Power Profiler for IoT

Hi There,

As probably many of you I am involved in designing embedded/IoT gadget. And one of the most important thing is estimate battery life…
So far I have managed it measuring the dc sleep state current with a DMM and profiling the current transients (BLE events, micros event routines, etc…) hacking a Dialog Semi evaluation ki. With the aid of the companion GUI I was able to sample the current consumption at 100kHz (if I remember correctly) and get it nicely visualize on the PC along with battery drain estimation (basically how many uColumb I am draining from the battery every time an event is running on the IoT thing).

I was curios to know if you have a better instrument to perform this kind of job…
Has anyone tested the ZS-2102-A ?
I have seen Nordic has a similar add on to plug on their eval board, I guess an option would be to hack that as well…

Cheers!

1 Like

Hi @andrea.longobardi85!

You want a Joulescope! I designed Joulescope to solve precisely this problem. Joulescope is the most affordable and easy-to-use precision DC energy analyzer. Joulescope measures current and voltage, then computes power and energy. It’s an electrically-isolated, USB-connected instrument that samples at 2 MSPS with open-source software for your computer.

As a consultant, I ran into the problem of cost-effectively measuring large dynamic current ranges far too many times. Your description of hacking various equipment and setups matches my experience (Lee Teschler wrote a nice summary of approaches here). I finally decided to do something about it, and Joulescope is the result. It’s the first product that I created for my own company.

Many of the people in this forum were early supporters and Kickstarter backers. I think the best thread on this forum is my Intro page. We continue to sell Joulescopes through our online store, and you can now buy Joulescope’s with ISO 17025 NIST-traceable calibration & recalibration, if your work needs that.

EEMBC just announced that Joulescope will be an approved instrument for their upcoming Wi-Fi benchmark. Since you mentioned Dialog, they have confirmed that Joulescope is just as accurate (and easier to use!) as the Keysight 6705, but at 1/10 the price.

If you have any questions, feel free to ask here, DM me, or post over on the Joulescope forum.

To make your decision easier, use discount code CE_LOVE for a 10% discount off everything in the Joulescope store through the end of the month.

3 Likes

My client and I both got a Joulescope independently of each other – and we both found it to be quite useful for seeing wide-swinging power demands. He uses it to monitor power demands of a cellular modem that he is bundling into his product – so he is particularly interested in the uA range performance of his product while being able to see the A range draw when the modem kicks in.

I like it because it’s an easy setup for an isolated current measurement into my PC.

2 Likes

We use the Otii Arc for exactly that. The battery simulation is a £440 add on per year, but even the base software is enough to do power and event profiling. Great as long as your supply voltage is 5V or lower.

Haven’t tried a Joulescope yet though…

Used Nordic power profiler in the past and recently got a Joulescope too. Seems to work pretty well. One thing still trying to determine is if it can be used for long term validation of IOT widgets. Ie can I leave it for a month to see if power profile goes weird for some reason.

1 Like

Joulescope is life-changing! Took me a little while to get used to its autoranging and to understand fully what it was showing me, but now I would never want to be without it.

1 Like

For now, we provide a downsample_logging.py python script to perform very long-term logging. The script is specifically intended for automatic recovery in the event of any errors including host computer reboots and power loss. We have at least one customer that has been logging Joulescope data for months.

Our next UI software update will include downsampling. The raw 2 Msps from Joulescope is about 8.1 MB/s, which really adds up. With downsampling, you can select the data rate and only save what you need. This will allow you to use the UI for much longer captures without filling up your SSD/HDD.

1 Like

@andrea.longobardi85 I love my Joulescope. I haven’t used the long term downsampling mode yet but I have used it to troubleshoot current spikes intermingled with sleep modes etc, which I imagine would be useful for your application. A great bit of kit and as the FW is FOSS, you can hack it if you want (although I observe that feature suggestions to @mliberty are a very efficient way to get what you need implemented).

3 Likes

Alright, I think I have found the what I need ! :smiley:
@mliberty thanks! Just placed an order!

2 Likes

Awesome! Your Joulescope will ship tomorrow (Sat). If you have any questions or issues that you can discuss publically, the Joulescope forum is the best place. I monitor it closely. You can also DM me or use the Joulescope contact form.

