Can a cheap multimeter be more accurate than built-in sensors?

[Mxsmanic]

Would an inexpensive digital multimeter give me more accurate readings
for voltages from the PSU than the built-in sensors that I see
displayed by programs that monitor them? I recall someone somewhere
saying that built-in sensors weren't that reliable.

How much should I expect to pay for a decent meter? I've seen some as
low as nine dollars and I'm wondering how accurate they are for the
price.

 

[Franc Zabkar]

Would an inexpensive digital multimeter give me more accurate readings
for voltages from the PSU than the built-in sensors that I see
displayed by programs that monitor them? I recall someone somewhere
saying that built-in sensors weren't that reliable.

The motherboard's voltage monitoring ADC sees the voltage on the
motherboard, not the voltage at the PSU. Consequently the ohmic losses
in the cables and PCB traces mean that the multimeter always reads
higher. The motherboard's ADC is also limited to a typical resolution
of 16mV per count when measuring the lesser voltages and 64mV/count
for the +12V rail.

How much should I expect to pay for a decent meter? I've seen some as
low as nine dollars and I'm wondering how accurate they are for the
price.

Even a cheap DMM should do better than +/-2%. My two DMMs show 4.96
and 4.97 when connected to a 5.000V precision reference. These
readings have not changed since I last checked them in March of this
year.

You can build your own single-chip precision voltage reference using
Maxim's MAX6350 (5.0V) or MAX6325 (2.5V):
http://pdfserv.maxim-ic.com/en/ds/MAX6325-MAX6350.pdf


"The MAX6325/MAX6341/MAX6350 are low-noise, precision
voltage references with extremely low, 0.5ppm/°C
typical temperature coefficients and excellent, ±0.02%
initial accuracy. These devices feature buried-zener
technology for lowest noise performance. Load-regulation
specifications are guaranteed for source and sink
currents up to 15mA. Excellent line and load regulation
and low output impedance at high frequencies make
them ideal for high-resolution data-conversion systems
up to 16 bits."


- Franc Zabkar

 

[kony]

The motherboard's voltage monitoring ADC sees the voltage on the
motherboard, not the voltage at the PSU. Consequently the ohmic losses
in the cables and PCB traces mean that the multimeter always reads
higher. The motherboard's ADC is also limited to a typical resolution
of 16mV per count when measuring the lesser voltages and 64mV/count
for the +12V rail.


Even a cheap DMM should do better than +/-2%. My two DMMs show 4.96
and 4.97 when connected to a 5.000V precision reference. These
readings have not changed since I last checked them in March of this
year.

You can build your own single-chip precision voltage reference using
Maxim's MAX6350 (5.0V) or MAX6325 (2.5V):
http://pdfserv.maxim-ic.com/en/ds/MAX6325-MAX6350.pdf


"The MAX6325/MAX6341/MAX6350 are low-noise, precision
voltage references with extremely low, 0.5ppm/°C
typical temperature coefficients and excellent, ±0.02%
initial accuracy. These devices feature buried-zener
technology for lowest noise performance. Load-regulation
specifications are guaranteed for source and sink
currents up to 15mA. Excellent line and load regulation
and low output impedance at high frequencies make
them ideal for high-resolution data-conversion systems
up to 16 bits."

That circuit is a good test of a cheap multimeter's accuracy
as it cannot be cost effective to have one calibrated (if
anyone would even attempt to do such service on a cheap
one?). The really, really cheap ones I have observed to be
off more than +-2% though, especially after they've aged a
few (too few) years. Unfortunately having the reference
doesn't easily allow one to calibrate those cheap ones, as
they dont' have internal adjustment and would have to be
reverse engineered and parts substitutions made, usually.

Even so, I fully agree that any (working as well as the
manufacturer *intended*) multimeter should be higher
accuracy than the motherboard sensor, particularly at
measuring voltage at any given point in the circuit(s),
which is obvious enough but to someone who doesn't have a
multimeter yet.

