Académique Documents
Professionnel Documents
Culture Documents
RF IF 6 dB PHASE
LPF VCO
AD JU ST
LO
L O W N O IS E LO O P RF
O R L IM IT IN G A M P L IF IE R A M P L IF IE R
A M P L IF IE R
Figure 1: Block diagram of a typical VCO-PLL system which includes an RF amplifier and cavity structure.
___________________________________________
*
Supported by US DOE Contract No. DE-AC05-84ER40150
†powers@jlab.org
amplifier, such as the difficult to find Lucent LG1605, be DC-coupled. Sources such as an Agilent E4422B
provides an expanded dynamic range with reduced work well for this application. This and similar RF
oscillation problems. sources have a low FM bandwidth, have stable low noise
The LNA section is followed by a mixer. Typically RF drive capabilities, and are flexible with respect to the
double balanced diode ring mixers are used. Devices operating frequency. They have an added advantage in
such as a Mini-Circuits ZFM-150 are perfectly adequate. that the output can be AM modulated simultaneous with
The two major considerations are that the intermediate FM modulation. This configuration is used when
frequency (IF) output must be DC coupled and the performing a measurement of the dynamic Lorentz force
operating level of the local oscillator (LO) should be as effects [5]. Remember when performing such tests that
high as practical. Typically 7 to 13 dBm mixers are used. the minimum RF amplitude must be maintained at the LO
Mixers with a LO much higher than 13 dBm will require port on the mixer for the system to function properly.
that one insert an amplifier between the coupled VCO The final low level section consists of a directional
output and the LO input. coupler, along with the amplitude and phase controls. The
The low pass filter (LPF) stage serves two purposes. directional coupler is used to provide the LO signal from
The first is to eliminate any the frequency content at the the VCO to the input of the mixer. The specific coupling
fundamental frequency or it’s second harmonic. The is determined by the output capability of the VCO and the
other purpose is to limit the loop bandwidth to about required LO signal level. In some cases an amplifier must
20 kHz which reduces the system noise without be used between the coupler and the mixer in order to
compromising the lock time necessary for cavities which provide sufficient signal level. Typically the phase shifter
happen to have rise times on the order of 1 ms or less. is a mechanical device such as a Narda 3752 or Arra
The variable gain amplifier provides another way to D3428B. When selecting these devices insure that they
adjust loop gain that, unlike the LNA, is independent of provide at least 190º of phase shift or 370º depending on
loop phase. Unless the following phase shifter is capable the configuration of the loop amplifier. For manual
of more than 360º of phase shift, this amplifier should systems, a series of mechanical attenuators are used to
have an invert switch. At the summing junction, the error adjust the RF drive level.
signal is summed with an offset signal, that is typically Figure 2 shows the phase shift for two such devices.
generated using two ten-turn potentiometers, one for Although called an attenuator, the ARRA device is a
coarse and one for fine frequency adjustments. The continuously adjustable variable coupler while the Narda
reference voltage for the potentiometer is typically a band attenuator is a switchable attenuator with 1 dB switching
gap voltage reference based circuit. This is done in order increments. Both of these devices would function well in
to ensure that the source is stable and low noise. In a VCO-PLL system that is used for cavity testing. If not
addition to custom circuit designs, devices like a Stanford designed to be relatively phase stable such devices can
Research SR540 amplifier can be used to implement the have phase shifts on the order of 180º over a 20 dB range.
loop gain and filter functions. 20
There are a number of choices for VCOs. The least
expensive devices are broadband devices such as those 15
PHASE SHIFT (deg)
LO C AL VCO
REM OTE CONTROL
S W IT C H IN T E R F A C E P C B
G A IN
POT +12 VD C
H I/L O CENTER FREQUENCY
G A IN A D JU S T M E N T
S W IT C H
3 dB P IN 30 dB 50 dB
P IN P U T PHASE RF IF VCO
LPF A2
T ATTN A D JU S T D R IV E
6 dB
A1
LO
SW 1 SW 2 3 dB
A3 P
T
T R A N S M IT T E D P O W E R
NETW ORK A4
V C O /P L L
50 POW ER M ETER A M P L IT U D E A N D
IN T E R F A C E PHASE CONTROL SW 3
PW R LLR F
M ETER D R IV E
20 dB 10 dB P IN VECTOR STEP
6 dB S W IT C H M O D ULATO R ATTEN
50
16 dB
10 dB A3 P VECTOR
10 dB R
P IN P U T M O D ULATO R R F O N /O F F
R 10 dB
D R IV E P C B S W IT C H
PW R
M ETER 6 dB
A3 P P /O X T A L B U F F E R
10 dB I
P IN P U T P , P , P , LO O P ERR
T R I
I 20 dB
PW R
M ETER
FREQ P /O A N A L O G I/O D IO P /O A N A L O G I/O
FREQUENCY R A D IA T IO N L E V E L
PW R1 G P IB P /O A N A L O G I/O
COUNTER PW R2 C O M P U T E R IN T E R F A C E H E L IU M P R E S S U R E
Figure 3: Block diagram of a typical low level RF system used for cavity testing.