1 Like

That’s killer, congrats @mliberty What has that meant for your business? Hopefully orders are through the roof?

2 Likes

Thanks @ChrisGammell ! Having Joulescope adopted by EEMBC has resulted in a small increase in sales, but they are still working on the benchmark. I would expect much more of an impact once the benchmark is ratified.

I am also still working on the downsampling feature, which they desparately want. Joulescope’s wider bandwidth make it look noisier than the Keysight 6705, but it’s about the same within the same bandwidth. Once that’s done, they have no reason not to use Joulescope :wink:

Overall, sales are going well. We have a lot of repeat, happy customers, and I hope to help many more engineers this year!

Thanks for asking, and thank you for your help in promoting Joulescope!

@mliberty I have been deeply involved with my ctxLink product for a while, however, I couldn’t let another opportunity to endorse Joulescope, I LOVE it.

1 Like

I just had a good long chat last week with Jerry Mendez (our local sales guy) and Tim Paasch-Colberg who is a product manager at Rohde & Schwarz.

He was showing me a rather cool T&M thing that they are currently developing for exactly this application. Basically the fact that their battery powered RTH scopes are completely floating and isolated but share the same super nice front end as their higher end scopes means you can get a really nice quiet signal down in the very very small signal range . Practically this enables using very small current shunts and still having the ability to measure both low power sleep currents as well as higher wake currents without the current shunt voltage browning out the device . They showed a slides of prototype that was very very impressive in this respect with regards to both it’s dynamic range and it’s bandwidth (hundreds of megahertz) . I’ll bug them and see if there’s an app note I can share :slight_smile: it was pretty impressive and we are working on getting one in house to test

Sounds like a competitor for the Keysight CX3300 family, arguably the leader in this space. (If you have to ask how much it costs, you can’t afford it.)

But the real question is how much bandwidth do you need? If you are measuring current directly into an IC without any bypass capacitance, you can make use of this large measurement bandwidth.

If you are measuring the current to a target circuit board, not so much. You have the impedance from your power supply, cables/wires, and any connectors. If you do a very good job with very short wires and super low-impedance supply, this could be 2.5mΩ in each direction, + and -. With even just 10 µF bypass capacitance on your board, your target board bandwidth is:

1 / (2 * π * 5e-3 Ω * 10e-6 µF) = 3 MHz.  

Most typical lab setups with banana jacks and 3’ of wires will be closer to 50 mΩ. Banana jacks have 1 mΩ connector resistance. 18 AWG is 6.3 mΩ/ft.

4 connectors + 6 ' of 18 AWG = 4 * 1 mΩ + 6.3 mΩ/ft * 6 ft = 42 mΩ

With this setup, you only get 379 kHz of system bandwidth. Coincidentally, this is how I determined the design objective for Joulescope’s target bandwith. Joulescope uses a 10 mΩ shunt in its highest current range setting. Other impedances (Joulescope’s MOSFET, PCB traces, the internal front-panel connector and the external connectors) add another ~15 mΩ. So, with 2 sets of 1’ banana wires (Joulescope sits between your power supply and target board), the system input resistance is:

4 connectors * 1 mΩ + 4' * 6.3 mΩ + 25 mΩ = 54.2 mΩ

This setup with 10 µF target bypass capacitance gives 294 kHz which is slightly higher than Joulescope’s 250 kHz bandwidth specification, but that spec is a conservative number.

Note that I am intentionally leaving out the supply impedance. A small LiPo battery can easily add 100s of mΩ.

So, if you want to measure current to a target board, save your $$$ and buy a Joulescope :wink:. Joulescope’s sensor inputs are electrically isolated, up autorange in 1 µs (no target brown-out), and measure 32-bits of effective range.

1 Like

Hi Matt, I finally got some spare money and ordered a Joulescope too! Hopefully the customs here won’t ruin my day with it, like they did with a LiDAR module a couple of days ago… :roll_eyes:

2 Likes

Awesome and thank you! If your customs office requests any additional information, DM me! Hope your Joulescope travels quickly and safely! :crossed_fingers:

1 Like

It did made it to Italy!!!

2 Likes

Excellent! I hope your Joulescope serves you well for many years. If you have any questions or issues, DM me, or post over on the Joulescope forum.

2 Likes