PMT-2000

PMT-2000

The PMT-2000 assembly.

Introduction

The PMT-2000 contains an MCA and plugs directly on to a photo-multiplier tube (PMT). It is powered through USB and uses and embedded FPGA plus a 32-bit ARM M0+ processor and SoC for acquiring histograms, computing count rates and alarms, even providing list mode operation and performing pulse shape discrimination. It can even be a complete portal monitor appliance which alarms on radioactive materials passing by in a vehicle.

PMT-2000

Introduction

The PMT-2000 is a low-power device with a PMT power supply, a switched-gain amplifier and a waveform digitizing ADC readout by an FPGA, all controlled by an embedded ARM SAM L21 ARM M0+ processor. The software running on its embedded 32-bit ARM processor can give this device quite some extraordinary capabilities. Besides the automatic gain stabilization, it can measure samples and background, compute alarms and even alarm on a passing radioactive source.

The PMT-2000 building blocks, here shown for a unit with negative high voltage.

Capabilities

PMT-2000 Capabilities
Capability Description
Analog The input of the PMT-2000 is DC-coupled to the PMT-anode for negative high voltage. Input pulses are sampled by a waveform-digitizing 12-bit ADC.
There are four electronic gain settings and a continuous gain adjustment va the high voltage. The operating voltage can be adjusted for optimum energy resolution. That sets the maximum measurable energy. After that, a 20-bit digital gain is used to map the energies into a desired MCA bin range; ie a given MCA calibration in keV/bin.
Gain stabilization The PMT-2000 uses either built-in or user-programmable look up tables vs temperature to adjust the PMT operating voltage as a function of temperature. Alternatively, the device can stabilize on the average energy deposited in a given region of interest.
Histogram size
Two banks:
2K × 32-bit; or One bank
4K × 32-bit
The MCA histogram memory is 16KB. There are two banks, one for sample counting and one for background counting. There is a histo_4k mode in which the two banks are combined into one large bank, providing a 4K × 32-bit histogram.
Counter and histogram The PMT-2000 can count pulses and acquire histograms in either of two active banks, one for samples to be measured and one for storing a background measurement. The device reports count rates and statistical 2-σ errors. Users can choose to see total counts or counts restricted to one region of interest.
Net Counter The PMT-2000 reports the difference between sample and background count rate together with the combined statistical 2-σ errors.
Analysis The PMT-2000 reports the probability that the measured sample count rate is compatible with the background count rate.
Dynamic alarming The PMT-2000 can analyze and report count rates in time slices of 50ms, ie at a rate of 20/s. The device automatically tracks slowly changing backgrounds and will alarm on a passing source. Its digital output can be used to drive an audio or visual alarm.
Loss-less counting The PMT-2000 implements a read-and-clear command, in which the ARM processor clears the counters right after copying data to the output buffer – for nearly loss-less reading of count rates.
Two-bank mode In two-bank mode, the device has an active and an inactive data acquisition bank. When the host reads MCA data, the device automatically selects the inactive bank and clears it when the read is complete.
Communication The PMT-2000 implements a USB-2.0 compatible USB 1.2 interface.
Security Software deployed on the PMT-2000 can not be read back.
Gain-stabilization parameters and lookup tables can be protected by the developer against read back by programming a lock bit.
On-board software can be updated at the customer site via USB using encrypted files.

Gain stabilization

The PMT-2000 can use a 20-point lookup table that describes the desired operating voltage vs temperature behavior. The embedded processor applies this to counteract the PMT vs temperature gain drift. Typically, the lookup table starts at lut_tmin=−30°C and increments in lut_dt=5°C steps up to 65°C. However, the developer can program lookup tables of their own choice into the non-volatile memory of the PMT-2000. If they wish, the developer can protect the lookup tables against read back.

Count rate measurements

The PMT-2000 provides independent count rate and histogram measurement in two banks, and this is supported by different operating modes. Count rates can be reported as a histogram total, or be restricted to events falling into a programmable region of interest.

When used in read_and_clear mode, the user reads the sample histogram bank frequently and the bank is cleared at the end of each read. The client simply issues a sequence of read commands.

In addition, there is built-in software to compare a sample count rate against the background count rate and compute the statistical probability that the sample counts are caused by the same activity as the established background rate. The developer can set an alarm threshold and let the trigger output of the PMT-2000 indicate an alarm when the sample is decidedly more radioactive than the background.

Time-slice operation

There are dynamic situations, where a radioactive source can be measured only for a brief moment. Examples are a vehicle passing through a radiation portal monitor, or a person with a backpack detector walking past a stationary source.

The time-slice operation supports these cases. The built-in software tracks slow changes in the environmental background. An alarm is created when during a summation time (L) of typically 4 seconds, the accumulated counts are significantly more than what is expected from the background. The alarm threshold is defined as the probability that the measured counts (N) during a period L, could have been caused by the established background rate over the same period (B).A threshold of 1.0e-4 means that we alarm when P(Counts ≥ N|BCK) < 1.0e-4.

For example, assume a summation time of 4 seconds and a background rate of 500cps for BCK=2000. Now assume that we count 2224 events in a particular 4s-period. The probability of the established background to cause 2224 counts or more in 4s is P(Counts ≥ 2224|BCK=2000) = 2.86e-7. This smaller than the alarm threshold of 1.0e-4, and the embedded program will generate an alarm.

If the alarm condition is permanent, the software resets all the logic after a period of H time slices and starts counting again. It now will accept the suddenly higher level of radioactivity as the new normal background.

Finally, a 'wait' parameter tells the system to wait a number of time slices after turn-on or reset before being ready to alarm. This is necessary so that the background will be known with sufficient accuracy.

