Inductor Q measurements

In circuit design at high frequencies one of the important parameters is the Q of

the inductor. Instruments have been designed that can accurately measure the Q
of an inductor that allow the circuit designer to attain close to perfection
knowing the actual Q value. The instrument the writer has intimate experience
with is the Marconi Instruments TF-1245A “Circuit Magnification Meter”. I am
fortunate to have been able to purchase such a unit with the two signal generators
that feed the instrument for the full range of 40kHz to 50MHz and 20MHz to
300MHz. These are the companion generators TF-1246 and TF-1247. The
Marconi Instruments Q meter was first introduced in 1957 and was made up until
about the mid 1970s.
Although other manufacturers also made Q meters such as HP, Boonton and
General Radio I have limited knowledge of these types. The HP version we have
at the local university but it is somewhat limited in frequency to only 75MHz, so
it finds little use in practice.
The Marconi Q meter is a simple instrument but very accurate. It not only can
measure the Q of an inductor but the precision tuning capacitor is accurate to
0.25pF resolution and with this it is possible to calculate the exact “in circuit”
inductance with great precision. The tuning capacitor is a work of art and I
happened to be involved with the company that made these for MI many years
ago. It is a heavily gold plated item and has two sets of vanes ganged together.
One section is the high frequency portion and is 7.5pF to 110pF. The lower
frequency section is higher in value and is 10pF to 400pF. Both of these tuning
capacitors have Q values in the several thousand, so they have almost no effect
on the measured Q of an inductor.
For measurements up to 50MHz both sections can be connected in parallel
giving a variation from 17.5pF to 510pF.
A picture of the TF-1245A is shown below with some covers removed as it has
recently undergone a major refurbishment to replace some ageing resistors and
capacitors.

Marconi Instruments TF-1245A Q meter

Measurement of inductor Q
Q or “quality factor” is the parameter that characterises a practical inductor. By
comparison to capacitors the equivalent parameter is known as ESR. Any real
world inductor has some inherent series resistance, either due to simple copper
loss (Ohmic loss) or due to the additional “skin effect” at high frequencies.
Whereas it is simple to make a dc ohmic measurement to establish the copper
loss value in ohms the skin effect also adds another resistance effectively in
series with the copper loss resistance. Hence, a simple dc ohmmeter
measurement doesn’t tell the true story.
In addition to the ESR value there is also another parameter that needs to be
known to calculate the tuning capacitor value to resonate an inductor to the
desired frequency. This is the inherent “parallel capacitance” that exists in a
practical inductor. For solenoid wound coils with several layers of wire on a
bobbin this parallel capacitance can be fairly significant. This value needs to be
subtracted from the calculated tuning capacitor value to attain resonance at the
correct frequency. For single layer solenoid wound air dielectric inductors or
inductors wound on toroidal cores this parallel capacitance is rarely more than a
few pico-farads and can be ignored in most cases.
The generally accepted equivalent circuit of a practical inductor is shown in
Figure 1.

Figure 1
The equivalent circuit has an inductance represented by L and a series resistance
Rs which caters for the ohmic and skin effect losses. Across this network is the
inherent stray winding capacitance Cp. To measure the Q the inductor is
resonated with a precision air-variable capacitor to the desired frequency.
The standard measurement of Q is with the inductor brought to series resonance
and the current then is at maximum. The equivalent circuit is shown in Figure 2.
It is well known that at resonance the value of reactance of the inductance and
capacitance are numerically equal but of opposite sign. It is the convention that
inductor have a positive sign and capacitors a negative sign. At series resonance
the two reactances cancel and the only portion left is the ESR. By establishing
the magnitude of the current flowing gives us the exact value of ESR. Knowing
this then Q is simply: Q=X/R
X can be accurately established from the capacitor value and the measurement
frequency using the standard equations.
Measuring RF currents is difficult to achieve accurate results. There is however
a simpler method.
If the two ends of the L and C network are connected together it becomes a
parallel connected network. The same circuit conditions apply and at resonance
the current circulating in the closed loop is at maximum and the voltage
appearing across the network is also maximum. It is easier to measure voltage
than current in practice. The signal being fed into the network has a very low
source impedance, ideally 0Ω, and so it is able to produce a large current into the
network. The source impedance of the MI Q meter is not more than 0.02Ω for
the low frequency range and a similar value for the high frequency range. As
this is very close to the ideal 0Ω then little error occurs.
The RF voltage appearing across the tuning capacitor is directly related to the Q
value as the same current flows in the capacitor as the inductor. Hence, by
measuring the RF voltage we achieve the same result.
The Q meter is essentially a very accurate VTVM with a precision RF detector
circuit

