Date: Fri, 4 Jul 2014 21:06:07 -0400
Reply-To: David Beierl <dbeierl@ATTGLOBAL.NET>
Sender: Vanagon Mailing List <vanagon@gerry.vanagon.com>
From: David Beierl <dbeierl@ATTGLOBAL.NET>
Subject: Start of Draft: Analog VOMs for vans, was Re: AFM test readings
Content-Type: text/plain; charset="us-ascii"; format=flowed
[Post rejected for length, so splitting it in two]
Dear Volks,
Here is quite a lot of info about analog meters in general for people
who aren't used to them, together with my thoughts about two general
classes of meter and several specific meters in each class, prices
ranging from under ten to over three hundred dollars. There are a
few oddballs but in general the classes are a) very small inexpensive
low-sensitivity meters that are fine for the AFM but not for oxygen
sensors; and b) moderately sized medium-high sensitivity meters
mostly about four and a half to six inches tall that are good for
both. I don't discuss any full-size bench meters or electronically
driven meters.
I can't begin to describe any of them completely, but for each one I
try to point out salient characteristics that may help focus your
choices both among the ones I talk about and among others you may
find. Many of them have very useful testing ranges for 1.5V and 9V
batteries that place a bit of load on them to get a much more
realistic reading than open-circuit voltage. I don't discuss the
presence or characteristics of any of these or most other features
not aimed at the van and particularly at the AFM and O2 sensor. I've
barely glanced at AC ranges. Also keep in mind that I've never laid
eyes on most of them.
One thing I neglected and am not going to go back and sort through
now is whether there is a DC range on a given meter that is
convenient for working with 12V systems. Your digital multimeter
will no doubt take care of that splendidly, but it doesn't hurt to be
aware of the question. And you should be aware that analog meters
have much larger errors in general than digitals; and accuracy is
specified in such a way that they can be effectively huge at low
readings. I hope and believe that although it could be better
organized, almost everything I've written here is useful and the rest
is at least interesting. But if you feel like skipping all the
background at least be sure you understand the accuracy part.
I put a lot of work into this, but it's basically a draft that
outgrew its beginnings and I know it could be much better organized
and lose some fat. But I'm done with it for the moment. If any
decent writer cares to have a whack at editing it I solicit your
suggestions. Likewise of course anyone please point out any errors
or gaps or bad explanations so I can fix them. I'm past seeing any
myself at this point, other than mentioning sensitivity a lot but not
thoroughly explaining it. I hope what I say in context works well
enough without the full explanation, but I'll fix it if I have to.
Mister Squirrel started this but all the explanations aren't for him
because I'm sure he could have written them himself.
Yours,
David
At 07:56 PM 7/3/2014, Rocket J Squirrel wrote:
>The question is -- who has a good value in an analog multimeter these
>days? Simpsons were my go-tos in Ye Days of Olde when Knights Were Bold
>and I had a budget.
SMALL METERS and a touch of explanation:
The little pocket multimeters for $20 or so from anybody would be
ideal for this use except that their lowest DC range is probably six
volts or so which is a bit high for this. Also they'll have
sensitivity (internal resistance) of 2,000 ohms per full-scale volt
(e.g. 6V range would have resistance of 12,000 ohms) so they'd be
useless for checking oxygen sensor activity in closed-loop mode,
which is the other place on the van where you want an analog meter or
scope.** However here's one from Extech with a 2.5VDC scale for
$20. No good for O2 sensor but great for AFM. Extech has been
around for a long time now, with various kinds meters that are
comparatively low priced, at least if you compare them with Fluke.
I've had at least one or two of their digital meters over the years
with no complaints. The one I'm sure of was stolen, which I can't
blame on the meter. Markings are clear, meter face crowded and
wastes space on an anti-parallax mirror which is just silly for a
meter like this, some ranges are not factors of ten of the 10/50/250
meter scales, which will drive you nuts. Test leads appear to be built in.
http://www.transcat.com/catalog/productdetail.aspx?itemnum=38073
http://www.transcat.com/PDF/38073.pdf
Here's a throwaway one with a 2.5 VDC range, $9 at Amazon:
http://www.amazon.com/Mastech-YG188-Pocket-size-analog-multimeter/dp/B00064CH6A/ref=pd_sim_hi_4?ie=UTF8&refRID=13CK3Z0VH2A2X08XNYT7
. Marking readability is mixed and I can't find a good enough image
to distinguish the scale numbering, however ranges are some variation
of 2.5/5/10 so the scales are likely easy to understand. Test leads
are built in.
