EMRFD Message Archive 7136

Message Date From Subject
7136 2012-01-03 05:25:01 ha5rxz Ohm Meter Calibration
Happy New Year to everyone!!

Somebody just asked me this question and I have no idea of the answer, I therefore turn it over to all of you in a search for clarity.

Suppose I have a very accurate 1,000 ohm resistor, when I measure it on my multimeter I get a reading of 998 ohms. For a correct reading of all other resistors on this range should I:

a) Add 2 ohms to all readings?

b) Multiply all readings by 1.002?

Gut feeling says multiply because the the problem of a zero ohm resistor can be dealt with but I'm not sure.


Peter HA5RXZ
7137 2012-01-03 05:34:19 Fábián Tamás L... Re: Ohm Meter Calibration
Happy New Year Peter!

To get a precise picture of how the multimeter cheats, you should
measure lots of various resistances.

Assuming that the deviation is linear, you should make two measurements.

You said that you have a very precise 1000 Ohm resistance. In this case
you actually have two: the 1K and a 0R (measurement leads shorted).

My opinion is the following:

You should short the leads and add / substract a value so the multimeter
reads 0 Ohms.

Measure the 1000 Ohm resistor, and multiply all readings so that you get
the correct value.

The order of the two correction circuits are critical. Things can get
even more complicated if you think about that the leads and the
multimeter circuitry itself can introduce various errors.

Any deviation you notice after such correction is because of the
nonlinearity of the multimeter (or the add / multiply circuitry you add).

I hope it helped.

2012-01-03 14:25 keltezéssel, ha5rxz írta:
>
> Happy New Year to everyone!!
>
> Somebody just asked me this question and I have no idea of the answer,
> I therefore turn it over to all of you in a search for clarity.
>
> Suppose I have a very accurate 1,000 ohm resistor, when I measure it
> on my multimeter I get a reading of 998 ohms. For a correct reading of
> all other resistors on this range should I:
>
> a) Add 2 ohms to all readings?
>
> b) Multiply all readings by 1.002?
>
> Gut feeling says multiply because the the problem of a zero ohm
> resistor can be dealt with but I'm not sure.
>
> Peter HA5RXZ
>
>



[Non-text portions of this message have been removed]
7138 2012-01-03 05:40:48 Tim Hills Re: Ohm Meter Calibration
Peter:

I was just wondering what your working on that would need better than
0.2% accuracy and what type of resistor your measuring.

There is such as thing as "close enough" and I find 5 or even 10% "close
enough" in most cases.

Tim Hills
Sioux Falls, SD

7139 2012-01-03 06:07:44 William Carver Re: Ohm Meter Calibration
I agree you first subtract the reading when the ohmmeter leads are
shorted. Some meters (my Fluke 187 for example) has a button for this
purpose.

After that (assuming the meter is an electronic DMM) the measured value
will deviate from actual resistance either because the test current
passing through the unknown resistor isn't the precise value it should
be, or the voltmeter reading the voltage across the unknown resistor is
off a bit. Either way, after that zero subtraction, you would multiply
all readings ON THAT SAME RANGE by the 1.002.

If you switch to a different range the voltmeter and/or test current
have been changed, so each range should be checked independently and may
have its own correction multiplier.

Bill - W7AAZ
7140 2012-01-03 08:56:17 ha5rxz Re: Ohm Meter Calibration
Being a very expensive (theoretical) meter it does of course read 0.00 ohms with the leads shorted. Thank you for your replies gentlemen, they have been passed
7141 2012-01-03 11:58:26 KK7B Re: Ohm Meter Calibration
One should always be aware of not just the component value, but the tolerance needed for any given application. 5 and 10% tolerance is good enough for some things. Any time you want balance, or have a circuit that is sensitive to differences between channels, you will want better. For a graphical look at when you might need 0.1% accuracy, see figures 9.58 through 9.61 in EMRFD. The conclusion from that section is that 0.2% tolerance is good enough for 3rd order all pass networks used in IQ systems, but many authors have chosen 4th order systems to cover the speech bandwidth. For those, there is some benefit to matching resistors and capacitors to better than 0.1%

Modern resistors and capacitors at ambient temperature will hold those tolerances over time, which makes analog phase shift networks attractive for systems that can be measured and built once, and still meet specs 2 decades later.

Enjoy the experiments,

Rick KK7B

>
> Peter:
>
> I was just wondering what your working on that would need better than
> 0.2% accuracy and what type of resistor your measuring.
>
> There is such as thing as "close enough" and I find 5 or even 10% "close
> enough" in most cases.
>
> Tim Hills
> Sioux Falls, SD
>
>
7142 2012-01-03 12:04:36 William Carver Re: Ohm Meter Calibration
Hey guys, he was just asking a theoretical question: if you want to
correct ohmmeter readings from one know resistor, do you multiply or
add/subtract.

