On Sat, 17 Apr 2004 12:22:03 -0700, Roy Lewallen posted this:

Ah, just the person I've been waiting for. How do you account for
current bunching on the conductors (that is, non-uniform distribution of
current around the conductors)? What reference, equation, or program do
you use? Nearly all "first principle" calculations of Q I've seen
grossly overestimate Q, and I believe the failure to take this into
account is at least part of the reason. I haven't seen a decent
analytical method of dealing with it, and an anxious to see how you do it.

Then there's surface corrosion and roughness, radiation, and coupling to
nearby objects. How do you deal with those? Have you identified some of
the other factors that often make a simplistic "first principle"
calculation disagree so badly with carefully made measurements?

Roy Lewallen, W7EL

James Meyer wrote:

If you have to "do the math", you might as well just calculate the Q
from first principles and forget the "measurement".

Jim


I was responding to a suggestion that one could do the math to calculate
what the Q would have been if you hadn't tried to measure it. I was pointing
out that if you could do that math, and get it correct, that you could do the
whole exercise with math and forget measuring anything.

And how do you know for sure that calculations overestimate Q when
measuring Q to verify the calculations disturbs the very thing you're measuring?

An engineer knows when to say "close enough". A mathematician is never
satisfied.

Jim

On Sat, 17 Apr 2004 12:22:03 -0700, Roy Lewallen posted this:

Ah, just the person I've been waiting for. How do you account for
current bunching on the conductors (that is, non-uniform distribution of
current around the conductors)? What reference, equation, or program do
you use? Nearly all "first principle" calculations of Q I've seen
grossly overestimate Q, and I believe the failure to take this into
account is at least part of the reason. I haven't seen a decent
analytical method of dealing with it, and an anxious to see how you do it.

Then there's surface corrosion and roughness, radiation, and coupling to
nearby objects. How do you deal with those? Have you identified some of
the other factors that often make a simplistic "first principle"
calculation disagree so badly with carefully made measurements?

Roy Lewallen, W7EL

James Meyer wrote:

If you have to "do the math", you might as well just calculate the Q
from first principles and forget the "measurement".

Jim


I was responding to a suggestion that one could do the math to calculate
what the Q would have been if you hadn't tried to measure it. I was pointing
out that if you could do that math, and get it correct, that you could do the
whole exercise with math and forget measuring anything.

And how do you know for sure that calculations overestimate Q when
measuring Q to verify the calculations disturbs the very thing you're measuring?

An engineer knows when to say "close enough". A mathematician is never
satisfied.

Jim

On Sat, 17 Apr 2004 21:05:34 +0100, John Woodgate
posted this:

I read in sci.electronics.design that James Meyer
wrote (in ) about 'A neat
and compact way to generate RF harmonics...', on Sat, 17 Apr 2004:
On Sat, 17 Apr 2004 17:26:15 +0100, John Woodgate
posted this:

I read in sci.electronics.design that James Meyer
wrote (in ) about 'A neat
and compact way to generate RF harmonics...', on Sat, 17 Apr 2004:
IOW, the Q without the probe will be higher than the Q when you insert
the probe to measure the Q.

Use an inductive current pick-off. That how the Marconi Instruments 1245
series Q-meters work(ed).

Nevertheless, *ANY* method used to probe the field associated with the
resonator will load the resonator and degrade the Q.

IIRC, the Marconi unit used a 10 nH inductor (maybe less) made of a
short length of silver wire, gold-plated to eliminate sulfide attack.
The effect on Q would be minimal in the extreme.

But what about the energy extracted from the probe to make the
measurement? Any load on the probe is transformed into a load on the resonator.
Compensating for that load requires a very small number be divided by another
very small number. Any error in either number makes a much larger error in the
result of the calculation.

