Transcript UF_–PNPI_HV_system_status

UF –PNPI HV system status
August 2008
Sergey Volkov
Nikolai Bondar
PNPI
HV system scope
•
•
•
•
•
•
•
•
Control computers
Host cards
Matsusada AU-5P60 type power supply
Master crates
RDB crate
Master boards
Remote boards
LV power supply VIENER
2 pc
6 pc
2 pc
2 pc
8 pc
8 pc
56 pc
1 pc
2016 individually controlled HV channels
Analyses troubles in operation of HV system.
HV system had the following main problem – very unstable behavior.
(Host Card driver sent warnings “pre-trip” and “trip”, even at zero HV)
We found the following problems:
1.
2.
3.
Distortion of the time alignment of control signal
Extra sensitivity of the Master board to the ripples of the input HV
Infant mortality of some channels of the Master and Remote boards
Distortion of the time alignment of control signal
For proper work of the control line the time alignment of the
control pulses should be within certain limits.
The system was designed for working with control line up to 100
meter of the twisted pairs cable with the time skew of 0.1ns/m.
(Maximum time skew should be less than 10 ns.) Unfortunately our
control line consists of few parts of different cable with total length
about 100 m. As a result the total time distortion was very close to
the critical value.
This problem was solved at CMS by changing the firmware at the
Host card. Unfortunately the Host cards with new firmware doesn’t
work in our control computers.
In our case we solved this problem by putting extra delay (15 ns) of
the ADC Hold signal (CVADC in the control bus protocol).
Extra sensitivity of the Master board to the ripples of the
input HV
The output voltage ripple is specified as 0.1% for primary power supply.
At 3 KV of output voltage this value may be up to 3 V. In our case we have 0.8 V
of the ripple with 2 main frequencies - 30 KHz and 100 Hz.
30 KHz frequency is decreasing accuracy of the current measurement of
the Master board in 2 times (accuracy about 5 uA).
This accuracy is not very good, but is not a crucial. It is possible to improve
resolution by using simple LC filter.
(Output cable capacitance is about 5000 pF, it is equal 10 kOhm of AC resistance. Master
board channel also has about 100 KOhm AC resistance.)
100 Hz frequency affects to the output voltage stability because of the
feedback frequency of the Master board stabilization circuit very close to the 100
Hz. As a results the resolution of the current measurement may decrease up to
trip value.
Unfortunately the 100Hz filter for 3-4 KV is a problematic thing.
So the CMS team decided to use saturated mode of the Master board
as a main working condition. In this case the voltage drop on the Master board
channels about 100 +/- 5 V at maximum consuming current ( ~1.2mA).
To implement this mode of operation it is very important to have remote control
of the primary power supply.
Infant mortality of some channels of the Master
and Remote boards
It is almost impossible to make burning procedure for HV modules because of fast
degradation of the some HV components at high temperature under nominal voltage.
So infant mortality still presents.
Founded problems:
- 4KV regulators for Master board - broken HV transistor
- 1KV regulator for Remote board - broken HV transistor
- Not stable current sensor
- Very low resolution of the current sensor
- Unknown problem (under investigation)
Total :
4KV HV regulator
1KV HV regulator
- 2 pc
- 3 pc
- 2 pc
- 2 pc
- 2 pc
2 pc
9 pc
According experience of the CMS team we expect the floating the value of
the voltage sensing resistor. So the recalibration of the voltage sensors is
necessary during all time of operation. This procedure should be
implemented in the system.
Conclusion
1. Distortion of the control signal time alignment fixed by In our case we
solved this problem by putting extra delay (15 ns) of the ADC Hold
signal (CVADC in the control bus protocol).
2. Extra sensitivity of the Master board to the ripples of the input HV should
be fixed by use saturated mode of the Master board
as a main working condition.
To implement this mode of operation it is very important to
have remote control of the primary power supply.
We have USB interface for HV primary supplies.
We need only software product
3. Infant mortality of some channel of the Master and Remote boards.
Unfortunately our board was under working conditions only few weeks.
So the failure of the channels still expected.
4. The recalibration of the voltage sensors is necessary during all time of
operation. This procedure should be implemented in the system.