Elenco M105, $17 at Amazon. This is an exception to the others in
this section, as it is 10,000 ohms per volt sensitivity instead of
2,000. On the ten volt range it should be usable to get an idea what
the oxygen sensor is doing. 2.5V DC lowest range. The ohmmeter is
very limited, 100 ohms midscale on the low range and 10 kohms on the
high range. But it's compact and well built and can serve both our
prime Vanagon applications. The reviewers love everything about
it. Markings are clear, meter is easily readable and all ranges are
a factor of ten of the 10/50/250 main scales. Well worth considering
if you want a quite compact meter that's still quite sensitive, and
can live with the +/- 5% FS DC accuracy and the crappy
ohmmeter. There's no range that's good for 12 VDC, but that's what
your little digital is for. 4.5x2.7x1.2 inches.
http://www.amazon.com/Elenco-M105-Range-Compact-VOM/dp/B0002HQVFO/ref=cm_cr_pr_sims_t
**Or an indicator based on the National Semiconductor LED bar graph
driver chip (LM-can'trememberthenumber), like the little indicator
board Ken Lewis sells. It's truly ideal for the purpose. It won't
load the sensor, it responds in milliseconds, gives you a ten-segment
bar or dot graph, draws practically no current if it's not lighting
up a segment; and if you want more resolution you can cascade as many
chips as you want. You might even be able to cascade Ken's boards
with a little creative wiring and a few resistors.
========================================================================
========================================================================
LARGER METERS and a lot more explanation:
All the meters below are at least 20 kohms per volt on DC ranges, so
on a 5V scale they'd have 100K resistance which shouldn't drag down
the oxygen sensor *too* badly but still lets you see the meter
pointer move. Most do not have a high-amps range and max out at a
quarter or half amp.
People without analog meter experience should at least see the note
on quoted accuracies under the Radio Shack entry and the note on ohms
scales under the UEI meter.
All or almost all of these meters use the same jacks for the current
ranges as for voltage. If you hook them to twelve volts or really
any voltage source when they're switched to a current range either a
fuse will blow or the meter will try to bend itself around the pin
(and might succeed) and the current shunt for that range will
smoke. Try to do better than my record of about five minutes when I
was given my first meter at age six or seven. Smoke *and* bent
pointer , whee. And of course I didn't dare admit it.
Digital meters these days use very low voltages and maximum currents
on the ohms ranges. Very hard to hurt things with them, and the
voltage isn't high enough to turn on semiconductors, which can be a
convenience (that's why they have to have a separate diode check
range). These analogs are different, they apply anywhere from 1.5
volts up through 9 or even 22 on the various ranges, and many can
probably put 150 mA or more through zero ohms on the x1 scale. They
*will* turn on semiconductors, and can potentially overheat or smoke
small components on the low ranges. The ones using more than three
volts on the high ranges can potentially ruin sensitive semiconductor
components by overvoltage. And they have to be re-zeroed on each use
and each time you switch ranges, especially when switching to a range
that uses a different battery. You can still check diodes but have
to choose the range thoughtfully, and the reading won't tell you what
the forward voltage is without either measuring it with a second
meter or working up a table of resistance readings vs voltage on the
ranges of interest for your particular meter.
Any meter you look at, check the correspondence between the ranges
and the scales printed on the meter. Any range that isn't a factor
of ten of one of the printed scales will drive you mad. I have a
little Radio Shack clamp meter around somewhere that's a clever
little thing, but hardly any of the ranges are direct reading and
it's impossible to use. Multiply this scale by two, divide that one
by three...wrong. You'd have to tattoo the manual on your wrist.
Radio Shack's $30 22-037 meter (20 kohms/volt DC, 2.5 VDC range,
+/-3% of full-scale DC accuracy,*** no high-amps range, 50 ohm
continuity buzzer and uses 3 volts for ohmmeter so it should light up
red/green/orange/yellow LEDs) sounds good but the recent reviews on
both RS analog meters point to dreadful build quality/QA or accuracy
troubles. Markings seem readable but the AC ranges are marked in
red and to me have lower contrast against the case. All
voltage/current ranges are powers of ten of the 10/50/125/250 scale
markings. But given the reviews I'd stay clear, especially since
reviewers compared unfavorably against previous RS meters.
==================================================
***(Long note, got even longer.) Analog meter accuracy is specified
much differently from digital meters and it can bite you if you're
not used to it. Digital meters on a given range are specified as +/-
m% of the reading, +/- n counts on the display. That implies that a
low reading on a given scale will have an error of similar proportion
to the actual value as a high reading will (the lower the reading,
though, the more the +/- so many counts part will influence it, so
high-count readings are still the most accurate).