Jim

On Sat, 17 Apr 2004 21:05:34 +0100, John Woodgate
posted this:

I read in sci.electronics.design that James Meyer
wrote (in ) about 'A neat
and compact way to generate RF harmonics...', on Sat, 17 Apr 2004:
On Sat, 17 Apr 2004 17:26:15 +0100, John Woodgate
posted this:

I read in sci.electronics.design that James Meyer
wrote (in ) about 'A neat
and compact way to generate RF harmonics...', on Sat, 17 Apr 2004:
IOW, the Q without the probe will be higher than the Q when you insert
the probe to measure the Q.

Use an inductive current pick-off. That how the Marconi Instruments 1245
series Q-meters work(ed).

Nevertheless, *ANY* method used to probe the field associated with the
resonator will load the resonator and degrade the Q.

IIRC, the Marconi unit used a 10 nH inductor (maybe less) made of a
short length of silver wire, gold-plated to eliminate sulfide attack.
The effect on Q would be minimal in the extreme.

But what about the energy extracted from the probe to make the
measurement? Any load on the probe is transformed into a load on the resonator.
Compensating for that load requires a very small number be divided by another
very small number. Any error in either number makes a much larger error in the
result of the calculation.

Jim

James Meyer wrote:
. . .
And how do you know for sure that calculations overestimate Q when
measuring Q to verify the calculations disturbs the very thing you're measuring?

An engineer knows when to say "close enough". A mathematician is never
satisfied.


I've measured quite a number of inductors both with a homebrew setup, in
which I account for the losses in the input and output networks, and
with an HP Q meter of specified accuracy. With simple input and output
networks consisting of a small series C and shunt R, the effect on Q is
predictable and easy to calculate. Results from the two methods agree
quite closely, even though they use somewhat different methods to arrive
at the Q, giving a fair amount of confidence in both results. And both
disagree quite dramatically in some cases to Q calculated simply from
theoretical calculations which include only conductor resistance
(including skin effect, of course), inductance, and shunt capacitance.
This is with inductors of only moderate Q -- calculation of very high Q
inductors, which is being discussed here, would require more attention
to second order effects -- as would measurement.

Thanks for the profound observation about mathematicians and engineers.
In which category does one put a person who's satisfied with
calculations made without thinking about, caring about, or considering
the errors caused by ignoring fundamental effects? Certainly not an
engineer as I use the term.

Roy Lewallen, W7EL

James Meyer wrote:
. . .
And how do you know for sure that calculations overestimate Q when
measuring Q to verify the calculations disturbs the very thing you're measuring?

An engineer knows when to say "close enough". A mathematician is never
satisfied.


I've measured quite a number of inductors both with a homebrew setup, in
which I account for the losses in the input and output networks, and
with an HP Q meter of specified accuracy. With simple input and output
networks consisting of a small series C and shunt R, the effect on Q is
predictable and easy to calculate. Results from the two methods agree
quite closely, even though they use somewhat different methods to arrive
at the Q, giving a fair amount of confidence in both results. And both
disagree quite dramatically in some cases to Q calculated simply from
theoretical calculations which include only conductor resistance
(including skin effect, of course), inductance, and shunt capacitance.
This is with inductors of only moderate Q -- calculation of very high Q
inductors, which is being discussed here, would require more attention
to second order effects -- as would measurement.

Thanks for the profound observation about mathematicians and engineers.
In which category does one put a person who's satisfied with
calculations made without thinking about, caring about, or considering
the errors caused by ignoring fundamental effects? Certainly not an
engineer as I use the term.

Roy Lewallen, W7EL

I read in sci.electronics.design that James Meyer
wrote (in ) about 'A neat
and compact way to generate RF harmonics...', on Sat, 17 Apr 2004:
On Sat, 17 Apr 2004 21:05:34 +0100, John Woodgate
posted this:

I read in sci.electronics.design that James Meyer
wrote (in ) about 'A neat
and compact way to generate RF harmonics...', on Sat, 17 Apr 2004:
On Sat, 17 Apr 2004 17:26:15 +0100, John Woodgate
posted this:

I read in sci.electronics.design that James Meyer
wrote (in ) about 'A neat
and compact way to generate RF harmonics...', on Sat, 17 Apr 2004:
IOW, the Q without the probe will be higher than the Q when you
insert
the probe to measure the Q.