location so that it could be thermally stabilized. Power Meter Interface
Additionally, having it external allows one to use an
The two critical considerations for the power meter
alternate VCO for different applications. The 3 dB and
interface are stability of the components and low VSWR
50 dB attenuators are carefully selected as part of the
in the power meter signal path. Neglecting either of these
system optimization. The 50 dB attenuator is a major
will lead to unnecessary errors in the measurements. In
contributor to setting the loop gain. In this case the VCO
addition to the function of protecting the power heads
is a broad band device with a sensitivity of about
from damage due to peak RF power levels, the attenuators
5.6 MHz/V, thus the high value of attenuation. As a rule
act as matching devices which absorb any reflections due
of thumb, the attenuator should be chosen to be 10 dB
to VSWR missmatch before they have a chance to
greater than the value at which the phase loop just starts to
multiply. The crystal detectors are provided so that the
oscillate, but still low enough that the loop will lock and
signals can be observed with an oscilloscope. Typically
have a moderate lock range.
the three traces that are observed are the transmitted
The 3 dB attenuator and the 30 dB coupler were chosen
power signal, the reflected power signal and the VCO
such that the crystal detector is not power starved and is
drive signal. The auxiliary RF ports on the transmitted
still in the square law range when the loop is locked and
and reflected power signals are provided so that a
not oscillating. While the incident and reflected power
spectrum analyzer may be used rather than the crystal
crystal detectors may be operated beyond the square law
detectors.
range without compromising the measurements, one must
be careful to ensure that the transmitted power crystal Interlocks
detector is operated in the square law range, (i.e. between
During vertical testing medium power amplifiers
10 and 25 mV at the output, depending on the detector
between 100 W and 500 W are used to drive the cavities.
and load combination) when making a decay measure-
No cavity protection interlocks are used during these tests
ment. As a matter of convenience and in order to ensure
at Jefferson Lab. Each facility and test should be
that there is adequate voltage available at the inputs to the
evaluated individually. Engineering and administrative
data acquisition card in the computer, the gain of the A3-
controls have been implemented to mitigate the hazards
amplifiers was set to 400. Front panel connections
associated with ionizing radiation produced by field
provide an easy means to observe these signals using an
emission. The two major engineering controls are lead,
oscilloscope. The A4 amplifier may be necessary
steel and concrete shielding surrounding six of the eight
depending on the output level of the VCO and the input
vertical dewars as well as a personnel safety system
requirements of the LO port on the mixer. In most
(PSS). The PSS interlocks have been implemented so as
instances it is not necessary.
to not allow one to apply high power RF to a cavity unless
Amplitude and Phase Control the PSS system can confirm that the shield lid is closed
and no radiation is detected by the general area monitors.
The output of the VCO is routed to the amplitude and
AC
phase control section. The first device in that section is a
PSS HPRF
circulator. It is there to ensure that frequency pulling due
to impedance mismatches is minimized. Directional R F S W IT C H
CONTROL
couplers are placed in the circuit to couple power out for
the mixer local oscillator and the frequency counter. In PSS HPRF
KLYSTRO N 40 dB 40 dB
PM
IN T E R L O C K
A R C , IR CH1
6 dB S W IT C H
CO UPLER VACUUM
R F IN T E R L O C K S
B E A M L IN E V A C U U M H IG H B A Y (K L Y S T R O N A R E A )
H E L IU M P R E S S U R E 20 dB
C O U PLER W ATER FLO W -1 0 d B
PM FOW ARD POW ER 20 dB
L IQ U ID L E V E L
CH2
10 dB*
10 dB*
D IG IT A L
STO RAGE
O S C IL L O S C O P E -1 0 d B
PM REFLECTED PO W ER
CH3
-1 0 d B 10 dB 1 W
PM F IE L D P R O B E P O W E R
CH4
10 dB 20 W
PM H IG H E R O R D E R M O D E A -C O U P L E R P O W E R
R F CO NTRO LS
CH5
AND MEASUREMENT
CO M PUTER
10 dB 20 W
PM H IG H E R O R D E R M O D E B -C O U P L E R P O W E R
CH6
KLYSTRO N
TEST TEST
RF SOURCE P R O T E C T IO N
C ELL C ELL
C IR C U IT S
P IN A T T N P IN S W IT C H
VACUUM
+ KLYSTRO N
GAUGE -2 .5
RF
CONTROL
C O N TR O LLER
+
+ L IN E A R IZ A T IO N A R C , C O O L IN G W A T E R , E T C
+1 IN T E R L O C K S
VACUUM C IR C U IT
GAUGE + -2 .5
C O N TR O LLER VACUUM
+ CONTROL
-5 V
VACUUM
S E T P O IN T VACUUM
READBACK G A IN
-V READBACK
dI
_
I dt dQ
+
I
I d
dt
dt
RF LO
cos 0 t t
1 kH z
+ LPF
0
10 dB STABLE RF 1 Hz
3 dB +
SOURCE LPF
L IM IT E R
90
RF LO dI
Q
dt
I
Q dQ
_
+ dt
1.4
from one end. In vertical tests the incident power cable
1.2
the field prob cable as well as any higher order mode
1 30 (HOM) coupler cables all have sections in the helium
35
0.8 bath. In cryomodules tests all of these cables with the
46
0.6 120 exception of the incident and reflected power also have
0.4 one end within the vacuum vessel. The only way to
180
0.2 208 measure the losses of a cable within a cryostat is to do a
two loss measurement either with a calibrated network
0
analyzer; or a source, a circulator and a power meter. The
0 100 200 300 400
VIBRATION FREQUENCY (Hz)
following are procedures used at Jefferson Lab for
calibrating cables during a vertical test. See Figure 12 for
Figure 11: An example of background microphonic noise the cable designations.
as measured with a cavity resonance monitor. The lower
trace is an FFT of the time domain plot shown in the To calibrate the cable from point A to point C. First,
upper trace. measure the one way loss of cable B-C.
