Transcript Current State of ALMY Readout, December 2007

Current State of ALMY Readout
Electronics, December 2007
Readout Electronics Consists of
1. Analog Board
2. Interface Board. This currently have three
options:
a. The old board, based on ATMEL 8-bit
microcontroller
b. New board, based on NXP (former Philips)
32-bit ARM microcontroller
c. Prototype board, based on (ADI) Analog
Devices 32-bit ARM microcontroller
3. Laser Controller Board
Analog Board
• Design of 2000
• Main components:
 Amplifier. Bandwidth ~2MHz.
Affects the measurement time.
 Multiplexer. Seems to be worst
radiation withstanding
component
 12-bit ADC with Serial Interface
 Local negative voltage supply
Analog Board
•
Radiation Hardness. Three options for multiplexers exist
until now. No new 16:1 multiplexer designs in the same
packages and power voltages (for direct replacement) found
in the Web.
–
–
•
•
ADI multiplexers gave very negative result
Maxim and Pericom still waiting for tests. No new sensors are
available for tests.
Very few chances to get a better result with listed
multiplexers too.
Possible solutions depend on what we need.
1.
2.
Radiation hardness is absolutely mandatory for all cases. The
multiplexers and the rest of electronics should be removed from the
hard radiation area. Penalty is increasing of noises and accordingly
worse resolution. Needs completely new approach to design, usage of
special cables and so on. Necessary to know in details the radiation
distribution to minimize cables length.
Radiation hardness experiments can be continued, besides non
radiation tolerant devices are still actual to have for different
applications. Using new components is possible to put all electronics,
analog and interface on one board.
Analog Board
• Noises. There are several sources of noises:
– Sensor itself and loading resistors. This cannot be changed by
electronics.
– Amplifiers. The amplifiers are of best ones. Only possibility to reduce
noise further is usage of a single-supply amplifiers.
– Multiplexers. The noise of multiplexer are greatly affected by the digital
address selection signals (see below in the interface part), power supply
voltage and the multiplexer technology. The new 16:1 and 32:1 low
voltage single-supply multiplexers seem to be better solution.
– Local negative -4V supply. It has very small noise. In case of replacing
the amplifiers and multiplexers to single supply powering it becomes
obsolete.
• ADC. It is affected by digital control signals as well as
multiplexers, especially in case of remote control.
– In case of usage microcontrollers with build-in 12-bit ADC, this part
becomes obsolete.
Interface Board
• Old board (2000 design), based on ATMEL 8-bit microcontroller
This board has different optional interfaces: RS-232, RS-485 and
CAN. It was successfully used for measurements during
several years.
Interface Board
– It works good if stacked directly on the analog board. But in case of
remote cable connection between the analog and interface boards for
separate radiation test, the signal degrade. See graphs below.
This boards was not designed for such usage. The situation is corrected in
the new interface board.
Fig. 1. Stacked
Fig. 2. Connected by 2m cable
Interface Board
– Another disadvantage. In the 8-bit microcontroller with very
limited speed and memory size (~8MHz, 8kB Program Memory
and 1Kb RAM) it is very difficult to implement the Gaussian Fitting
algorithm, which is eventually necessary for the completed design
and data correction algorithms.
Nonetheless there was a large party of such boards produced, as
well as analog boards, which can be used for many applications.
To avoid listed problems, it was decided to design a new Interface
board using the new generation microcontrollers. In 2005 was
designed a new board, based on the NXP (former Philips
Semiconductor) 32-bit ARM (Advanced Risk Machine) LPC2129
microcontroller. The PCB was ordered and several samples were
assembled. The main features of the new board are listed in the
next slide.
Interface Board
New board (2005 design), based on NXP 32-bit microcontroller
– Due to smaller power voltage 3.3V instead of 5V and, hence,
smaller amplitude digital signals with better controlled slopes, they
have significantly smaller impact on the analog measurements.
For comparison with the old one, graphs for the same conditions
as for the old board see below. There is no degradation.