 

[Conor]

Would an inexpensive digital multimeter give me more accurate readings
for voltages from the PSU than the built-in sensors that I see
displayed by programs that monitor them?
Yes.
I recall someone somewhere
saying that built-in sensors weren't that reliable.

How much should I expect to pay for a decent meter? I've seen some as
low as nine dollars and I'm wondering how accurate they are for the
price.

Look at digital ones.


--
Conor

If Pac-Man affected us as kids, we'd all be running around in darkened
rooms, munching magic pills and listening to repetitive electronic
music.

 

[Felger Carbon]

How much should I expect to pay for a decent meter? I've seen some as
low as nine dollars and I'm wondering how accurate they are for the
price.

http://www.case-mod.com/store/advan...5a2a4fb1bd93b0e60&keywords=digital+multimeter

$4.95 plus S&H. Price includes the 9V battery and very decent test
leads. Does not include automatic timed off switch, so if you leave
the meter on, the battery will be history in a very few days. The
whole thing costs less than a 9V battery alone, so buy two or three.
Give them away as party favors.

Calibration? Back when I built a Heathkit vacuum tube VOM, we used a
new Everready carbon battery, which can be depended on to measure
1.55V. Dunno if that's true of alkaline batteries (Duracell), so use
Everready, the bunny brand.

If you don't like the long URL above, go to www.case-mod.com and
search "digital multimeter".

 

[Andrew Smallshaw]

That circuit is a good test of a cheap multimeter's accuracy
as it cannot be cost effective to have one calibrated (if
anyone would even attempt to do such service on a cheap
one?). The really, really cheap ones I have observed to be
off more than +-2% though, especially after they've aged a
few (too few) years. Unfortunately having the reference
doesn't easily allow one to calibrate those cheap ones, as
they dont' have internal adjustment and would have to be
reverse engineered and parts substitutions made, usually.

Even so, I fully agree that any (working as well as the
manufacturer *intended*) multimeter should be higher
accuracy than the motherboard sensor, particularly at
measuring voltage at any given point in the circuit(s),
which is obvious enough but to someone who doesn't have a
multimeter yet.

Try using a proper multimeter: the pocket types are guaranteed for +/-
5 or 10% in general. My bench meter is guaranteed +/- 0.5% though that
precision costs more than any $9, and indeed it takes a new NAMAS

certificate every year. Pocket meters have their role but precision
measurement isn't among them.

 

[Mxsmanic]

Try using a proper multimeter: the pocket types are guaranteed for +/-
5 or 10% in general.

How much does a "proper" multimeter cost?

Pocket meters have their role but precision
measurement isn't among them.

How do they compare to motherboard sensors?

 

[kony]

Would an inexpensive digital multimeter give me more accurate readings
for voltages from the PSU than the built-in sensors that I see
displayed by programs that monitor them? I recall someone somewhere
saying that built-in sensors weren't that reliable.

Yes, almost any meter can be expected to be far more
accurate. However, like any other product there's also some
really poor no-name junk. The goal should not be to find
the absolute cheapest meter but rather, one fitting the
budget that you can expect to rely on for at least a few
years. Resist the urge to buy one with lots of features you
don't need, over a more quality meter at any given price
point.

There are $50 meters that seem to do a lot, and they're no
better (if as good) at the basics as a $30 meter.

How much should I expect to pay for a decent meter? I've seen some as
low as nine dollars and I'm wondering how accurate they are for the
price.

It's a hard question to answer, not knowing how much you
might use it for other tasks... or if you'd ONLY ever use it
for measuring PC power voltage levels. Many people prefer
autoranging meters, at least if they can also be set to
manual mode and have good resolution. Generally speaking,
in the lower cost meters there is a direct correlation
between size and quality. A really small pocket usually
isn't very good, not even considering the much smaller
display and crude fragile probes. Good probes cost more
than the entire $10 meter with cheap probes.