But on an analog meter, the voltage and current scales will be
specified as +/- m% *of full scale* so that the allowed percentage
error gets completely ridiculous as you get near the bottom of the
scale. A ten volt scale with +/- 3% would allow an error of +/-
three tenths of a volt anywhere on the scale, which would be a 30%
error at one volt. For this reason people are advised to distrust
the actual values of readings below about one-third or even one-half
scale (however a higher input will always produce a higher reading
than a lower input, which is not always strictly the case with a
digital meter). Also, +/- 1.25 of FS on DC is the best spec I'm
aware of for an analog meter (Simpson 270-5/5RT), and only high-class
meters meet 2% (3% is more common); whereas even a very cheap digital
meter may well have a basic DC accuracy of 1% of the reading +/- one
or two counts and fancy ones run a quarter per cent or
better. Harbor Freight's $6 meter is specified +/- 0.5% +/- one
count on the 200 mV DC range and +/- 1% +/- two counts on its other
DC voltage scales. Both analog and digital meters are generally most
accurate on their DC voltage scales.
The analog ohms scale is odd because it runs backwards and has much
better resolution on the low-resistance end of the scale which is
also the high end of the meter scale where the movement is most
accurate. Accuracy is specified in +/- m% of the total length of the
scale rather than in terms of a specific value, so you have to
measure the scale to know the accuracy in ohms for a given resistance
being tested.
Typically analog meters will do noticeably better than the spec on
low readings, but there's no promise. And errors on the high end
won't necessarily be linear. Inaccuracy in the network of resistors
used to change ranges will cause errors that may be different on
different ranges, but on a given range will be the same percentage of
different readings, high or low. But non-linearities in the meter
movement will cause varying errors that affect every range, causing
the pointer at a given point on the scale to deviate *by an angle*
that may be different from that at every other point but doesn't
change when you switch ranges. My hand-held analog meter which as I
recall cost about eighty bucks ten years ago has a very irritating
bump right in the vicinity of 12v (on a 15 or 20 volt scale, I
forget) and of course a similar location on other scales. Ten volts
reads as ten, fourteen volts reads as fourteen; but twelve reads as
12.3. It's in spec, but it bugs me.
Errors in zeroing the pointer (the real mechanical zero, not the
backwards ohmmeter zero) will cause an angular error that diminishes
proportionately at higher readings. All problems are exaggerated
around the zero point because in theory everything is perfectly
balanced and perfectly frictionless, and an infinitely small input
will cause an infinitely small force that will move the pointer an
infinitely small amount. In practice of course it doesn't work that way.
Most analog meters are specified to be read in horizontal position,
whether or not they have a tilt stand or feet on the short side of
the case. The zero may shift if you set them at other angles
depending on how and how closely the pointers are balanced, and the
accuracy will probably decline.
The next bit is about how meter movements are built,, how to
recognize a certain sign that you just paid too much for a fancyish
shock-resistant meter that turns out to be counterfeit, and the
compromises and interactions among price, sensitivity, physical size,
and how the pointer responds (the last is called ballistics). I
think it's both interesting and useful, but if you want to skip to
specific meters head down to the next double line of ==============.
The movement in most meters is of the type called D'arsonval, because
he invented it. The pointer is typically made of extremely thin
aluminum tubing, sometimes with the far end squeezed shut to make it
yet thinner. It's mounted on a shaft which seats in jeweled bearings
at either end, and there are two hairsprings that make electrical
connection to a coil of wire wrapped around a rectangular form that's
mounted surrounding the shaft. The pointer has a short stub
extending to the rear, and two more stubs extend to the sides. These
three stubs carry small coils of wire wrapped around them to balance
the pointer in all directions so that ideally it does not change
reading when the meter is held at a different angle. Adjusting these
weights is very delicate work, so most pointers probably aren't quite
balanced and will shift zero a bit when you hold them at different
angles. The bigger and more sensitive the meter, the more critical
the balance.
