Text preview for : 3590-1,0.pdf part of
| Home
Number 810
Application Note Optimizing Switch/Read Rates with
Series Keithley Series 2000 DMMs and
7001/7002 Switch Systems
Introduction
DUT
The period of time an instrument requires to perform a
specific measurement is only one factor in the overall equation
of test system timing. Typical testing applications often require
measuring several points of the DUT with the same instrument, DMM Switch DUT
such as a multimeter. A multiplexer is typically used to switch
the test signals automatically from point to point. Multiplexing
can be done with relays or solid-state switching devices. The
switching devices are normally housed in a mainframe that pro- DUT
vides the card cage facility, drive lines and test instruction para-
Figure 1. DMM/Switch Configuration
meters. Keithley offers a variety of flexible measurement and
switching solutions for these multipoint testing applications,
including the Series 2000 line of digital multimeters and Model Integration rate--The integration rate is the time it takes the
7001 and 7002 switch mainframes. A/D converter to "look" at the input signal. The integration time
affects the usable resolution, the amount of reading noise, and
The rate at which a test is performed is crucial to test the ultimate reading rate of the measurement. The optimum inte-
engineers because the faster the test is completed with reliable gration time setting for a given application depends on the mea-
results, the more devices that can be tested in same period of surement requirements. If speed is the most important considera-
time. Manufacturing/production facilities are always seeking opti- tion, then the smallest integration rate would be used (0.01
mum speed in their test systems. This application note addresses power line cycles or PLCs) at the expense of additional reading
the factors that affect test system speed. These factors are noise and less usable resolution. For maximum common mode
generally additive, but their effect on total system speed varies. and normal mode rejection, the larger integration rate would be
Test engineers use the word "throughput" to define overall used (up to 50 PLCs). This would produce the most stable, accu-
system speed. Throughput encompasses a variety of factors, rate and highest resolution readings. See Tables 1 and 2 for inte-
including measurement source and sense times, and switch gration selections for Series 2000 DMMs.
settling, actuation, and trigger actuation times.
Table 1. Model 2000 and 2010 Integration Selections
Fast Medium Slow Settings
Test System Configuration Model (Front Panel) (Front Panel) (Front Panel) (Remote)
2000 0.1 1.0 10 0.01--10.00
Although test systems may be configured with many
2010 0.1 1.0 5 0.01--10.00
devices, for the purposes of this application note, we will consid-
er a test system consisting of a digital multimeter and a switch- Table 2. Model 2001 and 2002 Integration Selections
ing device. The system may connect these devices through sever- Model* Fast Medium Normal HI Accuracy Settings
al ports, such as external trigger lines, input/output signals, and
2001 0.01 0.1 1.0 10 0.01--10.00
remote control connections (GPIB, RS-232, etc.) Figure 1 illus-
2002 0.01 0.1 1.0 10 0.01--50.00
trates a typical DMM/Switch-based test system.
*All settings available on front panel and remote
DMM Configuration Ranging--All published measurement rates are stated with a
The time required to make the measurement itself is the fixed measurement range. Autoranging should not be used when
first major factor in overall system throughput. When using a optimum speed is required. While some meters have fast
multimeter, several parameters can affect the measurement autoranging features, the actual speed is not predictable. Fixed
speed. While most multimeters have control over many of these range is the only way to ensure timing precision.
parameters, some of the more "basic" units do not. However, the Filter--All rates are stated with all filters turned off. Although
factors are the same for the simplest meter or for the most com- filtering stabilizes noisy measurements, it also slows the rate of
plicated one. It is the level of control of these parameters that is measurement. The Keithley 2000 Series meters all have digital
important for exact measurement timing. filters, which are based on reading conversions. With a digital
filter, a DMM takes a number of conversions and averages them
before displaying the result. The number of conversions is the
largest factor affecting rate when using filters. As the number of
conversion averaged increases, the rate decreases. Signal
Integrate
DMMs typically offer two types of filters (Figure 2):
Repeating Filter--A repeating filter takes a selected num-
ber of reading conversions, averages them, and yields a
reading. The filter is then cleared and starts collecting con- Reference
Integrate
versions all over again.
Moving Filter--A moving filter is a first-in/first-out stack;
in other words, the newest reading conversion replaces the
oldest. An average of the stacked reading conversions Zero Signal
Integrate Integrate
yields a reading. Therefore, after a selected number of con-
versions, a moving filter gives a new reading for every
new conversion.
The time to the first reading is the same for both types of Calculate Calculate
filters. After the first reading, the moving average filter produces
a faster rate than the repeating filter.