Figure 2 Series resonant mode and parallel resonant mode

The basic RF portion of the Q meter consists of an input for the signal generator,
a tuning capacitor with accurate dial markings and the RF detector circuitry. In
the picture of the TF-1245A the tuning capacitor dial has a main dial and a sub-
dial that has 0.25pF increments. This is similar to a receiver with a main tuning
dial and a “band spread” dial. Reading the two dials can establish the exact
resonating capacitance to great accuracy. The two dials are mechanically ganged
with a clever mechanism. By rotating the larger dial the tuning capacitor is
rotated from one stop to the opposite stop. The smaller dial being “geared
down” by the linkages moves the main dial a small number of degrees.
To find the exact capacitance the main dial is turned until the Q peaks to
maximum and the cursor position is noted. The incremental dial is initially set to
0. Suppose the main dial falls between two increments on the main scale, say,
between 160pF and 170pF. We move the main dial to the 160pF marking and
the Q indicated now falls a little. By adjusting the incremental dial we again
peak the Q meter to maximum and then read off how many extra pF we needed
to add. The actual capacitance is 160pF + the incremental dial value, say,
2.75pF as the increments are in 0.25pF steps.
Having determined the capacitance to resonate the inductor we can calculate the
reactance for the test frequency. Since XC and XL are the same value at
resonance it is simple to calculate the effective inductance value using the
standard equations. The Q meter has a chart on the top that gives the correlation
between capacitance and inductance at standard test frequencies. These are also
listed in the operating manual and shown below. The oscillator tuning dial has
these special frequencies marked with a small dot so finding them is simple.

Test frequencies and conversion chart for inductance

Two different input networks are provided to cater for the full frequency range.
The lower range uses precision resistors to feed the high RF current into the
network, but this is not suitable for above 50MHz. The higher range uses two
precision inductors instead of resistors. The signal generators are not 50Ω
source impedance. They are typically less than 0.5Ω and can generate a high RF
current in the series resonant network.
This current typically is about 0.1A and this generates a high RF voltage across
the tuning capacitor. For a Q of 50 the rms voltage is 1V, for a Q of 150 it is 3V
rms and for a Q of 500 it is 10V rms.

Interconnection and block diagram

The diagram below shows a simplified block diagram and how the instrument is
connected for usual measurements.

Interconnection diagram
Input signal terminating details
The terminating impedance for the two signal generators is different for the low
and high frequency sections.

Figure 3 Input circuitry of the TF-1245A

The RF detector is a miniature thermionic pencil diode (EA52) that is usable up
to at least 1GHz and has a completely flat response from dc to 1GHz with a high
efficiency.
Looking into the screened compartment that houses the tuning capacitor and the
detector diode is shown in the next picture. Everything is heavily gold or silver
plated to reduce skin effect losses.
 TF-1245A tuning capacitor and RF detector diode
The stators of the two section tuning capacitor are solid blocks of brass with
semi-circular troughs formed by gang milling with multiple slitting saws spaced
on an arbor. The saws are plunged into the block to the required depth. The
rotor portion is built up like a conventional tuning capacitor with thick brass
vanes. Grounding of the rotor shaft is taken care of by multiple sections of
finger-stock and grounding wires rubbing on the shaft and the moving vanes.
The anode of the EA52 is connected to the high frequency tuning capacitor by a
springy piece of foil which is also gold plated.
The round object above the EA52 is the low frequency resistor network which
uses thin-film deposited onto a ceramic substrate. The resistors are trimmed to
the exact values required during manufacture. The high frequency input inductor
is encased in a trough and not visible. Both input networks have a step-down in
impedance of 25:1.
The front panel contains two meters and various knobs for selecting the Q range
and the Set Zero knob to balance the VTVM circuitry. In use the signal
generator is set to the desired frequency and the level turned up until the left
hand meter indicates the “Set RF” level. Input RF level detection is by 1N21
cartridge diodes.
The right hand meter with the Q range knob gives the Q value measured on three
scales, 50, 150 and 500. For a Q higher than 500 the signal generator level meter
has an additional set of markings of 1.5 and 2. These are lower RF levels from
the signal generator and the Q meter reading is multiplied by either 1.5 or 2 to
give the correct measurement. It is possible to measure a Q value of up to 1,000
using these additional scale divisions.