- Measure the reference source power level with the
When building a cavity resonance monitor one must be reference power meter. (P1)
very careful that the cutoff frequency for the integrator - Connect the reference source to point B of cable
and low pass filters are precisely controlled and matched. B-C.
If there is an offset between the frequency of the stable - Measure the power level with the transmitted
source and frequency being measure, the I and Q signals power meter. (P2)
have a large magnitude component at that difference - The one way loss is P1-P2 (dB)
frequency. At higher difference frequencies, the second
harmonic components of I and Q bleed through to the
Measure the two way return loss of cable A-B
output giving false frequency components. A 5%
- Connect the reference source to the input terminal
mismatch in filter cutoff frequencies or gains will cause
of the circulator.
bleed through problems when the difference frequency is
- Connect the reference power meter to the load port
larger than 10% of the filter cutoff frequencies. During
on the circulator.
operation it is difficult to distinguish this bleed through
- Record the reading on the reference power meter
signal from a driven microphonics signal.
with the output port of the circulator open. (P3)
A new DSP based system is currently under
- Connect the output port of the circulator to port B
development at Jefferson Lab that shows a great deal of
of cable A-B and record the reading on the
promise.[11] Using a CORDIC [12, 13] algorithm for
reference power meter. (P4)
phase determination and a high resolution analog-to-
R EFLEC TED
R F D R IV E
POW ER
SOURCE
M ETER
P 3 P 4 . POW ER
M ETER
POW ER
M ETER
P1 P 2 P 3 P 4 30 dB
C AC (dB) (12) T Y P IC A L
2 REFERENCE
POW ER
B E M ETER
P 5 P 6 P 7 P 8 P 9 P 10 2.0%
C INC (dB) (13)
2 0.0%
P 6 P 5 3 * P 7 3 * P 8 P 9 P 10 -2.0%
C REFL (dB) (14)
2
-4.0%
Calibration Verification
-6.0%
Two ways to verify calibration procedures are to
calibrate the system using an external cable in place of the -8.0%
cable within the dewar then do one or both of the 804.0 804.5 805.0 805.5 806.0
following. For the field probe cable calibration and FREQUENCY (MHz)
reflected power calibration inject a known signal level into Figure 13: Difference between RF power measurements
the external cable and measure the power using the calibrated at 805 MHz and those taken at nearby
calibrated power meter. For the forward power frequencies for several different signal paths.
calibration connect the external cable to a remote power
meter; inject a signal into the drive cable using the RF CABLE BREAKDOWN
drive source; and measure the power using the remote
When performing vertical tests at 2 K the incident
power meter and the incident power meter. In both cases
power cables must pass through the low pressure helium
it can be a good exercise to vary the frequency over an 1
gas in order to get to the fundamental power coupler.
MHz to 2 MHz range and compare the values over the
The pressure at which helium goes superfluid is 35 Torr, repeat the above steps at moderate power levels to ensure
and systems are typically operated at pressures between that the lost power is consistent with the error associated
20 and 25 Torr. Operating in this pressure regime with the with the measurement.
typical dimensions of medium power RF connectors (i.e.
1 mm to 5 mm) means that the connectors are operating at System Improvements
or near the Pasching minimum of 4 Torr-cm. At this The primary solution to the problem of cable
pressure-distance product the breakdown voltage in breakdown is to never make RF connection rated for more
helium gas is minimized [14], and thus, there is a that a few watts in low pressure helium gas. At Jefferson
reasonable probability that glow discharges will occur. Lab we use silicon dioxide dielectric, stainless steel
Work done by MacDonald and Brown in 1949 indicated jacketed cables manufactured by Meggitt Safety Systems.
that the minimum pressure for RF breakdown in helium The variant that we use has the outer conductor welded
was between 8 and 30 Torr depending on the into a conflate flange. They make use of a hermetic fret to
geometery.[15] The probability of breakdown is further seal the exposed cable dielectric at connectors at each end
enhanced by field emission radiation. In general, this of the cable. This ensures that the high power connections
phenomenon is not new and has been extensively studied are only made in liquid helium and that there is no path for
by individuals working in the satellite industry where it is contaminating gas through the cable dielectric. Other
known as multipactor breakdown. [16] solutions involve using epoxy to seal the outer jacket of
the cable in a flange; building a secondary volume around
Observations the air side connection; and backfilling that space with
RF discharges have been observed in gas with as little helium gas so that any gas that leaks down the inner
as 10 W at 1500 MHz. Discharges have even occurred in dielectric of the cable does not contaminate the helium
connectors that were completely immersed in superfluid system or to machine the outer conductor and dielectric
helium with incident power levels on the order of 150 W, off of the cable and epoxy it into the a metal sleeve as part
full reflected, at the cavity. In all cases, discharges and of the feed through assembly.[17]
the resulting damage have been observed in the volume For the connections made in the helium bath, we also
within the connector space and in the connector back shell vent all connector volumes to the helium bath to improve
space. Both of these volumes do not contain dielectric the heat conduction of the space, especially the connector
materials. back shell space by drilling four holes, 3 mm in diameter
Once a breakdown is initiated it will be sustained in in the outer conductor of the connector or back shell.
