Content-Transfer-Encoding: 8bit
Message-ID: <5.1.0.14.0.20011121102933.00a7b730@gemini.herts.ac.uk>
Date: Wed, 21 Nov 2001 12:38:25 +0000
To: rsgb_lf_group@blacksheep.org
From: "James Moritz" <j.r.moritz@herts.ac.uk>
Subject: Re: LF: Measuring Q
In-reply-to: <001b01c17220$e9cc8820$531686d4@ericadodd>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

Dear Peter, LF Group,

This is one of the well-known techniques for measuring Q; as Rik pointed 
out, the important thing is to minimise coupling between generator and 
detector and the tuned circuit under test, to ensure the Q is not being 
decreased by loading by the test gear. The way to check is to reduce the 
coupling so the signal amplitude is reduced, say by about half, and measure 
the Q again - if it is significantly different, there is significant 
loading. You need to do this with both generator and detector, since either 
one can cause loading. With very high Q coils, the coupling must be very 
small; the equivalent parallel resistance of a high Q LF loading coil can 
easily exceed a megohm, so almost anything actually connected to, or even 
anywhere near a high potential point in the circuit will really clobber the Q.

What you really want to know is the equivalent series resistance of the 
coil. I usually do this as follows: Connect generator to meter and measure 
ampliude V1. Then connect a series resonant circuit using the coil to be 
tested in series with a suitable resonating capacitor across the generator 
terminals, and tune for a null in meter reading, ie. series resonance, and 
measure voltage V2. At resonance, the reactance of L and C cancel, and the 
remaining Rseries forms a potential divider with the paralleled source 
resistance of the generator (Rs) and load resistance (RL) of the meter. If 
you know what Rs and RL are, you can calculate Rseries:

Rseries = (RsRL/[Rs+RL])*1/([V1/V2]-1)

Typically, The meter needs to measure a reduction in voltage of  10 - 30dB, 
which should not be a problem for a level meter, or a scope. The main 
problem with this method is that harmonics of the generator signal will not 
be nulled out, and will produce an increase in apparent Rseries if that 
resistance is very low. However, this is not a problem if the generator 
output is clean (harmonics < 1%), or if a selective level meter is used to 
measure the voltages. It works well for series resistances less than the 
generator source impedances, which they should normally be. There is no 
other connection to the junction of L and C, which is the sensitive, high 
potential point in the circuit, and provided you know what it is, the 
source and generator impedance does not cause errors. A selective voltmeter 
with tracking generator is the ideal tool for this job.

Once you have Rseries, Q = XL/Rseries, = 2pi*f*L/Rseries

The resonant frequency of  the coil for the large coils we are using will 
depend on stray C between windings and connecting leads, so the apparent L 
will wary with frequency, as will Rseries. So again I agree with Rik and 
Andy that it is important to measure at close to the desired operating 
frequency. There will always be variations in stray capacitance between 
measurement of Q and connection to antenna, so the effective L will be 
somewhat variable. Also, with large diameter coils, nearby conducting 
objects will absorb energy from the coil, and again affect Rseries and L. 
So keep coil as far as possible from ground, metallic objects, walls etc. 
both when measuring and in use.

My 136k Loading coil has 80 odd turns of Decca litz wire on a sectional 
manhole former. L is about 4mH, and Rseries about 5ohms, making Q around 
700. For 73k, another sectional manhole is stacked on top, wound with about 
120 turns of 19/0.25 Teflon insulated stranded wire, which gives a total L 
of about 15mH, and a Q of around 300. The coils are wound in sections, with 
the total turns divided fairly evenly between the 14 slots on the former 
The required number of turns wound in to each slot before moving to the 
next, with the aim of minimising inter-winding C and maximising breakdown 
voltage, rather like the old-fashioned RF chokes. So G3LDO's Qs of less 
than 200 suggest either poor inductor performance or Q measurement errors. 
Having said that, with most combinations of antenna and loading coil, even 
reducing loading coil losses to zero would only lead to modest 10-20% 
increases in antenna current, because losses are dominated by the antenna 
itself. Other effects, like how wet the weather is, will produce similar 
variations. My main reason for winding big loading coils was to stop the 
things melting!

Q measuring seems to have gone out of fashion in the last few decades - all 
the major test gear companies have stopped making Q meters, which is a pity 
because the impedance meters which have replaced them do not cope well with 
measurements on high-Q circuits. Older textbooks, like Scroggie's "Wireless 
Laboratory Handbook", discuss Q measurement at some length.

Cheers, Jim Moritz
73 de M0BMU