Anthony Fremont wrote:
ehsjr wrote:
Anthony Fremont wrote:
ehsjr wrote:
sense to look for your meter to be accurate to 8 decimal
places for your .00000005 amp reading.
Now come on, the 8 decimal places is only assuming that the scale is
in an Amps range. The meter would be in the 500uA full scale range
where 50nA is only 2 decimal places.
Perhaps I did not make the point clearly.
When you are using your DMM and measuring something in
the neighborhood of 8 decimal places, like tens of nA,
The meter doesn't change accuracy based upon the scale it's using, it only
changes resolution. It remains .03% accurate. Whether reading Amps,
milliamps, or microamps.
Take a look at the specs. They most certainly
do change, depending on the scale you are using.
Read the Fluke app note Understanding DMM Specifications.
Noise becomes a significant factor at the low end of a
range within the meter, and in general when measuring very
small voltage or current. And the specs, regardless of
noise, vary from range to range.
http://assets.fluke.com/datasheets/2153ExtSpecs.pdf
http://us.fluke.com/usen/support/app...lukeProducts)#
your meter, regardless of scale, will be less accurate
than when it is measuring something in the 2 decimal
place neighborhood. The meter itself is more susceptable
I guess I wasn't clear or we're not understanding each other. The meter
will be in a ranger where 500uA is the full scale reading. 10's of nA is
two decimal places.
Ok, I see what you are saying. To get accuracy on
that scale, you would need at least 6 digits displayed,
and that's before you consider any error in the
circuitry. But I now undertsnd what you have in
mind based on what you said at the bottom of your
note, where a reading of anything from .03 ua
to .07 ua will meet your needs for your .05 ua
current. That's not the accuracy I thought you
were talking about. A +/- 20 nA variation on a
50 nA measurement is an error of 40 percent - which
I call innacurate.
But now that I understand what you have in mind,
I see your point. The way I was thinking about
it was too stringent for the example you posted,
so your example does prove the case of a kind
of measurement that fits into the under tens of
mA that I was talking about. Now that I understand
what you are saying, I think the confusion was at my end.
to uncertainty the lower you go. AFAIK, the current
shunt even for low current scales has a much lower
resistance than the 2meg or 100 k I mentioned. That
means that the meter has to work with a lower level
than the 110 mv those resistors produce.
Regarding scaling - DMM's have tens of mV in 2 decimal
places. Most DMM's do not have tens of nA in 2 decimal
places. To get an 8th decimal point current reading into
the 2 decimal point range, convert it to mV with a resistor.
To put it in another perspective, consider a Fluke 187.
It will give .01 uA resolution (2 digits after the decimal)
on the 500 ua scale at a claimed accuracy of +/- .25%.
We'll ignore the further 20 count uncertaincy. That's
a +/- 1.25 uA error. That measurement is useless for the
55 nA current measurement you need. The meter could show
500.00 or 500.05 or 501.25 or whatever and you would not
know whether you had 55 nA or not. On that scale, the
Your error calculation is assuming a full scale reading. The error
(neglecting the count uncertainty) at 50nA is only .125nA, it wouldn't even
show on the display.
But at 50nA it would read .03 to .07uA on my meter including the 2d
uncertainty, plenty good enough for me.
That statement clears it up for me, as I mentioned above.
To me it's a 40% error, but for what you are doing it
is accurate.
meeter cannot be accurate to 2 decimal places. And you
cannot throw away the third digit after the decimal - it
doesn't exist on the meter, the resolution is too poor.
The same meter, on the 3 volt (3000mV) scale is accurate
to within +/- .025% which is +/- 75 uV - again, ignoring
the further 5 count uncertainty. On the 3 volt scale
with the technique I mentioned where you throw away the
third digit after the decimal, the error is meaningless.
That digit happens to be accurate on this meter and scale,
so the error is meaningless, even if you keep it.
Try looking at the Extech I just ordered. .1%+2d 50000 count.
I'd like to - if you have a handy url, please post it.
If not handy, don't go digging for it. All ths talk
has piqued my interest in buying yet another DMM
(that I don't need - too many DMM's not enough time)
or at least drooling over the specs.
Is there an antidote for "test equipment lust"?
Here's how you do it with accuracy at the tens of _mV_ digit:
For 11 uA, put a 10K .01% resistor in series with
the supply and measure .11 volts across it. The voltage
would range from 0.109989 to 0.110011. Keep only
2 decimal places. Your computed current, worst case,
would be off by 1 uA
For 50 nA, use a 2 meg 1% resistor and measure .10
volts across it. The voltage would range from .099
to .101 taking the 1% into account. Throw out the
last digit. Your current computation would be off
worst case, by 5 nA.
Those are fine ways to measuring static current levels, but they
will not work for me. Until the PIC goes to sleep, the current draw
is much higher. So much so that it would never power up thru a 2M
resistor.
So I guess you're stuck with a need that the fancy Fluke
mentioned above cannot meet. How _do_ you measure the
55 nA?
That's why I didn't buy the Fluke. The meter I bought will give me 10nA
resolution. I know it won't be dead on when reading 50nA, but it will be
close enough that I know that I didn't leave some pull-ups turned on or some
other peripheral pidling away the juice. In current mode the Extech will be
good enough for me to be sure of what's happening. Any worse accuracy, and
I couldn't be sure.
What I would do is bypass the resistor with a switch so
the PIC can power up and run, and monitor it while it
is active by whatever technique you choose, so that you
know it is active. When it goes inactive, open the switch
to measure the voltage across the resistor.
Yes, I have done time-wasting methods like this before, that's why I want a
new meter, DSO and a logic analyzer.
:-)
With a voltmeter accurate to 2 decimal places.
I don't know why you would
If your volt meter has a 1V maximum at full scale and one can live
with 10% error, then I agree. If it has a 100V range, then you need
.01% accuracy on your equipment to make your measurements, right?
Anyone who is not smart enough to turn his meter
range down from the 100V scale to measure mV
is not smart enough to need nA measurements.
Measuring mV with the range set to 100 is stupid.
And 10% error for a DMM is stupid. I know you are
*not* stupid. So what is your point?
The 10% error is due to your technique not the DMM, you said so yourself,
and I quote:
" For 11 uA, put a 10K .01% resistor in series with
the supply and measure .11 volts across it. The voltage
would range from 0.109989 to 0.110011. Keep only
2 decimal places. Your computed current, worst case,
would be off by 1 uA
For 50 nA, use a 2 meg 1% resistor and measure .10
volts across it. The voltage would range from .099
to .101 taking the 1% into account. Throw out the
last digit. Your current computation would be off
worst case, by 5 nA."
By my calculations, a 5nA error on a 50nA reading is a 10% error or did I
miss something?
No, I did. I thought you were talking about meter accuracy
when you said 10% - you were talking about measurement
error.
I agree that these techniques are valid and worthwhile at times, but I will
stick with the convenience and accuracy of a $200 meter instead of buying $5
resistors. :-) I've got a tracking number and it should be here tomorrow,
I can't wait. I reall can't wait til my scope gets here. :-)))))))
Go ahead - make me drool! Enjoy the meter. :-)
And the scope. :-)
Ed