Test equipment for true integrated average current down at the uA level.

[Psi]

I have the need for a power meter, or other piece of test equipment, that can give me a true average battery current over 10 minutes on an IOT device running on coin cell.  The average current is around 4uA with peaks up to 40mA.

Using a good desktop meter is problematic due to the meter needing to perform a long auto-zero sample period and missing current spikes. Turning off the auto-zero results in too much drift to be useful.

What kind of test equipment should I be looking at for getting an integrated average current with an accuracy around +/- 0.2uA ?

It really needs to be something off-the-shelf so it can be certified/calibrated.

$3k is probably within budget but cheaper is better.

Any ideas?

[tggzzz]

You don't mention the spike duration, which would determine the maximum measurement interval.

In the absence of that, you only need to take one measurement every 10mins, at the appropriate time.

So, build an analogue integrator, reset it at the appropriate time and measure the output 10mins later.

[nctnico]

A good bench DMM with statistics should do the job. Since you are only interested in the average you can use a (low ESR) capacitor at the side of the DUT to filter out large spikes.

[bdunham7]

Using a good desktop meter is problematic due to the meter needing to perform a long auto-zero sample period and missing current spikes. Turning off the auto-zero results in too much drift to be useful.

Not all DMMs rely on auto-zero like that.  I assume you also can't rely on auto-range for the same reason?  How much drift are you seeing and what meters have you tried?  And how narrow are these spikes? A warmed-up and stabilized Fluke 8846A, for example, will have little to no zero drift in that 10 minute period, but still has 3-5 counts of noise at 10 NPLC (and much more at faster speeds) and that exceeds what you want for accuracy.  And even that greatly exceeds its specified performance, so if you need the instrument to guarantee the accuracy you specify, I'm not sure what to say. 

What kind of test equipment should I be looking at for getting an integrated average current with an accuracy around +/- 0.2uA ?

Your desired accuracy is 5ppm of the maximum current.  Assuming you will be using a 100mA range, it is 2ppm of the range.  Those are tight specs.  The TOTL Fluke 8588A has a specified uncertainty in the 100mA range (99%, 1 year, +/-5C) of 74ppm/reading and 13ppm/range.  And that is assuming a fairly long aperture time and slow reading rate so if your spikes are less than a large fraction of a second, some may get lost in the dead time between apertures.  If you speed up the acquisitions, your accuracy gets worse.

I'm not familiar enough with high-end SMUs to tell you if they can do what you need, but it would have to be something that can range quickly enough for your needs or has some other way of concurrently measuring in two ranges.  I can't see a single-range instrument quite getting there. 
[/quote]

[Psi]

You don't mention the spike duration, which would determine the maximum measurement interval.

If it could pickup 1ms spikes that would be fine.

what meters have you tried?

Only tried on Keysight 34461A


hm.. solution is not as easy as I first thought.

[Hydron]

Possibly a slightly crazy idea, but maybe use 2 power supplies, one at a slightly higher voltage and in series with a ammeter on a low current range, the other at a slightly lower voltage and in series with an ammeter at a higher range, connected together after the ammeters.

At low currents the first supply will source all the current, but if the current consumption spikes then the burden voltage of the first ammeter will become equal to the difference in supply voltage, and the second supply will start contributing the majority of the current needed during the spike.

At the end, add average current values from both ammeters for an overall total.

Should work with anything that drops output voltage slightly at an appropriate current draw (including a SMU going into compliance).

Accuracy requirements (for drift etc) would be highest on the meter measuring the high range, but you could possibly do some post-processing to null out any offset occurring when the current is below the range switchover level.

[MegaVolt]

To be honest, it is not very clear why such an accuracy of 2 ppm.
With an oscilloscope, we measure the parameter of the pulses.
Separately, we accurately measure the current in both modes by specifically turning them on for a long time.
We calculate consumption with an acceptable accuracy of a couple of %.