One brand I've heard recommended a few times for good value
is Protek. It's not hard to find anyone-and-their-brother
swearing by Fluke. If I had to suggest one meter with fair
quality, a great value for basic PC tests then it'd probably
be a (sears) Craftsman #81437. Most of the Craftsman meters
aren't noteworthy but the 81437 is a rebranded Fluke meter,
comes with a certificate of calibration too... and can be
recalibrated. At Sears stores they run about $50, maybe
less on sale, but you can pick them up on ebay for about 1/3
that, primarily because few people know they're Flukes in
disguise... the same Fluke branded meter goes for (average)
over $70 on ebay. I mention it because it's well worth the
extra ~ $10-20 over a $10 meter, but otherwise I'd suggest
avoiding anything in the $10 class and at least looking at
$20. I have that $5 meter another poster linked and it's
cute because of how small it is, but as far as meters go,
it's junk even for a low-end meter. You can also get the
same thing at Harbor Freight every-other month for $3 each
in a different color (yellow, IIRC).

Examples,

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=7539377376

http://www.sears.com/sr/javasr/product.do?pid=03481437000

Under $20 it's hard to find anything of decent quality, even
ignoring accuracy... that first $20 is almost a prerequisite
for a meter but if you really want the cheapest thing that's
not total junk then I would suggest this,

http://www.itcelectronics.com/pdtl.asp?P=6251

 

[kony]

Try using a proper multimeter: the pocket types are guaranteed for +/-
5 or 10% in general.

The proper multimeter for field, PC voltage testing is a
pocket type. It is pointless to take a PC to a bench to
gain a level of accuracy not needed, as many pocket meters
do in fact have better than 5 or 10% accuracy, particularly
those that come with (or are set away for) calibration.
These are, like anything else, more expensive pocket meters
meant for professional use, not a $10 hardware store
special. For example, some Fluke pocket meters come spec'd
and certified at under 0.1% accuracy (for this posts' use,
DC voltage). Of coure, they too can be certified and
recalibrated, IF the task(s) required that level of
precision.

While I have seen some in excess of 5%, in general they are
not ever spec'd as loose as 10%. I'd consider 2% about
average for a low-end meter... at least when new, though
after aging it may drift off more.

Inbetween poor and great, there's an ocean of meters.
"Typically" an average market price of around $40 or more is
needed to get above poor junk, not just accuracy but
features and sturdy construction. Even so, it can still be
expected that the average $20 meter is sufficient for basic
voltage tests many enthusiasts would perform, such as basic
measurements of the power supply rail DC level. It's not
the ultimate test but a distinct increase in accuracy over a
motherboard voltage sensor's report.

My bench meter is guaranteed +/- 0.5% though that
precision costs more than any $9, and indeed it takes a new NAMAS

certificate every year. Pocket meters have their role but precision
measurement isn't among them.

Great, so long as you need that level of precision. One
need not spend as much as the yearly certification alone for
a pocket meter sufficient to get useful data on many PC
voltage levels from an end-user or troubleshooter
perspective.

 

[Mxsmanic]

It's a hard question to answer, not knowing how much you
might use it for other tasks... or if you'd ONLY ever use it
for measuring PC power voltage levels.

I'd probably only use it for PC voltage levels, although I suppose in
a pinch I might want to check line voltage from a wall receptable, or
continuity in a cable.

Generally speaking,
in the lower cost meters there is a direct correlation
between size and quality. A really small pocket usually
isn't very good, not even considering the much smaller
display and crude fragile probes. Good probes cost more
than the entire $10 meter with cheap probes.

Size isn't a problem, since I won't be carrying it around much. It'll
just sit in a corner most of the time.

One brand I've heard recommended a few times for good value
is Protek. It's not hard to find anyone-and-their-brother
swearing by Fluke. If I had to suggest one meter with fair
quality, a great value for basic PC tests then it'd probably
be a (sears) Craftsman #81437. Most of the Craftsman meters
aren't noteworthy but the 81437 is a rebranded Fluke meter,
comes with a certificate of calibration too... and can be
recalibrated.

I can't get Craftsman in France, but I'm sure the other two must be
available.

