Vaughan Wesson, February 2003

Introduction

This technical note is designed to help users calculate response information for SRC digital recorders. It covers the Kelunji Classic with a KA1 or a KA2 analog board and the Kelunji D Series with a DD1 or DD2 digitiser board.

For each recorder, there are a number of stages the signal goes through from the input connector to the recorded numbers. These are:

Input Stage

This includes input protection and a may include a simple low pass filter. All instruments also have a finite input impedance that affects the output of the connected transducer

Amplifier Gains

Most board include some capability to vary the gain. This may be a software and/or hardware function

Analog Filters

Depending on the analog to digital convertor (ADC) being used, some type of analog anti-aliasing filter is required

Analog To Digitiser Convertor (ADC)

This is the device that converts the analog signal to a digital form at set intervals

Digital Filters and scaling

In some situations, additional filtering and/or scaling is performed on the digital samples

Models

This technical note only covers the Kelunji Classic and Kelunji D Series families of instruments, not earlier SRC models such as the Alpha or Yerilla. There are four models of Digitiser that will be discussed. Each of these are manufactured such that every unit of a particular model will have response values that are within one or two percentage points of the nominal values indicated here. For many purposes these values will suffice, but for critical studies, it may be necessary to characterise the particular instrument(s) being used.

Kelunji Classic

The Kelunji Classic came with either of two possible analog boards. The KA1 is a three channel board using a 12 bit ADC and automatic gain ranging. The gains are binary factors with 12 gains from x1 to x2048. The KA1M is a modification of the original KA1 to provide lower noise levels and better immunity to external noise, its response figures are identical to the KA1.

The KA2 is a three or six channel board using a 16 bit ADC. It has hardware and software selectable gains.

Kelunji D Series

The Kelunji D Series currently has two possible Digitiser boards. Either or both of these may be used in any D Series recorder. The DD1 is a three channel 22 bit board while the DD2 is a six channel 24 bit board.

Input Stage

The KA1 has a very basic input stage. It is only a single ended input as the return line is connected to ground. There is a 10k ohm current limiting resistor and a pair of back to back 5.6V zener diodes for protection. The signal then goes directly in to a buffer amplifier.

The KA2 signal and return line input sections are identical. They go through a 100 ohm resistor and in to an instrumentation amplifier. Each line also has a 1M ohm resistor and a capacitor to ground to provide a path for bias currents and a crude low pass filter to remove high frequency signals.

The DD1 input circuit is very similar to the KA2's but with a lower value (47k ohm) resistor to ground.

The DD2 input signals go though a voltage clamping device and then each have a resistor and capacitor to ground. In addition, there is a capacitor between the two signals. The DD2 analog circuit is differential all the way to the ADC.

Amplifier Gains

The KA1 has a fixed hardware (preamplifier) gain of 1. This can only be changed by changing resistors on the PCB. It then uses a hardware assisted software scheme to automate gain changing from sample to sample. The gains are 1, 2, 4, 8, 16, 32, 63, 128, 256, 512, 1024 and 2048.

The KA2 has hardware selectable preamplifier gains of 1, 10 or 100 and software selectable gains of 1, 2, 4 and 8. These are not designed to be changed dynamically. Typically a gain of 1 is used for an accelerometer and a gain of 100 for a seismometer.

The DD1 has software selectable gains of 1, 2, 4, 8 and 16. As a by-product of the way the gains are implemented, additional gain values are available but these are not whole numbers.

The DD2 has software selectable gains of 1, 2, 3, 4, 5, 6, 7 and 8.

Analog Filters

The various Kelunji analog boards use a variety of analog filters. The mathematics of these filters is a good topic for another technical note that has not yet been written.

The Kelunji Classic boards use a simple RC high pass (low cut) filter and a complicated low pass (high cut) anti-aliasing (AA) filter. For these boards, the AA filter must try and cut out all signals at frequencies equal to and greater than half the sample rate. For example when running at 100 sps, the AA filter should cut out all signals greater than 50Hz.

The Kelunji D Series boards both use sigma-delta ADC's that sample the signal at much higher sample rates (hundreds of kilohertz) and use digital filtering to avoid aliasing. In this case, a simple RC low pass filter is sufficient.

Analog to Digital Convertor

A wide variety of ADC's are used on the various boards. The Kelunji Classic boards both use monolithic ADC's of the "standard" type. For the KA1 this is an Analog Devices AD7582 four channel 12 bit convertor. The convertor accepts (single ended) signals of between -5V and +5V.

The KA2 uses a Cirrus CS5016 single channel 16 bit ADC. There is a four channel multiplexor in front of this to select which channel is being sampled. The convertor accepts (single ended) signals between -5V and +5V.

The DD1 uses an Analog Devices AD7716 four channel 22 bit sigma-delta convertor. It uses a sinc cubed digital filter whose amplitude can be computed from (Sin (pi.f/F)/(pi.f/F))^3. Where f is the frequency of interest and F is the output sample rate. The delay caused by the digital filter within the ADC is corrected for in the Guria software.

The DD2 uses six separate Burr-Brown (now Texas Instruments) ADS1210 single channel 24 bit sigma-delta convertors. These also use a sinc cubed filter with the same response as that given for the DD1.

Digital Filters and Scaling

Both the Kelunji Classic and Kelunji D Series provide the option of sending a continuous stream of data out one of the serial ports in a variety of formats. This is called telemetered data. This data may be scaled in the Kelunji, so this scaling should be taken in to account.

The Kelunji D Series has the capability to perform digital low pass filtering and decimation on the data streams for its continuous data and telemetered data. This is done using two "decimators". One decimates by a factor of two, that is, it produces an output data stream at one half the sample rate of its input data stream. The other decimates by a factor of five. It is possible to chain together any number of either of the decimators. For example, the Digitiser could be sampling at 200 samples per second, then using decimators with factors of /2, /5 and /5 a data stream at 4 sps would be produced.

The filters used in these decimators are FIR filters designed for at least 60dB attenuation in the stop band. That is less than 0.1% of any signal in the stop band will get through. The /2 decimator has a transition band from 0.2 to 0.3 of the input sample rate, while the transition band for the /5 decimator is from 0.08 to 0.12.

The tables below show the filter coefficients for the two filters. These are scaled by a scale factor of 1,048,576 (that is 2^20). The /2 decimator filter has 29 taps and the /5 one has 69 taps. Since the filters are symmetric, only one half the filter is shown.

Divide by 2 Decimator Filter

Divide by 5 Decimator Filter

Typical Calibration Amplitudes

The tables below show typical (zero to peak) amplitudes recorded when performing standard calibration checks. More entries will be added as these are determined. We also welcome entries from our customers for these tables.

Passive Seismometers

Active Seismometers

Accelerometers