This is only a preview of the May 2009 issue of Silicon Chip. You can view 31 of the 104 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Articles in this series:
Items relevant to "Dead-Accurate 6-Digit GPS-Locked Clock, Pt.1":
Items relevant to "230VAC 10A Full-Wave Motor Speed Controller":
Items relevant to "Precision 10V DC Reference For Checking DMMs":
Items relevant to "Input Attenuator For The Digital Audio Millivoltmeter":
Purchase a printed copy of this issue for $10.00. |
WHY CALIBRATE
your
TEST EQUIPMENT?
During preparation of the Precision Voltage
Reference project elsewhere in this issue, we
consulted with Trio Smartcal on the subject of
calibration. This article gives a brief overview.
By TONY TONG
You have invested in test equipment
to allow you to measure something.
It could be a multimeter to measure a
power supply voltage, an oscilloscope
to measure waveform amplitude or
a spectrum analyser to measure RF
power at a given frequency, etc. In
any of these cases the big question is
“how important is the result you get
from your test equipment to you and
your company?”
Many people consider calibration
to be a necessary evil. However the
implications and cost of using an inaccurate item of test equipment can be
much greater.
For example, let’s take a production maintenance technician who is
troubleshooting electronic control systems. Typically his multimeter has
not been checked since new (several
years ago) and has had a hard life in
his toolbox, being dropped a few times.
It now reads 5% low.
He is checking a control board for
the manufacturing line and is checking the 5V board supply which has a
tolerance of ±5% but it happens to be
faulty and is high by 7%. The board
is acting erratically but his meter tells
him the supply is OK, comfortably
inside tolerance. The technician then
spends another four hours looking
70 Silicon Chip
elsewhere for the problem. Not only
does this cost the company money for
his time but production has been down
for four more hours and this cost the
company $40,000 in lost output.
In a recent case here at TRIO Smartcal, we had a customer who was
setting up CCTV systems. He owned
a well-respected brand of TV signal
generator, about three years old. He
never suspected it to output anything
other than the specified signals. He
had set up several hundred systems in
the last year and felt that his waveform
monitor (not the signal generator) was
out of spec because it was not showing
the picture he expected.
We performance-verified the monitor and found it was well within spec.
It turned out that the signal generator
was out of spec instead and he had
probably set up every system incorrectly for at least the last year; not what
he wanted to hear!
Any measuring instrument should
have its performance verified regularly
if it is going to be used to measure
something that you rely on. And
regularly does not mean annually, as
many people believe. However most
test equipment manufacturers do recommend annual performance verification. The cost of verifying performance
and calibrating can be quite small compared to the cost of rectifying problems
caused by faulty measurements from
inaccurate equipment.
So, how do we minimise the cost
of calibration? There are several ways
and one or more may be applicable to
your situation:
(1) IDENTIFY WHICH items of your
equipment need to be calibrated.
This sounds obvious but often equipment that has been superseded or
not required for current operations is
still being calibrated every year, just
because no-one has looked into what
is required. Items that are not to be
calibrated should be labelled “Accuracy Not Verified – Do not use to verify
performance”.
(2) YOU CAN USE your calibrated
equipment to check your non-calibrated equipment, provided that the
calibrated equipment specifications
are typically 4-10 times better than the
un-calibrated item. This is called the
Test Uncertainty Ratio (TUR). Measure the parameter with the calibrated
equipment and then verify that the
non-calibrated units indicate similar
readings. This will increase confidence in the non-calibrated equipment
without increasing calibration costs.
More info on TUR can be found at
http://www.agilent.com/metrology/
uncert.shtml
(3) FIND OUT if your calibration
provider offers various levels of
calibration and select the best fit
for your equipment. If you run a
laboratory doing calibration work
for your customers or your lab is
NATA-accredited, then you will need
siliconchip.com.au
a full NATA report. This is an expensive option but you get a test report
from an accredited lab.
If you need to comply with ISO9001,
then a traceable calibration with data
is all you need. For those instruments
where you like to know they are within
specification but don’t need the full
traceable calibration with data, then
a Performance Verification or “Mini
Cal” is a lower cost option.
(4) YOU CAN EXTEND the calibration
interval. This applies if you have an
instrument which has been calibrated
annually for many years with good
tracking records to its performance.
Then you can make use the past data
reports, to identify several key measurements, extract them from the report
into a spreadsheet and graph them
against the published specification
limits. This will allow you to predict
the rate of drift.
Armed with this information, you
can confidently set extended calibration intervals. An example is a meter
with a specification limit of 1% as
a 1-year specification. Its initial cal
data shows it to be -0.5%, the next cal
shows the error to be 0%. The third
cal shows the error as +0.5%. Hence,
the drift is +0.5% per year so we can
expect it to be out of specification at
siliconchip.com.au
the next calibration interval.
If the drift was only .025%, we
could predict six years (from -0.5%
to +1.0%) drift and extend the calibration interval accordingly. If you
buy an instrument that comes with a
factory “cal-report” you already have
one of the reference points needed to
predict drift.
Doubling the calibration interval
would halve your costs for calibrating
this item without impacting on the
quality of the measurement.
(5) TAKE OUT a calibration contract. It
is often possible to take out a calibration contract with your service provider. This is normally taken for 1-3
years and discounts are available for
equipment quantity and for length of
contract. You may also get additional
benefits, including fixed prices for the
contract period.
You also reduce the delays and internal cost of creating multiple purchase
orders and billing events. This also
improves the turnaround times due to
the elimination of delays for quoting
and sending purchase orders. It also
allows the annual calibration costs to
be budgeted more accurately, which
improves your expenses forecasting.
(6) TAKE OUT LIMITED calibration.
This allows you to match your instru-
ment’s verified performance to only
that which you need. You do not pay
for calibration tests you do not require.
A case in point was an RF signal generator used in a laboratory. It was used
in a test rack to perform only one test
yet the user was getting a full calibration done every year.
When the generator was tested
for only the required signal, the cost
dropped by 75%. This was without
compromising the traceability or
quality for the measurements needed
for the specified task. All that was required was a label indicating that the
instrument had a limited calibration.
Supplementary information concerning limited calibration of the unit
should be made available for a person
who uses it.
(7) FINALLY, YOU CAN arrange to
have an expert from your calibration
provider visit your facility. He can
recommend ways to reduce your calibration costs and to improve measurement quality. There is usually a charge
for this service which is waived if you
proceed with a calibration contract/
plan.
For further information or a quote
for calibration contact Trio Smartcal
on 1300 134 091. www.triosmartcal.
SC
com.au
May 2009 71
|