µCurrent on Arduino very low power consumption (measuring 1 to 40µA, 1.5 to 2V)

[luxLumen]

Hello,
there are various posts on how to reduce Arduino's power consumption from ~40mA to 1-40µA (for volt see here). On very low µA consumption one builds a very lightweight "Arduio" made of a ATMEGA328P-PU but without all the other power consuming parts.

My question is, would the µCurrent make sense if measuring such low currents (1 to 40µA) or has the burden voltage no effect on such low currents (and voltages from 1.5 to 2V)?

[Simon]

The whole idea of daves microcurrent is to reduce the burden voltage but still have the resolution. Any multimter is primarily a voltmeter, every other measurement is derived and converted to volts. in order to measure current you put a resistor in series with the load, but this cuases a voltage drop so you use a low value resistor, but then the voltage can be too low to measure. So daves uCurrent basically uses a low value series resistor but amplifies the voltage so that you don't get a large voltage drop but retain accuracy and resolution of measurement

[luxLumen]

Sry, doesn't help me. I've already read and saw his kickstarter vid. So would it help in my measuring case?

[Simon]

probably, is it any better than your multimeter is the question and how accurate do you want to be, you ideally need to compare the series resistances to work it out. If you know the max burden voltage then divide it by the max current you can measure on that scale to get the series resistance in the meter.

[luxLumen]

is it any better than your multimeter is the question

Isn't it supposed to be better than any multimeter? ("A typical high end “accurate” multimeter will have a “low” 1mV/mA burden voltage (about as low as it gets), ...")

I have a peaktech 2010. It has 10MOhm. Hope this info helps.

[idpromnut]

is it any better than your multimeter is the question

Isn't it supposed to be better than any multimeter? ("A typical high end “accurate” multimeter will have a “low” 1mV/mA burden voltage (about as low as it gets), ...")

I have a peaktech 2010. It has 10MOhm. Hope this info helps.

Well it is, but like many things in life, it's not so simple. If you have a uCurrent, then go ahead and use it; it WILL be more accurate than the meter you have. But, the question I think is how much more accurate. For that, let's calculate the burden voltage of your meter with the worst case current you mentioned, 40 uA.

From Ohm's, we have V = IR, where V will be the burden voltage, I is the current that we're measuring and R is the resistance across your meter.

V = (0.00004) x (0.010) = 0.4 uV

So the burden voltage in your setup will be 4 uV. and it will probably not impact your measurement, unless you happen to be providing an input voltage that is less than 4 uV over what your Arduino requires to operate correctly.

EDIT: I can't math 

[Ziltoid]

Depends on your Equipent.

If you want to measure the current correctly you should compensate for the Voltage drop in the uCurrent oder Multimeter. See my "professional" Drawing.
R_uCurrent is (for the old one ) is
10 mOhm @ mA
10 Ohm    @ µA
1K Ohm    @ nA

40µA @ 3V is 75k Ohm

Error in nA range at DUT is 2,961V
µA 2,9996V
mA 0,99999986666668444444207407439012V

If you dont have a PSU with a Sense line place a Voltmeter between s+ and s- und tweak it until it displays 3V or whatever Voltage it needs.

Edith: u_CurrentArticle
 

[Rick Law]

Hello,
there are various posts on how to reduce Arduino's power consumption from ~40mA to 1-40µA (for volt see here). On very low µA consumption one builds a very lightweight "Arduio" made of a ATMEGA328P-PU but without all the other power consuming parts.

My question is, would the µCurrent make sense if measuring such low currents (1 to 40µA) or has the burden voltage no effect on such low currents (and voltages from 1.5 to 2V)?

If you have a high-accuracy DMM for current measurement but with high burden volt, I think using an adjustable power supply with two meters may be better than worrying about burden voltage.

First, measure the voltage supplied to the Arduino.  Leaving that meter connected to the Arduino, now connect the other current-measuring DMM between your Arduino (with volt meter) and the power supply.  Re-adjust the voltage at the Arduino to exactly as before adding the current meter.  The burden voltage is irrelevant now since you subtracted it off by increasing the PSU output voltage to compensate for the burden voltage - you subtracted it off.  With that, you can use a high-accurate meter regardless of its burden voltage.

[EEVblog]

If you have a high-accuracy DMM for current measurement but with high burden volt, I think using an adjustable power supply with two meters may be better than worrying about burden voltage.

Yes, and that is the traditional way of overcoming any issues with burden voltage.