Under $20 it's hard to find anything of decent quality, even
ignoring accuracy... that first $20 is almost a prerequisite
for a meter but if you really want the cheapest thing that's
not total junk then I would suggest this,

http://www.itcelectronics.com/pdtl.asp?P=6251

I'll try to save up a bit more and get something a bit better. No
sense in buying one if it's so junky that it provides no improvement
over the motherboard sensors.

 

[kony]

I'll try to save up a bit more and get something a bit better. No
sense in buying one if it's so junky that it provides no improvement
over the motherboard sensors.

I may have mislead a bit. Any cheap meter that works
*properly* should be more accurate than the motherboard
sensors, even more so because the probes allow taking
voltage at any specific point. The very best one could
hope for from the motherboard sensor is an accurate reading
of that specific point on the board where it takes the
samples.


For example, if the motherboard sensor reported 11.6V for
the 12V rail, the multimeter can check for 12V rail level at
the power supply plugs, at the fan headers, onboard power
regulation supply, drive plugs, etc. It is useful to know
where the voltage deviates from desired values when it does.

 

[Guest]

How much does a "proper" multimeter cost?

I've seen them for as little as $1.99, but $5-15 is more common.

How do they compare to motherboard sensors?

Motherboard sensors are commonly wrong by 2%, but I've seen errors as
large as 5%, while the worst rated accuracy for the cheapest digital
meter is 2% for DC volts, with 1% being very common, and the real
accuracy will typicaly be twice as good as the rating. Try to get a
meter rated for a count of at least 3 1/2 digits, that is, it displays
0000-1999 rather than 2 1/2 digits - 000-199, because the least
significant digit can't be guaranteed to be correct and may be off by a
count of at least 1, sometimes 5 for really bad meters. The display
error for readings just slightly beyond the 199 count can be huge. For
example the 2.5000V used for most DDR memory will be shown as 2.5, but
because the count accuracy can't be better than 1, this reading can't
be trusted to better than 4%, and if the count accuracy is 5, the
reading's accuracy is only 20%. With a 3 1/2 digit display the count
error is 10 times smaller, or 0.4% or 2%, respectively.

 

[David Maynard]

Try using a proper multimeter: the pocket types are guaranteed for +/-
5 or 10% in general.

5 and 10%? I don't know of ANY that are that bad, at least by claim.

Here's a typical 'throw away' 5 buck meter and even it claims +/-.5% on the
DC scale (up to 200) with 'worst' being 1.5% on AC.

http://www.surpluscomputers.com/store/main.aspx?p=ItemDetail&item=TOL10061

 

[Franc Zabkar]

How much does a "proper" multimeter cost?

I'd be very skeptical of the 10% claim. In fact even 5% sounds like an
exaggeration. IME a cheap DMM will be more than adequate for your
purposes. Just out of curiosity, I'll take my 5.000V precision
reference to my local electronics store and test the two cheapest DMMs
(AU$12 and AU$16).

How do they compare to motherboard sensors?

Here are the results of my testing earlier this year:
http://groups.google.com/group/alt.comp.hardware.homebuilt/msg/6ede1e075631c675?dmode=source&hl=en


- Franc Zabkar

 

[Franc Zabkar]

Try using a proper multimeter: the pocket types are guaranteed for +/-
5 or 10% in general. My bench meter is guaranteed +/- 0.5% though that
precision costs more than any $9, and indeed it takes a new NAMAS

certificate every year. Pocket meters have their role but precision
measurement isn't among them.

Today I checked several cheap DMMs at my local electronics store. I
tested 4 samples of the least expensive meter (Dick Smith Electronics
cat# Q1469, AU$12). Using the precision 5.000V +/- 0.001V reference
alluded to elsewhere in this thread, I measured 5.01, 5.10, 5.01, and
5.10. That's 2% at worst. Tests of other meters (Q1467P at 2 for
AU$17, and Q1459 palm DMM at AU$23) ranged from 5.01 to 5.06.


- Franc Zabkar