Interface Board
Fig. 3. Stacked
Fig. 4. Connected by 2m cable
Interface Board
• Another advantages:
– 32-bit architecture, 256kB Program memory and 16kB RAM is
enough for (almost) direct porting of the fitting algorithm from PC
to the microcontroller.
– A Humidity and Temperature Sensor is added on the board. It
immediately revealed, that with 60-80mA consuming (summary
for both boards), the average temperature of the board is about 34 Celsius above the ambient. This can distort the precise
measurements result. So in case of a new design it is necessary
to implement a sleep mode for the electronics, so it will awaken
only during the measurement, thus minimizing overheating.
– The internal 10-bit ADC of the microcontroller can be used instead
of the external microcontroller, mounted on the analog board.
Unfortunately, due to smaller resolution, the dynamic range of the
whole sensor is worse. Usage of the new generation of 32-bit
microcontrollers can solve the problem.
Interface Board
– Up to 64kB (currently 1kB) nonvolatile memory chip can be
assembled on the board. It can be used for storage of different
parameters like the sensor address, and for storage of arrays with
correction data. Very simple examples one can see below. The
Fig 5. shows the raw data of the sensor with bad strips. On the
Fig 6. bad strips numbers array is stored, and for these channels
0 value is substituted. On the Fig. 7 the mean value of the
previous measurement is used for substitution.
Fig. 5. Raw Data
Fig. 6. Zeros for bad strips
Interface Board
Fig. 7. Mean value substitution instead of bad strips data
Interface Board
– More complex data correction schemes can be implemented,
thanks to a large memory size. Below are graphs for two sensors
from the last party of ALMYs. The algorithm ignores (covers) the
bad strips. For all other strips saves in the non-volatile memory
an array of multiplicative correction coefficients.
Fig. 8. Raw data for GC38-I sensor. Laser spot is at small
peaks
Interface Board
Fig. 9. Corrected data for GC38-I sensor
Interface Board
Fig. 10. Raw data for GC47-IV sensor
Interface Board
Fig. 11. Corrected data for GC47-IV sensor
Interface Board
Disadvantages:
– Only 10-bit internal ADC is insufficient for good dynamic range, if
no external ADC chip is used
– There is no possibility to switch off the power to go to slip mode
Now we have 4 such new boards assembled and tested, and 6 bare
PCBs. To assemble these PCBs we have to order additional
components.
For a larger party of the new readout board, the new amount of
PCBs also have to be ordered.
We have to decide: or produce more such boards, or design a new
boards, which will have all listed advantages and more new ones,
as well as a smaller cost. The proposal below.
Interface Board
New prototype designed and made in December 2007, based on ADI
32-bit ADuC7020 microcontroller with 12-bit ADC and DAC
– The main goal of the new design is to check how the microcontroller of
the new generation with build-in 12-bit ADC will work with the analog
board. The graphs are similar to the previous ones and does not depend
on the cable length. With this prototype the minimal average noises
where achieved.
– Advantages of this new microcontroller:
– Very small size, can be directly mounted on the analog board, so all
readout can be placed on one board.
Interface Board
– Build-in DAC allows to make the ALMY sensor bias voltage
programmable to get an optimal working mode.
– Build-in temperature sensor.
– Disadvantage – lack of CANbus interface. Can be overcame by
usage of a standard RS to CAN converter for the whole “Laser +
Sensors” line. Just like standard USB to RS485 converter which
we use now.
Another type of perspective microcontroller to be checked is the
ST Microelectronics STM32 ARM Cortex-M3 with 12-bit 2
simultaneous sampling channels ADC.
– Compared to the ADI 12-bit ADC microcontroller, it faster and has
better performance for the Gaussian fitting calculation,
– Has more reprogrammable nonvolatile memory, enough for
storage both the program code and data correction arrays,
without an external EEPROM chip usage,
– Build in CAN interface,
– Disadvantage – not tested yet
Programmers for interfaces boards
Each type of microcontroller needs a special hardware
programmer and corresponding software to be
programmed during software design and production. I
have made in Yerevan in my lab two types for the new
interface boards and used them here for software
design.
