Text preview for : HP-Bench-Briefs-1975-05-06.pdf part of HP HP-Bench-Briefs-1975-05-06 HP Publikacje HP-Bench-Briefs-1975-05-06.pdf



Back to : HP-Bench-Briefs-1975-05-0 | Home

rli




- -
S


DIGITAL VS ANALOG
by Dick Gasperini, Editor
(r



It seems that just about every new




oonswmer items
and calculators are



cireuits?
\
Digital may s m confusing, but it drcuitr t no layiera rna&erafetmlmfor sewiee pemmnol; ajhm uwfer-
e
need not be; digital items (more k essemal. This arllcle h th

the most common.



on or off. An automobile
either energized or sitting i e therefore it is an analog control
thing with a car's horn. Brake lights circuit.
The simplicity of adigital circuit results
arelitordarklandsoaretail lights.. .
and the list continues.
from its two possible s t a t e s i t is either
"on" or "off ". Of course, there are a lot of analog
circuits around too. The light
dimmer on your car`s dashlights is
control circuit, This
possibilities than a
simple toggle switch- the lights
could be off or on very dim, or




There are a lot of familiar items that
are digital in operation. A light

I'
cbmp.ny 197s
oc(ewktE.~nd
WWW. H PARCHIVE.CQM
DIGITAL VS ANALOG




DATA TRANSFER

Let's suppose that we have a
frequency meter (counter) that can
measure any signal from 1 kHz up to
10 kHz. It is measuring the
frequency of a signal (such as an
oscillator at a radio station) and we
want to obtain a plot of the variations
in frequency on a strip chart
recorder, as shown in Figure 3. One
way to accomplish this is to output a
voltage to the strip chart recorder
that i s p r o p r o t i o n e d t o t h e
frequency. For example, 1Volt could
signify 1 kHz, 2V would mean 2 kHz,
etc. up to 1OV for the full scale
deflections of 10 kHz.




There are many examples of digital is either on or o f Same with a
control circuits that we work with clothes iron, baking oven and a
daily, such as those listed in Table 1. home furnace, even though these
have a temperature dial that is
analog-that is, one that can have an
Items that have a thermostat are infinite number of settings. The
generally digital in nature, although thermostat circuit determines if the
some people might argue that they temperature is above or below the
are analog. The answer can be thermostat setting and then either
determined by examining the power turns on the heating element at full
consumed by the appliance. An power or keeps it off. What we are
electric skillet, for example, will doing with the thermostat is setting
either be consuming maximum the threshold (more on this later).
power or none at all. Generally a See Figure 2.
light will be on when heat is being FIGURE Measuri
There is another category of inn the iuencv c
applied. That is, the heatingelement
products that we may have difficulty
classifying. Earlier, we agreed that
devices such as the toggle switch
that had only two possible states While this method works, it has its
would be called digital. Items such limitations. A change of 100 Hz in
as the volume control that had an frequency will move the pen on the
unlimited number recorder only lo,% full scale. A
of
be called analog. How about the of 1 Hz will not even be
seven speed blender in the kitchen? ible. Therefore, this method
Or the electric stove with eight lacks resolution that may be desired.
pushbuttons? Or the three speed
heater blower in your car? These An improvement will be to replace
FIGURE 2. ing furnace don't seem to fit either th the strip chart with a digital printer
usually digi eration-eit digital category. Since t and to have four signal lines-one for
Iplying h several definite repeatatjl each digit. For our rement of
each of the examples, this category 9431 Hz, the four w I have 9V,
has been termed staircase. 4V, 3V and 1V on them respectively.