The entire assembly is placed inside the gap in a circular magnet so
that it tends to rotate when current passes through the coil, and the
hairsprings also provide the restoring force that the magnetic field
works against. The upper hairspring has an adjuster attached so you
can adjust the pointer zero point. This movement serves very well,
but the jewels are easily cracked which makes the pointer stick or
otherwise misbehave, or they can get sticky for other reasons. And
at very low readings the tiny friction may be enough to disturb the
reading slightly. Here's a photo of what's visible from the front of
a D'arsonval meter with its cover/window removed. There's a pin
mounted eccentrically on a screw-head in the meter cover, and this
engages the fork you can see marked as Zero Adjust. As you rotate
the screw-head the fork moves first one way, then the other, so there
are two points in its rotation that will leave the adjuster in the
same
place.
https://www.google.com/search?q=d%27arsonval+movement&rlz=1C1KMZB_enUS518US518&es_sm=93&tbm=isch&tbo=u&source=univ&sa=X&ei=hta2U_PEJ8-NyATAq4HICg&sqi=2&ved=0CCoQsAQ&biw=1746&bih=905#facrc=_&imgdii=_&imgrc=povq0GZ2s8KY7M%253A%3BUtR2wXBFMGNlMM%3Bhttp%253A%252F%252Fwww.repairfaq.org%252Fsam%252Fmmanat1.jpg%3Bhttp%253A%252F%252Fwww.repairfaq.org%252Fsam%252Ffaqfil.htm%3B603%3B390
There's another movement that's used in meters needing shock
resistance, excellent linearity, better accuracy at low readings and
suchlike. It's called a taut band movement, and it's just like the
D'Arsonval except there are no bearings and no hairsprings and no
shaft as such. Instead a thin springy ribbon is mounted to the top
and bottom of the coil form, and the two ends stretched across a
stiff C-shaped spring. The force of the spring keeps everything
stretched out and the ribbons lie straight and flat with the needle
at the zero point. The ribbons also supply the connections to the
coil. When current passes through the coil it rotates, swinging the
pointer and twisting the ribbons, which makes them a tiny bit
shorter. This shortening pulls the ends of the spring together very
slightly but proportionately to the pointer rotation, and this
supplies the restoring force to bring the pointer back to zero. This
is an expensive way to do it, but the resulting assembly is very
resistant to shock and has no friction whatsoever. Here's a photo of
one of the first taut band movements, built by HP in 1964. It's
rotated 180 degrees from how you'd usually see it -- the short white
thing pointing down and to the right is the stub of the pointer, and
the adjuster yoke is sticking straight up. But you can clearly see
the lack of jewels and hairsprings. I'm making a point of this
because if you buy a meter that uses a taut band movement and it
turns out to be a counterfeit, almost surely the actual movement will
be of the D'Arsonval type which you can instantly recognize.
Every meter movement is a compromise. It's easy to make a
slow-responding pointer that doesn't tend to overshoot its reading,
not so easy with a fast one. It's easy to make a fast-responding
movement that uses a lot of power and drives a short pointer, but the
longer the pointer and the more sensitive the movement must be, the
harder it gets to make it respond quickly. So big sensitive 50
microamp meters tend to be slow, and big *cheap* sensitive meters
downright leisurely; but small 500 microamp ones can easily be
reasonably brisk. But a meter (even a fairly small one) that meets
the technical specs for a VU meter (90% of full reading in 300
milliseconds, and no overshoot at all) is going to be quite expensive
and not very sensitive.++
And if you for some bizarre reason want a big meter that is quick but
overshoots like mad, can hardly stand still -- you go to an outfit in
say New Hampshire that makes big quick expensive meters and special
order some with most of the damping taken out, and you take these now
pretty useless but impressive movements and mount them in something
called an E-meter that you sell for big money to every other
Scientologist in the land who wants to amount to anything. The
underdamped pointers make them considerably more impressive to
watch. The first few seconds of this may give you an idea of it:
http://youtu.be/_YCoqYyQXqc?t=2m42s
++There are (or were) meters labeled VU on every tape recorder and
all sorts of other audio gear, but very few of them met both specs to
legitimately wear the label. If they weren't slow they
overshot. And if they did either they weren't very useful for their
main purpose, which was to enable the operator to use a meter that
responded only to average levels *and his own knowledge of the
characteristics of what was being recorded* to set the levels
properly so that the peaks didn't distort but the average level was
as high as possible to minimize tape hiss. If you were recording
piano music for example, you had to set the levels to read very low
because each piano note starts with an extremely loud peak that
quickly decays; but for a group of singers you could run the needle
right up. If the meter didn't behave the way it was supposed to you
didn't know what the readings really meant even though the steady
reading would be the same as a "real" VU meter. All this quickly
went away as soon as we learned to make various kinds of electronic
peak-reading meters which could be both extremely fast and quite
cheap since they didn't have to swing a big heavy pointer all about,
though there was a stage when you still had an analog meter but also
one or two LEDs that would flicker if the peaks got too high.
End of part 1, completed in part 2.