 5%
	error is 50 mV.  Now.  Reading 50mV on THAT scale means a _100_%
	(potentially...) error and the meter is still in spec.  For
	any given meter the 'typical' I used may not apply, but if working
	in detail, these details affect the meaning of the results...

	best
	dwp