superfluid helium by the forward power even at levels Although improving the situation, connectors with this
down to 10 W. The theory is that a few watts of heat is modification have been known to fail at power levels on
produced in the connector, possibly through thermal the order of 200 W at the cavity when in a liquid helium
conduction down the insulated center conductor, from the bath. Another approach is to fill all potential spaces with
antenna within the cavity, or in the connector pin itself. an insulating material. In theory this should work but we
The liquid helium flashes to gas within the connector have only had limited success at 200 W, 805 MHz
space and breakdown occurs within the newly produced forward power. One option that we have pursued but not
low pressure gas volume. From the perspective of RF fully implemented is to pressurize the cable with helium
measurements, such events appear to be Q-switching gas above the triple point of helium. This volume
within the cavity. The measured gradient appears to be includes the connection to the vacuum feed through at the
reduced and the Q 0 as calculated using the dissipated coupler antenna. Most important is to critically couple
power will be reduced substantially. the cavities by carefully adjusting the input antenna or by
using a variable coupler so that you do not have to use
Determining if Breakdown Occurs more than 150 W of RF power at the cavity.
To determine if you have a cable discharge while it is
occurring. FUNDAMENTAL EQUATIONS
During the cryomodule production cycle there are two
Detune the frequency of the LLRF sytem far enough basic types of high gradient RF tests that are done on cold
to lose lock in the VCO-PLL. cavities. In the first test the basic RF properties such as
Measure the forward and reflected power. maximum accelerating gradient, field emission onset, and
Subtract the calibrated forward power from the quality factor, Q0, as a function of gradient are
calibrated reflected power to calculate the lost power. determined. At Jefferson Lab, as is done at many other
If any significant power is being lost, i.e. much labs, these tests are done in test cryostats where the
beyond the errors in the measurements, you probably cavities are held vertically. Ideally, these tests are done at
have a glow discharge in one of the connectors. or near critical coupling. In this way the RF source
requirements are only a few hundred watts, which is just
On occasion connectors damaged from this mechanism enough power to overcome the wall and field emission
will exhibit this anomalous loss permanently at all power losses.
levels. Therefore, one should turn of the RF power and
In the second type of test the cavities are installed in the Superconductivity for Accelerators, by Padamsee,
final cryostat and they are typically strongly over coupled. Knobloch and Hayes.
This presents a problem as the errors in lost RF power get The Appendix of this document contains a complete set
excessive when 95% to 99.9% of the incident power is of equations used for making cavity measurements, near
reflected back out of the fundamental power coupler. critical coupling using the decay method and the CW
Thus, during cryomodule tests the RF heat load is method as well as for making measurements of strongly
measured calorimetrically. over coupled systems. Additionally, the appendix
contains equations used for calculating the errors
Table 1: Common variables used when discussing associated with the calculated values.
superconducting RF cavities. The critical variable for calculating the RF parameters
Symbol Variable Name Units of a superconducting cavity is the shunt impedance, which
relates the stored energy to the effective accelerating
r /Q Geometric Shunt Impedance Ω/m gradient, peak electric field, and peak magnetic field for
G Geometry Factor Ω any given mode. It is determined using electromagnetic
simulation tools such as Mafia or Superfish. One should
E Electric Field V/m be careful in applying this variable as there are different
L Electrical Length M definitions of shunt impedance, R, and geometric shunt
impedance, (r/Q), in use[21]. For this paper, both
0 Cavity Frequency s-1 2
variables are based on the definition that R V P ,
U Stored Energy J which includes the transient time factor.
rs Surface Resistance Ω General RF Measurement Equations
TC Critical Temperature K The following are general RF equations that apply to
SRF cavities:
PX RF Power at Port X W 2
E L
U (15)
Pemit Emitted Power W 0 r / Q
2
R Shunt Impedance Ω U0 E L
P (16)
T Operational Temperature K Q Q r / Q
Q 0 G / rS || Q ElectronLo ading (17)
rresid Residual Surface Resistance Ω
2
f 1 . 95 T c / T
Q0 Intrinsic Quality Factor rS 10 4 ( K / GHz 2 ) e rresid (18)
T
Fundamental Power Coupler Q L Q 0 || Q FPC || Q FP Q FPC (19)
Q FPC
Coupling Factor RC Q L (r / Q ) (20)
Q FP , Q 2 Field Probe Coupling Factor I M E / RC (22)
RC Coupling Impedance Ω/m
The power delivered to the beam is:
I Beam Current A
PBeam LEI (23)
IM Matching Current A
Pdisp Dissipated Power W The coupling factor, β, is a measure of the efficiency of
coupling RF power into the system it is given by:
Decay Time s 1 C Preflected / Pincident
Shunt Impedance Per (24)
r Ω/m 1 C Preflected / Pincident
Unit Length
where C is 1 for the under coupled and -1 for the over
Summary of Variables Names and Units coupled case. For a cavity which are perfectly tuned and
Table 1 is a listing of the variables commonly used with the beam on crest, the power required by the klystron
when discussing superconducting cavities, their names is:
and associated units. The equations that follow were 1 L E IR C 2
extracted from several sources over the years [18, 19, 20]. PKly (25)
4 RC
They are the basis for many of the RF measurements and
associated calculations associated with SRF cavities. A 1 1 L
E IQ L r/Q 2 (26)
good general reference for this material is entitled RF 4 Q L (r / Q )
The Power reflected back to the circulator is: ( A) 1 A
(36)
PRef
1 L E IR C 2
(27) A 2 A
4 RC ( AA ) A
1 1
2 (37)
L
E IQ L r/Q 2
(28) AA A
4 Q L (r / Q ) The first two equations assume that the errors are
The time dependent complex differential equation Gaussian and uncorrelated. The factors of ½ and 2 found
where K is the incident wave amplitude in Watts , d in equations (36) and (37) are because the errors are
correlated. There are occasions, for instance the emitted
is the (time varying) detune angle, and f 0 / 2Q L :
power measurement, when using the simple equations is
ω 1 dE RC
not appropriate and can lead to non causal errors. In such
1-j d E 2K RC I (29) cases it is a simple matter to perform a Monte Carlo
ωf ω f dt L
calculation to determine the dependencies. Additionally,
Adding microphonics and the effects of the difference care must be taken when chaining together calculations.