Microcontroller firmware for the
ARM-based interfaces boards
The program (firmware) for microcontrollers for both types
of ARM-based interfaces is written in C, using free GNUlicensed WinARM package
Laser Controller Board
• Was designed in 2002
– Has several optional interfaces: RS-232, RS-485, CAN.
– Can be used also as COM <-> RS-485 converter.
Unfortunately the laser we use, is no more produced. So to work
with new types of lasers it should be slightly redesigned and
made more universal. At least it should work properly with all
types of new lasers from the same (Schaefter&Kirchhoff)
company
The next steps to be planed
• Make a decision if the work on the ALMY
readout should be continued. The first question
is availability of the sensors.
• If yes, then in which of two directions:
– Radiation hard solution or with a new ASIC, or with
remote readout located out of the radiation zone.
– Integrated one-board solution, based on the new
generation components. This can be afterwards
again tested for radiation tolerance and, besides just
used for a general purpose position sensor.
The next steps to be planed
If both, or, for the beginning, just the second alternative will be accepted,
then the next steps seem to be done.
1. Make a permanent ALMY corner in the laser lab. It should be
portable and small enough, and between the activities could be
compactly stored in a dedicated shelf. We can start during my this
visit.
2. Continue to work with Stuttgart to have a sensors of a necessary
quality.
3. Design a new readout electronics board. The proposed parameters
and features below. Design can be made in Yerevan (as the
previous one) and produced in MPI. This can take 3-4 months.
The main problem usually is availability of small quantities of
components we need. The best supplier company for such a case
is the USA Digi-Key company, which works in Europe too, but MPI
cannot order from it due to some financial problems. It is very
desirable to solve the problem for any new electronics designs.
4. After the readout boards will be produced and if a good quality
sensors will be received from Stuttgart, then I can again visit MPI
with my young colleague programmer, to test the new boards and
design a new, simple to use, readout software, instead of, or based
on, Alica3.
The new readout board proposal
• All electronics on single board
• Single power supply. This is necessary for new generation of
amplifiers and multiplexers, besides to minimize consumption
• Sleep mode to avoid heating over ambient
• New, small size 32-channel multiplexers can be used. In this case
all 64 strips of each coordinate can be connected on one side,
instead of 2, like now (is it desirable? to be discussed)
• New 32-bit microcontrollers with 12-bit ADC should be placed on the
board. No more external ADC chip is needed. Which one: ADI or
ST, should be decided after ST microcontrollers testing. This can be
done in Yerevan, using LED instead of the laser.
• Programmable sensor bias voltage.
• The board should have at least the RS-485 interface (as now), all
other necessary interface solutions to be external. If possible,
internal CAN should be implemented as well.
The new readout board proposal
• The board should have a low impedance analog outputs of the
amplifiers, for possibility to be used just as a simple analog board
with and external ADC and interface.
• If necessary, the size of the board can be reduced and the place of
the window changed
• The cost of the new board supposed to be about ½ of the existing
two board solution.
December 2007, Max-Planck-Institute for Physics, Munich
Varuzhan Danielyan, Yerevan Physics Institute
Just for information
Low Cost Programmable High Voltage supply for PMT and different types of
particle detectors. Designed and produced in Yerevan Physics Institute. More
than three years 50 such supplies work reliable in the Nor Amberd and Aragats
mountain (3200m) based Cosmic Ray Stations.
Voltage programming in two hardware selectable ranges: 900V to 2100V and
1500 to 3000V in 2V steps. Output polarity: Positive and Negative. Regulated to
accuracy ±2V.
Max. output current 1.2mA for ± 900V to 2100V range; 0.8mA for ±1500 to
3000V range.
Input voltage from +12V to +18V.
RS-485 half-duplex 2-wire 9600 baud interface to program and monitor the
output voltage.
Now a better one, with up to 4kV output range, 4 times better stability and
resolution, reduces low-frequency pulsations is in the final stage of design.