Transcript Simulating Noise and Crosstalk problems with GEM detectors

Simulating Noise
and Crosstalk
problems with GEM
detectors
Matti Rahkala, HIP.
Part One
Comparing of different Order Low Pass Filters.
Smoothing High Voltage source output.
Frequency response of 1 to 4 Order
RC-filters. Equal amount of the sum of
Resistance and Capacitance
1.0V
0.8V
0.6V
0.4V
0.2V
0V
100Hz
V(Out4)
V(Out3)
1.0KHz
V(Out2)
10KHz
100KHz
V(Out1)
Frequency
1.0MHz
10MHz
The same, but at Log scale.
3800 times improvement (72 dB) at 1 MHz
between 1. and 4. order filter
10V
1.0V
(1.0000M,1.5916m)
1. Order Filter
1.0mV
1.0uV
(1.0000M,417.920n)
4. Order filter
1.0nV
10pV
100Hz
V(Out4)
V(Out3)
1.0KHz
V(Out2)
10KHz
100KHz
V(Out1)
Frequency
1.0MHz
10MHz
Impulse response. Equal V(t)*dt
integrals. Decay time vs. Amplitude
1.5V
4. Order
3. Order
2. Order
1.0V
1. Order
0.5V
0V
0s
V(Out4)
50us
V(Out3)
V(Out2)
100us
V(Out1)
150us
200us
Time
250us
300us
350us
400us
Part two. Noise and Crosstalk
simulation of the GEM Strips
Frequency response of the GEM
Strips
1.0KV
Output
1.0V
Strip Voltage
First adjacent Strip Voltage
100uV
2. adjacent Strip
10nV
3. adjacent Strip
1.0pV
1.0Hz
V(B3)
10Hz
V(B2)
V(B1)
100Hz
V(In)
V(Out)
1.0KHz
10KHz
Frequency
100KHz
1.0MHz
10MHz
100MHz
1
Resistor – Capacitor contribution
to The output Noise Level
1.5m
2
80uV
1.428 mV RMS
Output Noise density, V/SQRT(Hz)
RMS Output total noise as fuction of upper Frequency
60uV
1.0m
40uV
0.5m
20uV
0
>>
0V
1.0Hz
1
10Hz
100Hz
1.0KHz
SQRT(S(NTOT(ONOISE))) 2
V(ONOISE)
10KHz
Frequency
100KHz
1.0MHz
10MHz
100MHz
Cont … Fighting against Noise
1. Detecting and attenuating the Common
mode Noise Current
2. Forgotten enemy – microfonic
phenomena. It needs furter
investigations with a proper analysig
tools. I propose use of Acceleration
sensors and cross correlation
measurement as an analytical tool.
3. Last but not least; Tribo electricity.
Huge problem if ignored with
measurement Setup. The Triboe effect
is easy to demonstrate. It gives Huge
noise response.
1. Power Supply Noise


Common
mode current
coming from
Power supply.
Measured by
toroidal
current
transformer. 2
mA pp (top
trace)
Corresponding
Charge
Amplifier
output noise
coupling. CM
to normal
mode Noise
pick up.
(Bottom trace)
HF-Noise envelope Before attenuation ferrites. Slow sweep,
8 mVpp Noise envelope at Amplifier output.
Sniffer tool. Simply toroidal transformer. 2 rounds
secondary winding- 50 ohm termination.
Simply filter. 1.64 mVpp noise at
output
Two
rounds
1.593
mVpp
Noise
Level
Three rounds.
nine ferrites.
1.050 mVpp
noise output
Some conclusions and
Recommendations



Don’t use switching power supplies if you can
avoid it.
CM-transformers (ferries) helps sometimes. CMnoise source impedance is too high for filters
alone. Noise Current takes lowest impedance
path to ground. The Art is: With Ferrites you can
steer CM-noise Current to choose less
hazardous path to the Ground.
Ground Noise, You can’t attenuate it much but
You can steer it to the harmless path!