cavity center frequency f 0 and that of the RF source, f , For instance, determination of Q FP for a strongly over
and the beam current, I, being off crest by B leads to the coupled cavity includes any fixed calibration error in the
equation (30) as the power required of the klystron.[20] transmitted power signal. If the calculated value of Q FP
is used to later calculate the gradient based on the
E I R cos 2 transmitted power signal (using the same cable calibration
1
0 C B
factor), the error in the gradient should not include the
2
L
PKly (30)
transmitted power cable calibration factor. In this case the
RC 4 2 Q f E I R sin
L 0 C B error in the transmitted power based gradient should only
f0
contain the error in the linearity of the power meter
measurement and the error associated with Q FP .
When testing strongly overcoupled cavities 1 ,
Q L Q 0 and Q L Q FP which means that
CRITICALLY COUPLED CAVITY
Q L Q FPC , in this case.
MEASUREMENTS
Q L 2 (31)
When a cavity is near critical coupling, the process for
2 4 (r / Q ) determining the cavity parameters of E , Q 0 , and Q FP is as
E PIncident Q L (32)
(1 ) L follows. The RF frequency and phase are controlled by a
4 VCO-PLL. The phase is carefully adjusted to minimize
Because 1 , 4 and, in this case.
1 the reflected power, which also maximizes the transmitted
(i.e., the field probe) power. Then a decay measurement
(r / Q )
E 4 PIncident Q L (33) is made which determines the values of E , Q 0 , and Q FP .
L
Once the results of the decay measurement is completed,
Although using the above forward power to calculate
Q FP is used along with the RF power measurements to
gradient is a reasonable technique, practical experience
says that there can easily be as much as 25% difference calculate E and Q 0 .
between the gradient as calculated using the forward
technique power and the emitted power technique or The Decay Measurement
using a well calibrated field probe. This difference can be The decay measurement is initiated by pulsing RF
reduced by properly tuning the phase locked loop, power on and off so that one can determine if the cavity is
variable frequency system, or the cavity for a fixed over coupled or under coupled. Figure 14 shows the
frequency system. shape of the reflected power pulse for the different
coupling conditions. Stable gradient is established and
Error Analysis the steady state forward, reflected and transmitted power
Most of the error analysis done when making cavity levels, ( PFwd , PRef , and PTran respectively) are recorded.
measurements can be done using a few fundamental
Next, the cavity drive signal is turned off and the decay
equations as follows:
time constant, , of the resulting transmitted power
( AB ) A
2
B
2
transient is determined using a crystal detector, a spectrum
(34) analyzer or a pulsed RF power meter.
AB A B
The decay time, , is used to calculate a value for the
2 2
( A B) A B loaded-Q, Q L . Q L , PFwd , PRef , and PTran are used to
(35)
A B A B 2 calculate Q 0 . The gradient may then be calculated as:
(r / Q ) RF power meter as discussed earlier or using a spectrum
E PLoss Q 0 (38) analyzer set up to do a zero span, time domain
L
measurement. The crystal detector measurement relies on
and the field probe-Q can be calculated as: the fact that the crystal detector is operating within the
2 L square law range. In this power, or output voltage, range
Q FP E PTran (39)
r / Q the output voltage is proportional to the RF power.
As stated earlier, the operator must determine if the Using the half power, decay time constant technique, a
cavity is over coupled or under coupled prior to properly terminated crystal detector can be used to make a
calculating the cavity parameters using the decay +/- 3% measurement of the cavity decay time constant if
measurement technique. Typically a crystal detector is the peak detector voltage is below 10 mV. If, for
placed on the reflected power signal and the resultant example, the same crystal detector were inadvertently
signal is observed on an oscilloscope. Figure 14 is a used at 100 mV, the measured decay time would be
depiction of the reflected power waveforms produced by a overestimated by about 40%, the calculated Q 0 would be
properly tuned cavity under different, near critical 40% higher than the actual value and the calculated cavity
coupling, conditions. gradient would be 18% higher than the actual value.