[Simon]

But not much good if current consumption varies greatly as your supply voltage will be all over the place and possibly dangerously high if current draws drops

[EEVblog]

But not much good if current consumption varies greatly as your supply voltage will be all over the place and possibly dangerously high if current draws drops

Yes. I should have added, for reasonably steady currents.

[jay]

Measuring 40µA or less can be tricky when it's rapidly changing like in an MCU case. I do that often because i'm MCU SW guy specialized in low power applications related. The burden voltage is a real problem only if you can't increase your supply voltage to compensate. Many data logging multimeters sample at 10Hz or so. It's far less than you need for measuring an MCU. However the multimeters are often work so that you actually get roughly the right value but it's essentially low pass filtered. I learned the hard way that not all the multimeters work the same way. Also many multimeters are really inaccurate in the low end of µA range. Often it doesn't really matter because when optimising PM you don't need much absolute accuracy. µCurrent with decent oscilloscope is enough to give quite nice data that you can process further on a computer. When doing the most demanding measurements at work I use 25k euro equipment designed for the purpose but in daily development I use a Fluke multimeter and a Tektronix midrange oscilloscope measuring over a current sense resistor, which gives quite accurate data. The current probes from Tektronix have been a serious disappointment.. Except in an earlier job when measuring smartphones that have power consumption changing from 1 mA to 2 A rapidly.

[Simon]

Well you get what you pay for, multimters were not made for the job. An oscilloscope will make sense as in Daves recent video.

[Rick Law]

I was going with "Give the man a fish....teach the man to fish..."

The OP ends with the question "My question is, would the µCurrent make sense if measuring such low currents (1 to 40µA) or has the burden voltage no effect on such low currents (and voltages from 1.5 to 2V)?"

The best way to see if something has an affect is to measure it if feasible.

So, if the OP got reminded of using two DMM to zero out the burden voltage - which involves measuring the burden voltage (he knows how much he has to bring up the voltage to zeroing it out).  So he knows the impact and answered his own question.  The OP would also connect the dot and realize he can also zero out the burden voltage of the uCurrent exactly the same way.  Then, kill two birds with one stone.  He actually knows the impact, and knows a way to fix it if it is really important.

If the voltage is too fast changing to measure, then the OP can judge "is it important" to know the max? the min? or the average...

[Simon]

I think with this question you need to see the implications of what you are missing out on. As already mentioned the multimeter seems to be totally the wrong tool for the job and it will lead into a false sense of security. Seeing i believing in this case, show the results on an oscilloscope through a uCurrent and it will become clear where the multimeter has it's limitations.

[Rick Law]

I think with this question you need to see the implications of what you are missing out on. As already mentioned the multimeter seems to be totally the wrong tool for the job and it will lead into a false sense of security. Seeing i believing in this case, show the results on an oscilloscope through a uCurrent and it will become clear where the multimeter has it's limitations.

Please enlighten me on what I am missing.

To explain what I don't understand, I must restate my point:

The same way one can adjust or measure burden voltage is the same whether one is using uCurrent or DMM.  The OP can connect this two dots.

So, what I said without being too blunt was: use the dmm, dig around and try to measure the burden voltage of uCurrent.  If it is too low to measure, the OP got the answer "does it matter" which was the question posted in the OP.  In the process, the OP will figure out exactly what the best next step is.

I can't think of reasons to know exactly "this nano-second what the current draw is."  So I assume in all likelihood, the average, or the min/max, or the trend are a lot more important for the OP.  But he will have to get there himself since he knows what goal he is hunting for.  Digging around will help him connect his dots.

Is just an Arduino.  What risk is there at 1-2V 40uA?  Perfect opportunity to dig around and learn.

So, with my point restated more bluntly.  My question is: So, whats wrong with digging around a bit to answer one's own question? Exactly what am I missing proposing that?  (No sarcasm intended) I sure like to know if I am missing something big.

[luxLumen]

Thank you for your replies.

I must say idpromnut helped me with this calculation:

V = (0.00004) x (0.010) = 0.4 uV

So the burden voltage in your setup will be 4 uV.

I guess this error is acceptable for 1.5-2V.

However the multimeters are often work so that you actually get roughly the right value but it's essentially low pass filtered. I learned the hard way that not all the multimeters work the same way. Also many multimeters are really inaccurate in the low end of µA range. Often it doesn't really matter because when optimising PM you don't need much absolute accuracy. µCurrent with decent oscilloscope is enough to give quite nice data that you can process further on a computer. When doing the most demanding measurements at work I use 25k euro equipment designed for the purpose but in daily development I use a Fluke multimeter and a Tektronix midrange oscilloscope measuring over a current sense resistor, which gives quite accurate data. The current probes from Tektronix have been a serious disappointment.. Except in an earlier job when measuring smartphones that have power consumption changing from 1 mA to 2 A rapidly.