[voltsandjolts]

Maybe just decouple the board input supply with big low-leakage capacitor, then measure current from psu to the capacitor/board.

[nctnico]

Maybe just decouple the board input supply with big low-leakage capacitor, then measure current from psu to the capacitor/board.

By definition all the current flows through the DMM. Depending on the construction of the DMM it should integrate narrow peaks no matter what so measuring the average current consumption should be very straightforward (but with the auto-ranging disabled). As a check you can use a function generator producing narrow peaks and see at what width/frequency where the DMM starts to lose precission.

[bdunham7]

hm.. solution is not as easy as I first thought.

You mentioned needing an off-the-shelf certified device.  Is that for a specific reason?  I can think of multiple ways to handle your problem, some of which have already been mentioned, but I can't think of any off-the-shelf solutions that are calibrated and have your specific requirement (let's call it DC crest factor) in the specs.

There are two ways to deal with signal like yours--analog integration and the digitizing approach.  With digitizing, the sample rate affects your accuracy as much as the sample accuracy.  So to measure the 1ms spike, you need to digitize at 100kSa/s to get 1% accuracy.  I think any digitizing system set up to catch your spikes is going to be totally inaccurate as to the residual 4uA current.

I realize you don't need to actually measure the average current to 5ppm.  What you need is a reasonably accurate analog integration system that will simply add the charge from the spikes into a longer period without such a large swing in the signal voltage presented to the ADC of whatever instrument you are using.  As mentioned, DMMs have that, but since there are typically gain and buffer stages before the integrator part, there is a limit to how much over full range you can go.  Suppose the meter can accurately integrate spikes of 4X full range--that still leaves you using the 10mA range of a DMM, which will struggle to measure your 4uA residual.  And on the 8846A, for example, it would be hopeless as the 10mA range is very noisy.

So assuming you don't find some off-the-shelf solution that I don't know about, the way to go would be analog integration outside the DMM.  This may be as simple as an external capacitor across the inputs as has already been mentioned, but I would recommend an external shunt.  I'd also consider getting a high-quality polypropylene film cap so that leakage and linearity are not in question. What you need to figure out is how many of those spikes there are and how long they are on average, so you can have some idea what average current will be.  Then you can select a current range or shunt and voltage range not too much higher than that and a capacitor large enough so that the maximum signal presented to the DMM doesn't go over range, or at least not too much over range.

The result can then be digitized over a 10 minute period if your DMM supports that.  You can also just measure the current with a long aperture, but there will still be dead time between apertures.  Whether that is an issue depends on how often the spikes occur and how variable they are in length.  Whether all of this passes muster as a 'certified' or 'calibrated' test is another issue.  Certainly the DMM and shunt can be calibrated and the precise value of the capacitor doesn't matter, but if someone is looking over your shoulder or demanding paperwork, you may have some convincing to do.

[I wanted a rude username]

The Nordic Power Profiler Kit II is an auto-ranging energy meter in the style of the Joulescope except an order of magnitude cheaper. It promises to do exactly what you want. But it was only released in December, and is sold out everywhere ...

[Berni]

One option is to run your device from a very large capacitor, like small value supercap.

You can charge it up to a known voltage, leave to stabilize and then use a multimeter in high impedance input mode to watch the voltage drop as current is drawn from it. You can calibrate your capacitor for any imperfections by repeating the measurement with a fixed resistive load or constant current source.

Just choose a capacitor value that will not discharge enough for the device to die, but still discharge enough for the voltage to be easily measurable.

[bdunham7]

The Nordic Power Profiler Kit II is an auto-ranging energy meter in the style of the Joulescope except an order of magnitude cheaper. It promises to do exactly what you want. But it was only released in December, and is sold out everywhere ...