In all of the reflected power pulses, the first peak has Another source of decay measurement errors is changes
the same magnitude as the reflected power signal when in the loaded-Q during the decay measurement. Usually
the cavity is detuned. When the cavity is over coupled, this is due to non linear effects such as field emission
the emitted power pulse, i.e. the second peak, is larger loading. As the energy stored in the cavity is emitted out
than the first peak. When the phase loop is properly tuned of the fundamental power port, the gradient in the cavity
the minimum after the first peak goes through zero. is reduced; the field emission loading is reduced; and the
When the cavity is critically coupled, the leading and loaded-Q is increased. This also occurs if the cavity has a
trailing peaks are of equal magnitude and the reflected strong Q-slope. The logarithmic slope of the decaying
power goes to zero at the end of the pulse. When the power is . In general is a function of E, or
cavity is under coupled the secondary peak, if there is equalivently Q 0 E . In such cases the decay slope at the
one, is smaller than the first peak and the reflected power
signal does not go to zero during the pulse. start of the decay must be used, or a systematic error will
lead to calculated Q 0 and E values that are larger than
the actual values.
Lost Power Measurement Errors
O VER C O U PLED >1 UNDER CO U PLED < 1 /3 Because the lost power is a difference between three
power meter measurements, the error is given by the
following:
2 2 2
C R IT IC A L L Y C O U P L E D =1 U ND ER C O U PLED 1< < 1 /3
PLoss PFwd PRef PTran
(42)
PLoss PFwd PRef PTran 2
Thus the error in the lost power increases dramatically
FORW ARD POW ER F IE L D P R O B E P O W E R
when the reflected power approaches that of the forward
power. Remember, when the cavity is critically coupled
Figure 14: Upper four traces are depictions of reflected
1 ; the reflected power is equal to zero and virtually
power waveforms for various coupling states. The lower
two traces are the corresponding forward and field probe all of the forward power goes into wall heating. As 1
pulse shapes. increases much above three or below one third the
reflected power starts to become a substantial fraction of
CW Measurements the wall heating power and the error in the lost power
Once a value has been determined for the field probe-Q increases. This is the major contributor to the error is CW
the calculations become much simpler. The gradient is Q 0 measurements and decay measurement based of the
given by: gradient. Figure 15 are plots of the error in gradient and
(r / Q ) Q 0 as a function of . The calculations assume that the
E PTrans Q FP (40)
L power meter measurements, including cable calibrations
and the quality factor is given by: are 7 % , the linearity of the power meters is 2 % , that
2 L is known to 3 % and that the power meters are
Q 0 E PLoss (41)
r / Q operated well above their noise floor. Under these
conditions, the error in the decay based gradient
Decay Measurement Errors measurement and the CW Q 0 measurement vary because
Crystal detectors are frequently used to measure the of the errors in the lost power calculation. Thus it is best
decay time, . Other alternatives included using a pulsed to try to make all of the measurements when 0 . 5 2 .
25% are introduced by the sampling system that can be reduced
(C) by proper choice of system parameters.
20% The uncertainty in the stored energy is given by the
Error In Calculated Value
following:
15% 2 2
U C R PCAL t ( N m ) C R Pmin
(B) U (45)
10%
( Pem itted ) m t ( Pem itted ) N
L E IR C
2
1 L
needed from the klystron E IQ L r / Q
2
4R C 4Q L ( r / Q )
L E IR C
2
1 L
reflected to the circulator E IQ L r / Q
2
4R C 4Q L ( r / Q )
Time dependent, complex diferentia l equation where K is the incident w ave amplitude
ω 1 dE RC
1-j d E 2K RC I
ωf ω f dt L
The equation for the power required for cavity center frequency f 0 detuned by f
and beam current, I 0 , off crest by B :
2
1 L f
E I 0 R C cos B 2 Q L
2
PKlystron E I 0 R C sin B
4 Q L (r / Q ) f0
Theory and Practice of Cavity RF Test Systems - Appendix A
Decay fit Parame ter for the P Reflected Signal Fit Starting at index point m
r / Q ( / m ) Cavity Shunt Impedance
L ( m ) Cavity Active Length
U ( J ) Cavity Stored Energy
f 0 ( Hz ) Cavity Res onant Freq uency
N - Data set sample size for pulsed measuremen t
m - Data Sample Set Index When the Incident P ower is Removed From the Cavity
k - Data Sample Set Index When the Incident P ower is Applied to the Cavity
Δt - Data sample interval.
T - Period of the pulsed waveform (Typically 60 Hz or less)
E (V/m) - Cavity Gradient
E emit , trans , fpwr ( V/m) - Cavity accelerati ng gradient as indicated by subscript
Ports - number of RF ports on the cavity, includes beam pipes, HOM couplers, etc.
Pincident Pim C i
Preflected Prm C r
Preflected
Pincident
1
( Overcouple d Cavity)
1
Q L 2 π f 0
Theory and Practice of Cavity RF Test Systems - Appendix A
N
Ports
U Pem it j i * t
j 1 im
Pemit j is the
th
Where power emitted from the j port on the cavity.
r /Q
E em it 2 F * U *
L
Preflected
Pincident
1
Note : 1 for strongly over coupled cavities
1
For a perfectly tuned cavity :
4 r /Q
E * Pincident * Q L *
1 L
Assuming that 1 and that the cavity is perfectly tuned.
r /Q
E fpwr 4 Pincident * Q L *
L
2 πf0 * U
Q tran
Ptransm itte d
2 πf0 * U
Q hom a
Phom a
2 πf0 * U
Q hom b
Phom b
r / Q
E trans Q ext 2 * Ptransm itte d *
L
Assumes known value of Q ext 2 which is typically based on an emitted
power measuremen t.