Schematic
The original schematic is a bit confusing as some vital information is not
adequately annotated. I redrew the schematic to simplify the circuit by omitting
some of the finer details. The parts left out are the δQ switching (delta Q) which
allows the bandwidth of the tuned circuit to be incrementally established.
The VTVM section is straight forward but has a couple of odd methods to
achieve fine and coarse balance, that in practice could be dispensed with and the
components removed. Coarse balance is by varying the grid voltage of the
reference half of the 12AU7 differential amplifier by a divided down portion of
the 85V supply.
One notable feature is the complex filament supply. This a “reactor
transformer” that monitors the mains voltage and adjusts the filament current to
hold it constant. The mains voltage is not directly measured but the 250V
unregulated dc supply before the gas regulator. If the mains voltage changes so
does the raw 250V dc feed from the rectifiers. The current bled through this
“transductor” from the 250V unregulated supply, which partially feeds the 150V
gas stabiliser tube, changes the coupling between the primary and secondary
windings and hence effectively varies the turns ratio and hence the output
voltage. The mains transformer low voltage winding is nominally 21V rms.
This portion of the circuit controls the filament current of the 12AU7 and the
EA52 effectively connected in series across a nominal 12.6V ac supply. The
EA52 is fed by a 1:1 transformer to provide RF isolation. Both tubes draw
300mA at 6.3V nominal. The filaments are brought up slowly at switch on and
the current is limited to a maximum of about 450mA. This causes a gradual
heating and protects the filaments from current surges that can shorten the life if
a normal filament winding were used. The filament current and hence voltage
applied is set by an internal pot to give precise control to within a few mA
variation. This contributes to the accuracy as the differential amplifier and the
RF detector diode sensitivity is very reliant on correct emission levels. The
stated accuracy of Q values is better than 2% across the whole range.
Two gas regulators are used, one for the nominal 150V anode supply and a 85V
for the coarse balance and δQ biasing controls and switching (not shown).
The redrawn schematic shown below.
 TF-1245A schematic
The two signal generators are simple LC oscillators with turret tuning to switch
in the required band coils and tuning is by an air-variable capacitor. One of the
oddities is the choice of the tubes. The low frequency oscillator (TF-1246) uses
a 5763 (QV 3/12) RF beam tetrode, which is a logical choice. This is a Hartley
oscillator with the anode bypassed to ground and the cathode feeding a tapped
inductor. What is surprising is that the regulated dc supply also uses 5763 tube
as the regulator element. Quite why Marconi Instruments made such a choice is
a mystery.
The high frequency oscillator (TF-1247) uses a push-pull oscillator using a uhf
QQVO 2/6 dual beam tetrode and the power supply uses a 5763 as the series
regulator tube. In both oscillators the power output is varied by the voltage
applied to the screen grid of the oscillator tube. The power supply is also a split-
rail type so the anodes are connected to ground for the dc condition.

Block diagram of the TF-1247 oscillator

The frequency stability and read out accuracy of the frequency isn’t very exact
but the output level control is very precise. In practice as long as the frequency
is within a percent or so it is quite adequate. Q only varies slowly versus
frequency so a small error in frequency does not have any significant effect on
the actual value indicated. Q normally rises as the frequency is increased until
the self resonant frequency and then it begins to fall past this frequency.