It looks to me like the Joulescope with the ISO 17025 calibration would be exactly what the OP needs and within budget...

https://www.joulescope.com/products/joulescope-precision-dc-energy-analyzer?variant=31385646235750

[Psi]

The Nordic Power Profiler Kit II is an auto-ranging energy meter in the style of the Joulescope except an order of magnitude cheaper. It promises to do exactly what you want. But it was only released in December, and is sold out everywhere ...

excellent I will investigate that more. Thanks

It looks to me like the Joulescope with the ISO 17025 calibration would be exactly what the OP needs and within budget...

https://www.joulescope.com/products/joulescope-precision-dc-energy-analyzer?variant=31385646235750

That does look really good. Thanks!!

[voltsandjolts]

Coin cell response to pulse loads:

[bdunham7]

That does look really good. Thanks!!

Thank I wanted a rude username!  I hadn't seen the Joulescope until he referred to it.  I've wondered if something like that would eventually become available and there it is.  The key is the ability to shift ranges in 1 microsecond.  If you get one let us know how well it works and how easy it is to set up.

[tautech]

I hadn't seen the Joulescope

They've been mentioned here a few times over the years and my aging memory cells couldn't bring back the name to recommend them to Psi. 
At one time we considered stocking them however thought they'd be a slow seller but these days lots more work is of micropower design it's likely a few units/year would sell.

[I wanted a rude username]

I assumed Psi already knew about the Joulescope ... but you know what they say about assumptions ...

The key is the ability to shift ranges

Bingo. Even a 24-bit ADC with a suitable sample rate (and they do exist) wouldn't cover the enormous current range we deal with (particularly when BLE or other radios are involved), from < 100 nA to > 100 mA ... the bottom end would be lost below the noise floor.

[nctnico]

One of the approaches I have been contemplating is to use a log amplifier to compress the dynamic range and feed it into an oscilloscope. By using a math function to reverse the log you can use the cursors to get the current reading. Or use math to get an average.

[I wanted a rude username]

There's got to be a catch with that approach, right? I mean an IC like the LOG114 has something like eight decades of range, but the datasheet suggests it only increases effective ADC range by 2/3 ... even if you use a 12-bit DSO that'll be effectively 20 bit range ...

[pigrew]

Another option is TI's EnergyTrace, which is supposed to be accurate down to 0.5uA.

It is a switching voltage supply that instruments its DC-DC converter to record the amount of energy dumped into the load. (It doesn't directly measure the current, AFAIK). The MSP-FET can output between 1.8 and 3.6V. Some of the TI cheaper MSP430 dev boards contain EnergyTrace if you'd like to try it out for <$20, but they are restricted to only output 3.3V.

[nctnico]

There's got to be a catch with that approach, right? I mean an IC like the LOG114 has something like eight decades of range, but the datasheet suggests it only increases effective ADC range by 2/3 ... even if you use a 12-bit DSO that'll be effectively 20 bit range ...

IIRC Analog has some more suitable devices but I have not looked at this project for a long time. Part of it is a power supply stage where the current through the feedback resistor is used to sense the current instead of using a shunt resistor (in order to keep the supply output impedance low).

[tom66]

We used a R&S NGM200 which can supply power and measure current down to 1uA.  The key advantage for this kit is it can sample up to 1MSa/s,  and up to 500kSa/s if the data is downloaded via a PC.  The interface is fairly simple, within a few hours I was able to write a Python script which took every single sample and integrated it to produce an accumulated mAh figure.  Ethernet worked better for this (USB seemed to choke above 250kSa/s.)

We used this to measure the battery performance of an IoT system, where the sleep current was ~1uA but peak consumption was >100mA.

We found the built-in battery simulator on the NGM200 to be useless.

[rsjsouza]

Another alternative could be TI's EnergyTrace, which is built into some Launchpads or a standalone combination of the XDS110 + the HDR accessory, which gives some 256kSPS from about 1uA to 800mA of range.

https://software-dl.ti.com/ccs/esd/documents/xdsdebugprobes/emu_energytrace.html

You may use the CCS IDE or a command line interface with CSV export.