Theory and Practice of Cavity RF Test Systems - Appendix A
Q0 Measurements
CW measurements where Pdissipated is the average dissipated power measured calorimetrically.
2
E L 2 πf 0 * U
Q0 *
Pdissipated r /Q Pdissipated
For pulsed operation the gradient is not constant throughout the measurement. In this case the
field probe transmitted power is recorded as a function of time with at a sample interval t; the
gradient can be calculated using the transmitted power method; and Q0 is calculated as:
N
1 2
T
E trans i * t
ik L
Qo *
Pdisapated r /Q
Where Pdissipated is the average dissipated power measured calorimetrically, and T is the period of
the pulses. It should be noted that the numerator in the above equation is used to account for the
non square pulse shape. Values of Q0 calculated using this method will be different for different
gradient pulse shapes or CW operations. If CW values are desired, it is best to make such
measurements with pulse widths that are much greater than the cavity fill times.
To measure the dissipated power calorimetrically one isolates the cryomodule from the helium
supply and return lines and records the rate of rise of the pressure under three conditions. These
are static heat load with RF and resistive heaters on; static plus a known resistive heat load
applied to the bath; and static plus a unknown RF heat load due to the cavity losses. The
dissipated power is then calculated using the following.
dP dP
dt RF ON dt STATIC
PDISSIPATED P
dP dP HEATER
dt HEATER ON dt STATIC
dP
where is the rate of rise of the pressure under the different conditions .
dt
Theory and Practice of Cavity RF Test Systems - Appendix A
Derivation of Measurement Errors - Cryomodule:
Starting Parameters:
Pmin Sensitivit y limit of power sensors used
Pcal Fractional uncertaint y in absolute power measured
C Fraction al uncertaint y in cable calibratio ns
PLin Fractiona l uncertaint y of the linearity of the power meter measuremen t
Δ Error of a variable . The units are the same as the variable, i.e. Ptran is the error
in Ptran in Watts.
Parameter Uncertainties:
2
2 Pim
Pincident Pincident ( C )
Pim
2
2 Prm
Preflected Preflected ( C )
Prm
2
2 Ptm
Ptransmitte d Ptransmitte d ( C )
Ptm
2
2 Ph am
Phom a Phom a ( C )
Pham
2
2 Ph bm
Phom b Phom b ( C )
Phbm
2
2 ( Prm ) i
( Pemit ) i ( Preflected ) i ( C )
( Prm ) i
2 2
Preflected Pincident
P P
2
reflected incident
Theory and Practice of Cavity RF Test Systems - Appendix A
2 2
U U C r Pcal ( Pem it ) m t P N t ( N m ) C r Pmin
P N term from not integratin g to
(( Pem it ) m t ) Term is due to jitter for starting the integratio n
t ( N m ) C r Pmin Term is due to power meter noise floor
E em it U
E em it
2 U
2 2
E fpwr Q L Pincident
E fpwr
Q
2 P
L incident
2 2
E em it PtransX
Q extX Q extX 4*
E P
em it transX
2 2
E trans Q transX PtransX
E transX
2 Q P
transX transX
Where Q trans and Q trans were entered values that were determined under different
operating conditions and calibrations than the current measurement.
2 2
E trans Q trans PLin * PtransX C X * Pmin
E trans
2 Q Ptrans
trans
Where Q trans and Q trans were determined under same operating conditions and
calibrations than the current measurement.
2 2
2 E fpwr Pdissipated
Δ Q 0 ( fpwr Q 0 ( fpwr
) ) E P
fpwr dissipated
2 2
2 E trans Pdissipated
Δ Q 0 ( trans ) Q 0 ( trans )
E P
trans dissipated
2 2
1 1
Δ
Q L Q L Q L Δ P Lin
Theory and Practice of Cavity RF Test Systems - Appendix A
Starting Parameters:
Pim , Prm , Ptm , Pha , Phb Actual measured CW RF power (Watts), just prior to turning the
RF OFF for the decay measuremen t
C i , C r , C t , C a , C b Cable calibratio n factors relating measured RF power to associated
RF power at the cavity.
C Over/under coupling factor 1 for under coupled, - 1 for over coupled.
Decay fit darameter (seconds) for the emitted (as measured at the reflected power signal)
or the transmitt ed power signal as measured when the incident R F power has been
turne d OFF.
f 0 ( Hz ) Cavity resonant frequency
Pincident Pim C i
Preflected Prm C r
Preflected
Pincident
C
* 1
1
Q L 2 π f 0
This implicitly assumes that Q0 and the external-Q of all of the RF ports are independent of the
stored energy, i.e. linear, flat Q0 , and constant coupling factor for gradients at or below that of the
starting point of the decay measurement.
