Consumption profile equipment

Hi Joao,

I think today if I had the choice, I'd go for a bench supply and a datalogger (see BK Precision or Hioki for instance), or a nice multimeter like DMM6500, and log to it or disk.

Some of the all-in-one units intended for IoT use suggest they simulate a battery, but the only decent simulation of a LiPo is in Fortran, and so unless they are running that, I don't know how much value that is.

The instruments above would be more versatile (albeit more costly).

For low cost, a microcontroller and an ADC board could be used (could be an off-the-shelf ADC eval board).

It depends a lot on exactly what you want to measure, most power supplies and even data loggers struggle with the dynamic range of the current drawn by IOT like devices,

1A maximum is not uncommon but the sleep current might be down in the uA or less region.

Very few systems can measure and log this if you want to catch fast spikes of high current  - the Keysight boxes can but you are looking at £5k+.

As Shabaz points out, if you want to simulate a battery then that's a whole extra can of worms.

I looked at the Qoitech and I did not like what I saw, the claimed operating temperature range is 15C - 25C - this looks like a company that doesn't know what it's doing.

The detail in their specs is just a joke comapred with that from Keysight, R & S, Keithley etc. And they seem to want loads of extra money for software licenses.

Like Shabaz, I've always ended up using multiple instruments, bench psu with computer control, Keithley DMM6500 (nice programmable DMM with fast logging if needed), or maybe use Kesysight DAQ.

Can you tell us a bit more about the typical current profiles of things you want to measure ?

MK

The JouleScope should tick all boxes - 15V, 3A, 250kHz bandwidth. Its not exactly cheap, but for what it does its also not too expensive. Look for more current reviews - from what I remember the software had some quirks initially, but they might be resolved now.

It looks quite good, and actually claims to switch the current sense resistors really fast - which must take a lot of effort since the resistance is so low on the 3A range.

It's cheaper than the Quotech device and much faster.

Has any one (reading this) actually used one ?

MK

Hi Michael,

In my current work, we have basically all boards being powered by solar panels, batteries ( Lithium, Lead-acid battery), or supercap. Those boards are designed to:

• Wake up time to time read many different sensors applied in agriculture: temperature, ultrasonic wind sensor, soil moisture, salinity... ; (Normally currents are way below 100mA)
• and via LoRa/NBIOT/CAT/2G/3G Modems report the sensors data to a server; ( when transmitting data can range from 40mA in LORA to 300mA on the modems )
• also, connect and power other boards  ( sensor node made by us too );
• To have switches to cut energy from the peripherals which even sleeping can consume some mA. (some boards in sleep mode go for 9uA-12uA)

Therefore in order to have a better estimation of battery life and the power consumption of our boards and also the sensors attached to them I am looking for such a tool.

...if you want to catch fast spikes of high current...

I think I don't have issues with fast spikes in these applications, at least up to now.

Like Shabaz, I've always ended up using multiple instruments, bench psu with computer control, Keithley DMM6500 (nice programmable DMM with fast logging if needed), or maybe use Kesysight DAQ.

I would go for it, but I'm not sure if the current budget can handle it. I will investigate this option; in the meantime, I'm looking carefully for the option pointed by Hendrik Lipka

I hope I could clarify a bit more,

Thanks

Thanks, Hendrik

I am checking your recommendation and I will look for the recent reviews. It's already cheaper than the Qoitech

Hi Shabaz,

Some of the all-in-one units intended for IoT use suggest they simulate a battery, but the only decent simulation of a LiPo is in Fortran

Thanks for this info, I didn't know that.

The instruments above would be more versatile (albeit more costly).

As I was answering Michael below, I would go for it. I will search for the options and maybe I have enough material to convince my bosses.

Could you check the answer I gave Michael below? I gave more details on the application and maybe it lights up some more ideas on you, but just if it does not bother you. Thank you!

Hi Joao,

Your comments make sense. Also, incidentally, if there's a difference from (say) a few hundred nA between the lowest current sleep mode and the next level of current draw (say 1mA), then it can be possible to can spot that, at a useful bandwidth, and still be able to handle the highest expected current, with a single sense resistor (i.e. no switching) and a very low-cost INAxxx board (from TI), if the sense resistor is sized well per device-under-test, because a lot of IoT DUTs won't consume more than a few hundred milliamps of current (and your mentioned values also support this), and some will be just in tens of milliamps if they use low power radios. However if there are several sleep ranges close to each other then that's harder. Anyway, that could be part of an ultra-cheap option, sometimes needed perhaps. There's a short discussion here: Using the INA210-214 for Current Measurement for IoT Devices where it also mentions the possibility of moving the sizing problem to plugging in/out different gain INAxxx chips, instead of swapping out sense resistors, for different desired ranges.

Here's a thought - what we want is to measure over a huge dynamic range but most of the time accuracy is not that important - 1% would certainly be adequate.

If you were to measure the voltage drop across a standard diode (or a transistor wired as a diode) it is accurately logarithmic with respect to current over many decades.

And if you either control or measure its temperature it should be possible to remove the temperature dependence.

Then you end up with a single range that can cover maybe six decades of current from 100mA to 100nA.

I've messed about with the idea, and even bought some suitable transistors (you need low base and emitter resistance).

Something like FZT849FZT849.

MK

Hi Michael,

A log method would be really neat, since accuracy isn't needed for most of these use-cases. A problem with the few existing all-in-one IoT tester devices, is that they still cost a lot, and sometimes several would be needed simultaneously, so then it's a huge cost for verifying transmit/idle/sleep behaviour that didn't need the accuracy in the first place. Battery life estimation isn't going to be much more accurate with those devices either.