Seeing i believing in this case, show the results on an oscilloscope through a uCurrent and it will become clear where the multimeter has it's limitations.

I don't need precise oscilloscope data measurements. If I would be able to afford an oscilloscope, I would (and should as we saw in the kickstarter vid) be able to afford a µCurrent.
I don't know about Tektronix probes but if it's true, then they really should improve them (especially their caliber) because mA is is already way too much, at least for devices which should (be small and portable) run for months or years from a small battery.

to zero out the burden voltage

If it's true what idpromnut is saying (4 uV burden voltage) and I'm using 1,5-2V, I hardly can be that precise in zeroing out this small voltage.

I hope I understood everything correctly that you all were saying.

[Rick Law]

Thank you for your replies.
...

to zero out the burden voltage

If it's true what idpromnut is saying (4 uV burden voltage) and I'm using 1,5-2V, I hardly can be that precise in zeroing out this small voltage.

I hope I understood everything correctly that you all were saying.

LuxLumen, that actually was what I thought you may likely discover.  If it is so low you have problem measuring it (gee-- before and after is almost exactly the same), then you know the burden voltage is..gee--so small I can't see, so what is the point in worry about something so small...

Probably habit from having been a TA (teacher's assistant) in graduate school.  I have seen many bright students that some how misses a clue/connection.  I found the most useful way is not "hey this is the answer", but point out a path of exploration that will lead to the inevitable answer - particularly when that path also contains an explanation of why.

Then again, that means I too can miss a clue/connection or I simply made a mistake.  That is why I am interested in hearing what Simon think I missed.

[idpromnut]

It is hardly my saying any sort of truth; it is Ohm's Law   

But having run the experiment of using a uCurrent as the "probe" for a scope, I can tell you it is very interesting to be able to more closely see the correlation between what my firmware is doing versus current consumption. But, perhaps that is for a later time. In this forum, we obey the Law of Ohm!   

[Rick Law]

It is hardly my saying any sort of truth; it is Ohm's Law   

In this forum, we obey the Law of Ohm!   

I am exempted.  My degree is in Physics.  I am therefore the Law of Physics...

Just a minute, I have to stuff my ego back in the house.  It is too big so I left it on the lawn when I got home.

Rick Law

[David Hess]

...

The current probes from Tektronix have been a serious disappointment.. Except in an earlier job when measuring smartphones that have power consumption changing from 1 mA to 2 A rapidly. ...

...

I don't know about Tektronix probes but if it's true, then they really should improve them (especially their caliber) because mA is is already way too much, at least for devices which should (be small and portable) run for months or years from a small battery.

Typical inductive current probes for oscilloscopes are not intended to operate at such low current levels and their wide bandwidth limits resolution because of noise.

A Tektronix P6021 AC current probe has a maximum sensitivity of 2 mA/mV with its passive terminator and 1mA/50mV with a Type 134 current probe amplifier but this later configuration is noise limited to about 150 microamps unless its 30+ MHz bandwidth is reduced by filtering.  A Tektronix AC/DC P6302/AM503 current probe performs about the same.

Jim Williams pulled this off using an AC current probe to measure the 530 nanoamps of current through a 32.768 kHz crystal but was able to take advantage of the fact that he was only looking at a specific frequency to filter the output and reduce noise:

http://www.linear.com/docs/25519

For an AC or AC/DC current probe, you can always wind the wire carrying the current through the probe aperture multiple times to raise the sensitivity.  Wrap it 10 times and the sensitivity is increased by 10 times.

[David Hess]

But having run the experiment of using a uCurrent as the "probe" for a scope, I can tell you it is very interesting to be able to more closely see the correlation between what my firmware is doing versus current consumption. But, perhaps that is for a later time. In this forum, we obey the Law of Ohm!   

I have made these sorts of measurements in the past by using a transistor cascode in the ground return of the circuit being measured.  This creates a virtual ground and the transistor acts as a ground referred transimpedance amplifier which is convenient for oscilloscope measurement.  Sensitivity is set by a single resistor at the collector and a shunt capacitor can be used to set the measurement bandwidth and limit noise.  You need a negative test supply to supply the ground current for this to work.

The emitter resistance of the transistor is about 26 ohms per milliamp making the burden voltage non-linear with current which is convenient but if an operational amplifier is added, then that value may be divided by the open loop gain of the amplifier yielding milliohms to microohms.

Ohms law is more of a guideline than a rule.