Theory and Practice of Cavity RF Test Systems - Appendix A
Decay Measurement Formulas Continued
* QL
Q
*
1
Q0
Q1
1
Ptransm itte
2 d
Ploss
Phom a
3
Ploss
Phom b
4
Ploss
*
1 (1 2 3 4 )
Q 0 (1 1 2 3 4 ) Q L
Q0
Q2
2
Q0
Q3
3
Q0
Q4
4
r / Q
E acc ( V/m ) Q 0 P Loss
L
Q 0 P Loss
U ( Joules )
2 πf 0
Theory and Practice of Cavity RF Test Systems - Appendix A
CW Formulas
Starting Parameters:
Pim , P rm , Ptm , Pha , Phb Actual Measured CW RF Power (Watts), just prior to turning the
RF OFF for the decay measuremen t
C i , C r , C t , C a , C b Cable Calibratio n Factors relating measured RF power to associated
RF power at the cavity.
C Over/Under coupling factor 1 for under coupled, - 1 for over coupled.
Decay fit darameter (seconds) for the emitted (as measured at the reflected power signal)
or the transmitt ed power signal as measured when the incident R F power has been
turne d OFF.
Q 2 Transmissi on Probe External Q as determined from a previous Decay measuremen t
f 0 ( Hz ) Cavity Res onant Freq uency
Pincident Pim C i
Preflected Prm C r
Preflected
Pincident
C
* 1
1
Q 2 Ptransm itte d
Q0
Ploss
Q0
Q1
1
Q 2 Ptransm itte d
Q3
Phom a
Q 2 Ptransm itte d
Q4
Phom b
Theory and Practice of Cavity RF Test Systems - Appendix A
CW Measurement Formulas Continued
Q0
2
Q2
Q0
3
Q3
Q0
4
Q4
1 2 3 4
*
1
Q0
QL
1 1 2 3 4
r / Q
E acc Q 2 Ptransm itte d
L
Q 2 Ptransm itte d
U ( Joules )
2 πf 0
Theory and Practice of Cavity RF Test Systems - Appendix A
Derivation of Measurement Errors - Decay Measurement:
Starting Parameters:
Pim , Prm , Ptm , Pha , Phb - Actual Measured CW RF power meter reading (Watts), just
prior to turning the RF off for the decay measurement.
Pmin - Sensitivity limit of power sensor used.
Pcal - Factional uncertainty in absolute power measured.
PLin Fractional uncertainty in the linearity of the power meter calibration
C - Fractional uncertainty in cable calibrations.
Δ - Error of a variable. The units are the same as the variable, i.e. ΔP tran is the error
in Ptran in Watts.
Parameter Uncertainties:
2 2
Preflected Pincident
2 P P
reflected incident
Theory and Practice of Cavity RF Test Systems - Appendix A
Derivation of Measurement Errors - Decay Measurement Cont.:
2 2
* *
1 1
Q L Q L
2
* Q L
2
Q
*
Q
*
*
1 QL
2 2
Ptransm itte d Ploss
2 2
P P
transm itte d loss
2 2
Phom a Ploss
3 3
P P
hom a loss
2 2
Phom b Ploss
4 4
P P
hom b loss
2
* 2 2 2
1 1
2 3 4
1 2 3 4
* 2
2
2 2 2 2 Q
Q 0 Q 0 1 2 3 4 L
1 2 QL
1 2 3 4
2 2
Q 0 1
Q1 Q1
Q
0 1
2 2
Q 0 2
Q 2 Q 2
Q
0 2
2 2
E Q 0 Ploss
E
2 Q P
0 loss
Theory and Practice of Cavity RF Test Systems - Appendix A
Pim , Prm , Ptm , Pha , Phb Actual Measured CW RF power meter reading (Watts).
Pmin Sensitivity limit of power sensor used.
Pcal Factional uncertainty in absolute power measured.
PLin Fractional uncertainty in the linearity of the power meter calibration
C Fractional uncertainty in cable calibrations.
Δ Error of a variable. The units are the same as the variable, i.e. ΔP tran is the error
in Ptran in Watts.
ΔQ 2 Uncertainty in Q 2 as determined from a decay measurement.
Parameter Uncertainties:
2 2
Preflected Pincident
2 P P
reflected incident
Theory and Practice of Cavity RF Test Systems - Appendix A
Derivation of Measurement Errors - CW Measurement Cont.:
2 2
Q 2 2 Ploss
Q 0 Q 0
Q
PLin
P
2 loss
Q 0 error assumes that the Q 2 used was measured using the same power meter and cable
calibrations.
2 2 2
Q 2 Ptran Ploss
Q 0 Q 0
Q
P P
2 tran loss
Q 0 error assumes that the Q 2 used was measured using different power meter and cable
calibrations.
2 2 2
Q2 Phom a Ploss
Q3 Q3
Q P P
2 hom a loss
2 2 2
Q2 Phom b Ploss
Q4 Q4
Q P P
2 hom b loss
2 2
Q0 Q2
2 2
Q Q
0 2
2 2
Q0 Q3
3 3
Q Q
0 3
2 2
Q0 Q4
4 4
Q Q
0 4
2 2
* *
1 1
2
* 2 2 2
1 1 2 3 4
1 2
* 2 3 4
2 2
Q0 1
Q1 Q1
Q
0 1
2
Q0 2 2 2 2
QL QL 1 2 3 4
Q 1 2
0 1 2 3 4
2 2
E Q2 Ptransm itte d
E
2 Q P
2 transm itte d