On Tue, 2012-01-03 at 19:58 +0000, KK7B wrote:
>
>
>
> One should always be aware of not just the component value, but the
> tolerance needed for any given application. 5 and 10% tolerance is
> good enough for some things. Any time you want balance, or have a
> circuit that is sensitive to differences between channels, you will
> want better. For a graphical look at when you might need 0.1%
> accuracy, see figures 9.58 through 9.61 in EMRFD. The conclusion from
> that section is that 0.2% tolerance is good enough for 3rd order all
> pass networks used in IQ systems, but many authors have chosen 4th
> order systems to cover the speech bandwidth. For those, there is some
> benefit to matching resistors and capacitors to better than 0.1%
>
> Modern resistors and capacitors at ambient temperature will hold those
> tolerances over time, which makes analog phase shift networks
> attractive for systems that can be measured and built once, and still
> meet specs 2 decades later.
>
> Enjoy the experiments,
>
> Rick KK7B
>
> >
> > Peter:
> >
> > I was just wondering what your working on that would need better
> than
> > 0.2% accuracy and what type of resistor your measuring.
> >
> > There is such as thing as "close enough" and I find 5 or even 10%
> "close
> > enough" in most cases.
> >
> > Tim Hills
> > Sioux Falls, SD
> >
> >
7143 2012-01-03 14:38:10 KK7B Re: Ohm Meter Calibration
Hi Bill,

Happy New Year. I think we established that his original question is under constrained. We provided some more input on additional measurements to make--then went off on one of our usual tangents in this group. A good theoretical question can often kick off a useful practical discussion.

His theoretical questi
7144 2012-01-03 14:50:07 William Carver Re: Ohm Meter Calibration
Understanding a measurement and its limits is crucial and the question
was a very good one. The kind of thing students should understand before
they tackle their first project or lab. I guess the tangential nature of
some of the replies bothered me Rick and it shouldn't have, hi.

I have reworked the rock planter (smaller), added a rock wall on one
side of the driveway and flattened the area where you turned the fifth
wheel around. It should be a piece of cake next time. So any time you
are passing through the area come visit and park for the night if you
like.

73 - Bill
7145 2012-01-03 19:24:17 Tim Hills Re: Ohm Meter Calibration
I wasn't trying to play down the original question. In my pre-coffee
haze I missed that it was a theoretical question rather than one based
in a current design/construction problem.

I think others have pretty much covered the theoretical and practical
aspects so I'll just head for bed and try again tomorrow ;)

Tim Hills
Sioux Falls, SD

On 1/3/2012 13:58, KK7B wrote:
>
>
>
> One should always be aware of not just the component value, but the
> tolerance needed for any given application. 5 and 10% tolerance is
> good enough for some things. Any time you want balance, or have a
> circuit that is sensitive to differences between channels, you will
> want better. For a graphical look at when you might need 0.1%
> accuracy, see figures 9.58 through 9.61 in EMRFD. The conclusion from
> that section is that 0.2% tolerance is good enough for 3rd order all
> pass networks used in IQ systems, but many authors have chosen 4th
> order systems to cover the speech bandwidth. For those, there is some
> benefit to matching resistors and capacitors to better than 0.1%
>
> Modern resistors and capacitors at ambient temperature will hold those
> tolerances over time, which makes analog phase shift networks
> attractive for systems that can be measured and built once, and still
> meet specs 2 decades later.
>
> Enjoy the experiments,
>
> Rick KK7B
>
> >
> > Peter:
> >
> > I was just wondering what your working on that would need better than
> > 0.2% accuracy and what type of resistor your measuring.
> >
> > There is such as thing as "close enough" and I find 5 or even 10%
> "close
> > enough" in most cases.
> >
> > Tim Hills
> > Sioux Falls, SD
> >
> >
7146 2012-01-03 19:30:09 William Carver Re: Ohm Meter Calibration
I apologize for coming on so strong, Tim. Guess I replied in an internet
haze which is far worst than pre-coffee.

Bill W7AAZ
7150 2012-01-04 09:50:20 Glen Re: Ohm Meter Calibration
The response to Peter's question is likely "multiply by 1.002".
But I'd like to point out situations where "add 2 to all readings"
is the appropriate solution...

I have a very accurate DC voltmeter, that has resolution down to
microvolts. I measure a voltage with one polarity, then switch the
leads and measure again - the absolute value is not the same. The
cause of the difference is thermocouple voltages caused by me holding
one voltage probe in my hand. This causes an offset error, similar to
Peter's example where the ohmmeter doesn't read "0.00" when the leads
are shorted. For both these cases, "add 2 to all readings" is the
right response.
BTW:
The response "multiply by 1.002" assumes that the ohmmeter is a
perfectly linear device. Any non-linearity makes this answer suspect.
Many manufacturers don't specify the limits of their accuracy specs.
A meter having 1% accuracy might suffer most from offset, from
non-linearity or from scaling errors. We often assume that the
scaling error dominates and that "multiply by 1.002" is the correct
solution. A number of measurements can reveal what proportion each of
these errors contributes to the overall "1%" error spec:
"Does it read zero when shorted?" reveals offset errors.
"Does it read 1000 ohms with a know 1K resistor?" reveals scaling errors
A number of different-valued calibrati