Problems associated with the instrument

Operational problems are normally few. One area that will cause an incorrect
reading is any contact resistance between the inductor and the terminal posts. If
this is not very low resistance it degrades the potential Q. The terminals are
silver plated and for the low frequency section they can accept normal 4mm
banana plugs. If this connection is not perfect then erratic measurements occur.
Often simply polishing the plugs will increase the reading by as much as 20%, so
attention to these areas is important. Often it is better to use solder tags attached
to the ends of the inductor and to tighten the terminals well to eliminate any poor
connections.
Another contact problem is the BNC connectors that connect the oscillator units
to the main unit. The special short 50Ω coax used to link the oscillators to the
main unit is very mismatched as the load impedance of the main unit is only
0.5Ω. The centre pins/sockets and the grounding bayonets need regularly
cleaning to ensure a good connection.
For a large air wound inductor any metal within less than ≈1-coil diameter will
severely “de-Q” the inductor, because it acts like a shorted turn. Having the
inductor with very short leads close to the main unit front panel reduces the Q by
a large factor. It is preferable to use longer thick wires to increase the distance.
The Q meter when supplied had an insulating right angle shelf that attaches to
the two large screws at the bottom. This allowed the inductor to stand vertically,
which reduces the de-Q-ing factor. Many units have lost this useful attachment
over the years.

Picture of MI Q meter with shelf attachment

Another potential problem area is the turret mechanism used in the oscillator
units. As the picture above shows the band selector knob is a large pointer knob.
This has a detent mechanism that holds the shaft in the selected position after
rotating it. The rotation is through 360° and at each step it makes a loud “clunk”
as the knob is turned. Attempting to rotate the knob too quickly can break the
delicate fingers in the internal wafer switch. Repairing or replacing these
damaged contact fingers is a delicate operation and extra care needs to made
when changing ranges to avoid this mishap. There is a warning note on the front
panel of both oscillators to move the range switch knob slowly!
 Range switch cautionary note

Most of the units made are now over 40 years old and electrolytic capacitors in
the power supplies have a limited life. Some of the carbon resistors have a
tendency to drift higher in value with age. The tubes normally will last the life
time as they are being run well below the maximum ratings. However, the 5763
tubes in the low frequency oscillator and the power supplies do get tired and
replacements are useful to have. Wafer switches need regularly cleaning with a
contact lubricant as do the various wire-wound pots used.
The low frequency signal generator on the lowest frequency ranges uses pile
wound Litz wire for the inductors. These are a source of trouble as the rosin flux
solder used to attach the delicate lead out wires eats away at the copper after a
long time. This causes the wire to go open circuit and the oscillator to fail. The
higher frequency ranges use either enamelled copper wire or tin plated copper
wire and these do not suffer so much.
The rubberised insulation on the connecting wires does degrade and can become
conductive leading to erratic operation and stopping of the oscillator at certain
tuning capacitor settings. The oscillator produces considerable power and the rf
voltage across the tank circuit is several hundred volts at full power. This can
cause arcing between the components and wires. The only certain cure is to
rewire all the oscillator sections with better wire. The sections are packed tightly
together so it is a tedious operation to do correctly.
Working on the signal generators is also difficult as the power supply section is
attached by an umbilical cord, but with care this can be unbolted and moved
away from the area of work. The turret assembly is contained in a large screened
box and when the cover is removed the large Bakelite circular turret holding the
coils etc can be removed by undoing three 2BA nuts. The circular disk can then
be carefully wiggled off the central hub and placed on the work bench.
Replacing the disk needs to be done with care. It is aligned by a dowel pin but
the disk must be offered up with an offset so that the pegs which contact the
spring fingers do not become damaged. The pegs can only enter the spring
fingers sideways and not from above. Once on the hub and correctly bottomed
fully it can be rotated to the nearest detent position and the nuts replaced. The
oscillator tube is underneath the turret and impossible to get at without removing
the disk. To replace the oscillator tube is at least a 2 hour job as the coils also
need to be re-aligned for bottom and top of the range on all sections! The
oscillator coils have slugs to set the low frequency end and trimmer capacitors to
set the high frequency end. There are 8 oscillator coils to align.

