Giga-bit VLBI, Progress on Giga-bit Sampler
Kashima Space Research Center
Communications Research Laboratory
893-1 Hirai, Kashima, Ibaraki 314-0012, Japan
Giga-bit samplers (see Figure 1) were tested at the Nobeyama radio observatory
(see the previous issue of TDC news for details of the VLBI giga-bit sampler).
Both CRL-VLBI and NRO-spectrometer group carried out the test. The DSO (Digital
Sampling Oscilloscope) sampler consists of a modified DSO part and an interface
for successive processor. The interface unit is working as a de-multiplexer of
the data stream for each purpose. Nobeyama 25-multi-beam receiver for 45m radio
telescope will use the DSO sampler with 1 Gsps (sampling per second) 2-bit
4-channel. Although there is a difference in the interface unit channel
configuration, we agreed to unify the DSO sampling part. This will simplify the
maintenance procedure at the manufacture. Radio astronomical society will be
able to accommodate each other with sampler on site. In September whole
astronomical giga-bit sampler had been gathered at Nobeyama. We measured all
DSO sampler at the same bench. Several static and dynamic parameters to
evaluate AD performance had been measured. Table 1 shows the unified
astronomical sampler major specification.
Figure 1. TDS784A Gigabit samplers under test.
Table 1. TDS-spec Comparison between the VLBI & Spectrometer.
Notes: 1:unified version. 2:single recorder use MSB bit only.
|Sample rate(M) || 1024/512/256 || 1024/512/256
|Channel Input || 4/4/4 (*1) || 4/4/4
|AD Quantaization || 2 || 2
|Output Datarate || 128MHz x64 || 128MHz x6
|Total Datarate || 8Gbps || 8Gbps
|Input Datarate || 128MHz x64 || 128MHz x64
|Handle rate(M) ||1024/512/256 || 1024/512/256
|Select channels || 1/2/4 || 4/4/4
|Quantaization || 1 or 2(*2) || 2
|Output connector || D-Sub || SCSI-halfpitch
|Total datarate || 2Gbps || 8Gbps
|Instrument || Recorder || Auto-correlator
|Datarate || 1Gbps (x2) (*2) || 2Gbps (x4)
Each channel of the sampler works 1024 Msps 2-bit for scientific data sampling.
We have checked AD performance. Although the TDS784 is completed as the
time-domain measurement instrument, longterm frequency-domain AD characteristics
should be examined for astronomical purposes. We have measured following
parameters for the first step .
Differential Liniarity Error
In 2-bit sampling, AD deviation from ideal transition spacing (DLE) exists.
Improper code-width between the transitions will give bad AD linearity.
Statical distribution of output-code bin will not represent actual level
distribution. In our astronomical data processing, there is no DLE calibrating
hardware. DLE is expected to be very small. Following equation,
defines DLE measured with LSB. Table 2 shows the measured TDS784 DLE
between 3 transitions under 2-bit sampling. Measured DLE is small enough to
neglect in the latter part.
Table 2. Differential Linearity Measurement Result of two
|DLE2 || DLE1 || DLE2 ||DLE1
|ch-1 || 0.072 || 0.082 || 0.010 || 0.030
|ch-2 || 0.032 || 0.068 || 0.040 || 0.004
|ch-3 || 0.074 || 0.046 || 0.038 || 0.016
|ch-4 || 0.074 || 0.076 || 0.042 || 0.004
- Offset Error
DC offset also results in the deviation of code appearance. Trouble will occur in
the latter part digital processing especially in correlation. Few mV DC offset
had confirmed in the AD raw data. To cancel out this,
additional offset can be
applied from the TDS front panel. Rather important longterm drift and
temperature dependence will be measured separately.
Auto-correlation result of a period was used to check the spectral
characteristics. UWBC correlator in the Nobeyama observatory was used for
preliminary test. 64 MHz harmonics was observed in cetain channel data. The bug
Updated on November 20, 1997.
Return to CONTENTS