All told, the time-slice firmware provides an unprecedented, and highly configurable, but fully autonomous alarming system for portal monitors. This is ideal for very low-cost mass-produced pedestrian monitors, hand-held sweepers and similar applications.

PMT-2000 Assembly

Outline drawing for PMT-2000 plug-on MCA. Dimensions are [mm] and inch.

Brief Specifications

  • Ideal for NaI and slower scintillators
  • Histogram: Sample and background at 2K×32-bit, or one bank at 4K×32-bit.
  • Accurate count rate measurements
  • Background-subtracted spectra with statistical error analysis
  • Portal monitor background tracking and alarming refreshes 10 times per second
  • Two-bank listmode with programmable time stamp resolution
  • Embedded neural network for pulse shape discrimination in phoswiches and neutron detectors.
  • Open-source application programmer's interface; Python API
  • Windows/Linux wxMCA graphics user interface
  • Power: 4.3V to 5.5V, 30mA, 150mW @ 5V
  • PMT supply: 1400V/250µA
  • USB 1.2 interface compatible with USB 2.0
  • On-board software is secure against reverse engineering.
  • On-board software can be updated at the customer site via USB using encrypted files.

Performance

The PMT-2000 provides high-quality spectra with a very low energy trigger threshold.

Here we show energy spectra for certain MCA + detector combination.

We emphasize the low-energy behavior by showing a zoom-in on the lower 100keV.

The two spectra were acquired with 1keV/bin for a useful energy range of 1600keV.

We show a typical performance for a premium quality integrated NaI detector (6.4% fwhm at 662keV) with negative high voltage.

Integral NaI(Tl): 50×50mm; Premium; Divider: N81L

Typical spectrum @ 23kcps.

The lower 100keV part of the Cs-137 spectrum, showing the effective trigger threshold of around 10keV.

Sample Measurements

The MCA-2000 (PMT-2000 and SiPM-2000) can store a background spectrum, and then compare samples vs that background.

It evaluates the likelihood that the sample is more radioactive than the background.

The user can set a false-alarm limit (eg 1:106) and let the MCA raise an alarm.

It has an output that can drive an LED, or control a light tower.

It is possible to set a region of interest (ROI) to narrow the attention to a certain kind of radiation.

Work flow: At the begin of a campaign, the user may measure the background for 10 minutes. Afterwards, they switch in one sample after another and check if the count rate is statistically speaking, significantly higher than background.

The first column of graphs shows the response to a sample that is equal to background.

The second column shows the response to radiation from a 3.8µCi source at 2.15m distance from a 50mm NaI-detector; ie 0.23µrem/hr, 2.3nSv/hr

In both columns the top graph shows how the measurement of the net count rate becomes more accurate over time. The middle curve is the most likely count rate, whereas the top and bottom curve show the 95% statistical accuracy range.

In both columns the bottom graph shows the measured signal strength. The signal strength is defined as -log10(false-alarm probability). False-alarm probability is the chance that the observed excess count rate is due to a purely statistical fluctuation.

A more detailed description can be found in the user's manual .

Sample counting for background sample

Measured count rate excess when measuring background again. As expected, the likely count rate is zero. Top and bottom graph show the 95% confidence interval.
 

Measured signal strength of the repeat background measurement. As expected the likely signal strength converges to zero.

Sample counting for a weak source

Measured count rate excess when measuring very weak radiation. The excess count rate, in a [580keV, 750keV] region of interest, is about 1.0cps. Top and bottom curves show the 95% confidence interval.

Plot of the signal strength. As more statistics accumulates over time, the confidence that the measured count rate excess is real grows. A strength of 6 indicates an only 1:106 chance that the counts are caused by background alone.

Portal Monitor Feature

The MCA-2000 (PMT-2000 and SiPM-2000) has a built-in portal monitor capability.

It can evaluate count rates 20 times per second and issue an alarm when the count rate exceeds the background rate.

The user programs the alarm threshold by indicating the acceptable false-alarm probability; eg 1.0e-6.

The unit continuously monitors the count rate and updates the measured background count rate to keep up with slow changes in the environment due to rain, wind and dust. For a portable system the background may slowly change with the terrain without causing an alarm.

The PMT-2000 performs a statistical analysis every 50ms and alarms if the alarm probability exceeds a programmed threshold of typically 1:10k to 1:1000k (1k=1000)

It is possible to set a region of interest (ROI) to narrow the attention to a certain kind of radiation.

The panels on the right show the result of a walk by of 3.8µCi (140kBq) near a 50mm NaI(Tl) detector.

The first column shows the total counts per 50ms when a source passed three times; all counts in the top panel, counts within an ROI in the bottom panel.

The second column shows another two source passes together with the computed signal strength. At a false-alarm probability of 1:106 the MCA-2000 would have confidently raised an alarm in both cases.

A more detailed description can be found in the user's manual .

Logging rapidly changing radiation intensities.

Plot of the response to a 3.8µCi Cs-137 source moving past a 50mm×50mm NaI detector, in precise 50ms time steps.

Same as in the previous graph, but for a narrow region of interest around the 662keV pewak, resulting in lower signal and background counts.

Alarm computed by the MCA-2000

Plot of Number of net events above background per 50ms time slice in response to a weak source passing by at times t=2...5s and t=10...13s.

Plot of the signal strength p=-log10(probability), measuring the probabbility that the observed net events are caused by background fluctuation. A strength of 6 indicates a 1:106 chance that the counts are caused by background alone.

Downloads and Pricing

Configure your PMT MCA