I WWW. HPARCHIVE.COM
MTBF-WHAT DOES IT MEAN?
.I .I
.-

The digital recorder would thus print
9431. If the frequency changed to
8.965 kHz, the four output lines
would have voltages on them of 8V,
9V, 6V and 5V, respectively, and the
digital recorder would print 8965.
This system, which gives the
resolutions desi red while `retaining
the full operating range, is called the
staircase method of transferring
data and was used for many years.
It fell into disuse because of the
difficulties of decoding the signal.
That is, generating the correct
staircase signal (for example 4V) is
relatively easy, but the digital
recorder has difficulty monitoring
was reduced to two (0 and 1) in the indicator is a "hot engine" light. If a hot
the signal and determining if a 4 binary number system, the term engine condition exists, the switch will
should be printed (or is at a 3 or a "digital" was still used to describe close, turning on the light.
5?). this. The name has been retained
The staircase method of data and will very likely never change, This is a very simple circuit, but we
transfer can correctly be termed even though we really ought to call will use it to build more complex
"digital" since a unique definite this binary. circuits in the next issue of BENCH
repeatable number (digit) is being -
BRIEFS.
-

indicated by a definite voltage level. The top circuit in Figure 4 is an analog
solution for monitoring the temperature This article copyrighted 1975 by
That is, the voltage on one wire will Richard E. Gasperini for publication
indicate one of ten possible digits. in an auto engine.
-
in a book on .Digital Trouble-
When the possible number of digits The digital solution to a temperature shooting.


Y

MTBF - WHAT DOES IT failure modes of the instrument. (HP erage failure rates for these parts. Let`s
applies this method to new instruments assume that ICs and transistors fail at a
MEAN? to predict the failure rate.) rate of 0.1% per year, and assume
capacitors and resistors fail at 0.01%
by Rod Parks per year. Most reliability models as-
1
Reliability and Mean Time Between MTBF = X No.of hours sume a "chain" of components. That is,
Failure (MTBF) data is often mistak- Z(No. of ea. of operation a failure of any one component consti-
enly thought of as an absolute num- type of corn- per year tutes a failure in the instrument. There-
ber. The person (HP customer) who ponent) X fore, we compute the failure rate by
uses the MTBF figure for the pur- (expected simply computing the summation of the
pose of deciding to buy HP's or a failure rate quantity of parts times the failure rate.
competitor's instrument must be of component) For our example:
cautioned that MTBF calculations
are very subjective. Due to under- The process of deriving a theoretical
lying assumptions made by the model that will accurately predict (or 100 x 0.1% = 10%
manufacturers (HP or competitor), specify) the reliability of a product is 60 x 0.1% = 6%
as much as a 3 or 4-to-1 difference one of simplifications (subjective). 50 X 0.01% = 0.5%
in MTBF calculations may result. 50 X 0.01% = 0.5%
For example, assume that we want
Generally, there are two types of to calculate the failure rate for a Predicted failure rate: 17% per year.
MTBF calculations: a) prediction new product. This instrument con-
based on component count, and sists of only the following parts: Of course an actual calculation
b) actual history. would be more refined than our
100 IC's
60 Transistors simple example. We would probably
50 Resistors use failure rates for categories of
50 Capacitors parts or we may use the failure rate
The prediction technique uses a for each particular component (e.g.
method that represents the expected Past history will have established av- one IC may have a failure rate of
0.02% and another may fail at a 1% MTBF's based on empirically de- users' repair histories. In 3
rate). Refining the prediction rived data assumes that the manu- some cases, the actual failure
process in this manner would give facturer has the "facts". rate may be a factor of 10 or
increased accuracy. 15 times the reported failure
However, the problem now becomes
rate.
There are two main sources of error one of interpreting the empirical
with this prediction method. data. Subjectivity may cloud the 3. MuI ti ple fai Iure.
results. For example:
The reliability of a newly developed An instrument fails this
component is not always known and 1. What constitutes a failure? (Cer- month and is "repaired". Next
we will have to estimate it. Also, so- tainly a bona fide component fail- month, the same instrument
called known components are not ac- ure should be counted.) fails with the same symptoms
tually known for the application and is repaired again. The
(stress). In addition to controlled stres- a. Failure to meet specs? manufacturer may interpret
ses, the actual instrument operating a b. Operator misinterpretation this data as one failure on the
50