Low frequency signal generator internal parts

One of the inherent problems with the VTVM section in the Q meter main unit is
the presence of leakage paths as the differential amplifier input stage has a very
high resistance from grid-1 to ground. In the original design this was a 100MΩ
carbon film resistor and these often go open circuit. A normal DVM cannot
measure these very high value resistors as it over-ranges past about 40MΩ. Most
units have had the 100MΩ changed to a metal-film 30MΩ resistor. This entails
a re-calibration.
The 12AU7 pin out is a little unfriendly for differential amplifier service. The
grid pin is between the anode and cathode pin and any moisture or dirt on the
base causes a small leakage current to the grid pin. (In older tubes often the grid
pin was the top-cap on the tube, which was a better method). The insulation
spaghetti sleeving used when the units were originally manufactured tends to
crumble and deposit bits onto the tube base that increase the leakage current
problem.
Although the tube base is a PTFE type this can still track small leakage currents
when the humidity is high and dust is present. The rosin flux residue from when
the pins were soldered absorbs moisture and gradually becomes conductive,
causing a leakage current to flow. A 1µA leakage current with the 100MΩ grid
resistance is an off-set voltage of 100V. To reduce the off-set voltage to less
than 1V requires the leakage current to be less than 10 nano-amps.
The writers unit had this problem and it caused an erratic jumping of the set zero
position which, after many hours of investigation, was traced to the dirt on the
tube base. A thorough washing with alcohol cleared the problem. As a
precaution, the area was sprayed with silicon spray to improve the moisture
repellent properties.

12AU7 tube base showing flux residue and crumbled insulation

The large resistor attached to the grid-1 pin of the 12AU7 is a 30MΩ metal film
high stability type originally used in Tektronix ‘scopes as part of the high voltage
monitoring for the CRT supply.

Other measurements possible

As well as Q measurements an instrument such as the Marconi Instruments Q
meter can also determine other parameters with great accuracy. These additional
measurements are exact capacitor value at the operating frequency, capacitor
tuning range and minimum and maximum values, dielectric constant of
insulating materials (permittivity) and permeability of magnetic materials.

Exact capacitor value.

For this measurement a suitable inductor is selected for the frequency of interest.
The unknown capacitor is connected in parallel and the parallel network is then
connected to the Q meter. The signal generator frequency is set to the operating
frequency.
The main tuning capacitor dial is turned until resonance is obtained and the value
read off the tuning dial. The unknown capacitor is then removed and the tuning
dial re-adjusted to bring the network back to resonance. The unknown value of
capacitor is the difference between the two dial readings.
At vhf/uhf the marked value of the capacitor is often different to the actual
capacitance. This is due to the leads having some small but significant
inductance. By measuring this way we can ascertain the effective capacitance
value with the leads at the length they need to be for the assembly envisaged.
For a variable capacitor the same technique is used. A suitable inductor is
resonated with the capacitor fully meshed and fully open. This gives the two
effective values and hence the tuning range.
Dielectric Constant measurements
Marconi Instruments could supply a dielectric measurement fixture that attached
to the Q meter. It consisted of a pair of circular disks one of which could be
varied in distance with a micrometer adjustment. In use a suitable inductor was
connected in parallel and the piece of insulating material placed between the two
disks and the micrometer adjusted to just nip it so all the air was excluded. The
material was free to slide and the micrometer reading noted.
The network was brought to resonance at the frequency at which the dielectric
was needed to be measured. The tuning dial capacitance was noted and the
insulating material then slid out. This reduced the effective capacitance so the
tuning capacitor dial was re-adjusted to bring the network back to resonance.
The effective permittivity of the dielectric is the ratio between the first
capacitance value and the second at resonance.

This same method can be used to measure the dielectric loss by using the δQ
function.

Permeability measurement
This uses the same method as the permittivity measurement except the inductor
can accept a ferrite slug or rod. The frequency is again chosen at which the data
is required and with the ferrite rod inside the inductor the resonance point is
found and the tuning capacitor dial setting noted. The ferrite rod is removed and
the tuning capacitor will need to be increased to bring the network back to
resonance. The effective permeability is the ratio between the second
capacitance value and the first. For example, with the rod inserted Ct = 50pF
and with it removed Ct = 150pF. The effective permeability is hence 3.
This method can also be used to adjust an inductor to an exact value and to check
the tuning range of a ferrite slugged can type. By adjusting the slug from fully in
to fully out the range of inductance can be calculated. Also with the known in
circuit capacitance the tuning capacitor dial is set to this value and the signal
generator is set to the required frequency and the slug then adjusted to peak the
resonance.