A. Type: Moving Average, readings = 10 Auto Zero ON Auto Zero OFF
Conversion Conversion Conversion
#10 #11 #12
#9 #10 #11 Figure 3. Typical A/D Conversion Period
#8 #9 #10
#7 #8 #9
Reading #1 Reading #2 Reading #3
#6 #7 #8
#5 #6 #7
#4 #5 #6
#3 #4 #5
Zero Integrate This is how long the meter "looks" at the
#2 #3 #4
#1 #2 #3 internal zero. This allows re-zeroing the
Conversion Conversion Conversion
input amplifier. This is designed to keep
the signal from drifting over the course of
B. Type: Repeating, readings = 10 several conversions.
Conversion Conversion Conversion
#10 #20 #30
#9 #19 #29 Calculate This is the period in which the meter
#8 #18 #28
#7 #17 #27 firmware calculates the measurement.
Reading #1 Reading #2 Reading #3
#6 #16 #26
#5 #15 #25 It is normally a very small portion of the
#4 #14 #24
#3 #13 #23 overall conversion.
#2 #12 #22
#1 #11 #21
Conversion Conversion Conversion With Autozero enabled, the A/D conversion sequence
includes all four periods. With Autozero disabled, the A/D con-
Figure 2. Filter Types version does not go through the zero integrate or reference inte-
grate periods. Therefore, the conversion period is much shorter
Autozero--Normally, every A/D conversion has four distinct and the measurement speed is increased. However, it also means
periods, as described below. Disabling the Autozero function the meter is not checking reference or zero in the A/D conver-
eliminates two of these four periods, enhancing measurement sion, so the reading could drift slightly after a period of time.
speed. The actual drift is a result of small changes in the temperature
Signal Integrate During this period, the A/D converter of the measurement circuitry. To avoid this drift, keep Autozero
"looks" at the input signal. The signal inte- enabled until just before the actual measurement sequence is per-
grate period is essentially the PLC setting. formed. Re-enable Autozero once the measurement sequence is
completed. This technique keeps the measurement circuitry on
Reference Integrate This is the length of time that the meter
track before and after the measurement. As long as the measure-
"looks" at the internal reference signal.
ment does not last too long, e.g. longer than ten minutes
This allows a reference to compare the
input signal with a known level. (depending on the meter used), the readings shouldn't drift
enough to affect the accuracy of the measurement.
Trigger Delay--In a multiple channel test system, a relay is Switch configuration
used to make the connection to various signals. Relays normally
The next factor affecting test system speed is the switching
have bounce or settle times of several milliseconds. Starting the
time. The switching time is the time it takes to disconnect one
A/D conversion before the signal is settled could result in a
signal and connect another one to the measurement instrument.
noisy measurement. The trigger delay is the period from when a
Three factors affect switching time (Figure 5). All three factors are
trigger is received to the time the meter starts the A/D conver-
additive in the switch configuration calculation of switch speed.
sion (Figure 4). This delay is generally used to allow the input
signal to settle before starting the A/D conversion. Keithley's Settle Time The time it takes for the switching device to
Series 2000 meters have settable trigger delays, which, if applied settle. A relay, for example, has a settle time
carefully, can help optimize measurement integrity and speed. specification of 3ms.
Actuation Time The time required for the switching main-
frame to start closing the switch. The Model
7001 and 7002 have actuation times of 6ms.
Trigger Trigger Execution The maximum time from activation of
Time (Trigger Trigger Source to start of switch open or
Latency) close. In general, this is several hundred
microseconds.
Measure
Trigger Execution Time
Trigger Delay
Figure 4. Trigger Delay
Function--The measurement function to be performed (DCV, Actuation Time
ACV, etc.) is a factor in the measurement rate because each
function is measured using a different technique. In general, the
resistance function takes longer to complete a measurement than
the DCV function, which is simply a single A/D conversion
sequence per measurement. In contrast, the Resistance function Settle Time
could require several A/D conversion sequences per measure-
ment. If a meter measures temperatures using RTDs or ther-
mistors as temperature sensors, that temperature measurement
function is usually the meter's slowest function. Typically, DCV Total Switch Time
measurement is usually the meter's fastest function.
Display--Updating the meter's front panel display takes time, so Figure 5. Switching Times
turning off the display allows the meter to complete the measure-
ment task faster, particularly meters with information-intensive Total system speed is equal to the DMM configuration
displays. These meters typically take up to 5% more time to factors plus the Switch configuration factors. For example, to
update the display. In systems where speed is critical, the display switch and measure six DC voltages:
should be disabled.
Model 2001
In applications where it's necessary to see the display, the
following technique can be used to optimize speed.