Use an inductive current pick-off. That how the Marconi Instruments 1245
series Q-meters work(ed).

Nevertheless, *ANY* method used to probe the field associated with the
resonator will load the resonator and degrade the Q.

IIRC, the Marconi unit used a 10 nH inductor (maybe less) made of a
short length of silver wire, gold-plated to eliminate sulfide attack.
The effect on Q would be minimal in the extreme.

But what about the energy extracted from the probe to make the
measurement?

The sensing inductor was connected to the grid of a triode tube, with,
IIRC, a 1 Mohm grid leak. With 1 mV across the sensor, that's a whole
1 uJ/s of energy extraction.

Any load on the probe is transformed into a load on the resonator.
Compensating for that load requires a very small number be divided by another
very small number. Any error in either number makes a much larger error in the
result of the calculation.

Check your math. It's small errors in *differences*, not in ratios, that
result in large errors in results.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

I read in sci.electronics.design that James Meyer
wrote (in ) about 'A neat
and compact way to generate RF harmonics...', on Sat, 17 Apr 2004:
On Sat, 17 Apr 2004 21:05:34 +0100, John Woodgate
posted this:

I read in sci.electronics.design that James Meyer
wrote (in ) about 'A neat
and compact way to generate RF harmonics...', on Sat, 17 Apr 2004:
On Sat, 17 Apr 2004 17:26:15 +0100, John Woodgate
posted this:

I read in sci.electronics.design that James Meyer
wrote (in ) about 'A neat
and compact way to generate RF harmonics...', on Sat, 17 Apr 2004:
IOW, the Q without the probe will be higher than the Q when you
insert
the probe to measure the Q.

Use an inductive current pick-off. That how the Marconi Instruments 1245
series Q-meters work(ed).

Nevertheless, *ANY* method used to probe the field associated with the
resonator will load the resonator and degrade the Q.

IIRC, the Marconi unit used a 10 nH inductor (maybe less) made of a
short length of silver wire, gold-plated to eliminate sulfide attack.
The effect on Q would be minimal in the extreme.

But what about the energy extracted from the probe to make the
measurement?

The sensing inductor was connected to the grid of a triode tube, with,
IIRC, a 1 Mohm grid leak. With 1 mV across the sensor, that's a whole
1 uJ/s of energy extraction.

Any load on the probe is transformed into a load on the resonator.
Compensating for that load requires a very small number be divided by another
very small number. Any error in either number makes a much larger error in the
result of the calculation.

Check your math. It's small errors in *differences*, not in ratios, that
result in large errors in results.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

James Meyer wrote:

On Sat, 17 Apr 2004 07:43:49 GMT, Robert Baer posted
this:

John Larkin wrote:


Well, all the usual methods: resonance width, phase shift, ringdown,
stuff like that. I work with gadgets with Qs over 1e9, and people
measure them without difficulty.

John

Ringdown is the easist way when Qs are extremely high.

You must still account for the energy you extract from the circuit in
order to measure the ringdown. Even the energy needed to drive a high impedance
probe is significant when the Q gets high.

IOW, the Q without the probe will be higher than the Q when you insert
the probe to measure the Q.

Jim

Not a problem; use two different loads.
Just like measuring the internal resistance of a battery or a curent
meter...
...Never done directly.

James Meyer wrote:

On Sat, 17 Apr 2004 07:43:49 GMT, Robert Baer posted
this:

John Larkin wrote:


Well, all the usual methods: resonance width, phase shift, ringdown,
stuff like that. I work with gadgets with Qs over 1e9, and people
measure them without difficulty.

John

Ringdown is the easist way when Qs are extremely high.

You must still account for the energy you extract from the circuit in
order to measure the ringdown. Even the energy needed to drive a high impedance
probe is significant when the Q gets high.

IOW, the Q without the probe will be higher than the Q when you insert
the probe to measure the Q.

Jim

Not a problem; use two different loads.
Just like measuring the internal resistance of a battery or a curent
meter...
...Never done directly.