Equipment advice: Battery lifetime testing

[tszaboo]

Do you know equipment, that measures the power usage of some DUT? I know this is typically very simple, however I have some requirements.
The sleep current is around 20uA, with occasional RF transmission, up to 2A pulses should be measured.
If it simulates battery behaviour, with internal resistance it is better.
The mAh or mAs is important.

We used a bench multimeters to do this kind of testing, but the overhead for calculating the actual usage is too long. We want to do this testing each time a new firmware is released, so the less time we have to spend on it the better.
And the drop on internal shunts is not really acceptable. And there is also the problem of slow sample rate.

One candidate I'm looking at is the Keithley 2281S-20-6 Precision DC Supply and Battery Simulator. We have it's small brother, the Keithley 2280S. I think the difference between them is mainly software, maybe it is the extra battery discharge unit? Doesnt matter that much, it is a nice equipment in theory, slightly inconvenient to use in practice. To be honest, all the recent Keithley equipment was a slight disappointment, and I'm somewhat reluctant to spend more money at them.
At this point I can convince management to spend a few thousand euros on this equipment (that is a few+1 thousand USD).

Do you guys know anything else that can be used for such purposes? I never had an SMU, is that something that can do these tests? Or probably an overkill.

[Kean]

Qoitech Otti
https://www.qoitech.com/

A couple of forum discussions here about it - the big negative for most is the annual licensing model for the software
https://www.eevblog.com/forum/testgear/qoitech-otii/
https://www.eevblog.com/forum/testgear/battery-energy-consumption-how-to-measure-it/

[genghisnico13]

This previous thread has a couple of options:
https://www.eevblog.com/forum/testgear/battery-energy-consumption-how-to-measure-it/

[tszaboo]

Qoitech Otti
https://www.qoitech.com/

A couple of forum discussions here about it - the big negative for most is the annual licensing model for the software
https://www.eevblog.com/forum/testgear/qoitech-otii/
https://www.eevblog.com/forum/testgear/battery-energy-consumption-how-to-measure-it/

Otti: Looks interesting, though the sample rate is not great. I think I visit them for a demo next week in Nürnberg.
Also, to cover all the equipment we sell, it would need 15V output.

Lots of good info in the thread, the Rhode equipment is also interesting, though probably obsolete. I will pay them a visit also anyway.

[Kean]

Great, please come back here and let us know what you discover - good and bad.
I'd also like more than 5V capability to cover my various projects (in particular 2S LiFePO4 and maybe 4S LiIon).
I've been considering the ITech IT6411 which is similar to the Keithley 2281S.

[tszaboo]

Great, please come back here and let us know what you discover - good and bad.
I'd also like more than 5V capability to cover my various projects (in particular 2S LiFePO4 and maybe 4S LiIon).
I've been considering the ITech IT6411 which is similar to the Keithley 2281S.

Those Itech units seem to be OK on first sight, but if you go into details, it falls apart.
It has two ranges, on the high range (5A) the resolution of current is 1mA, that is 13-14 bits. Not very promising. And it doesnt say anything about range switching. This is from the user guide:
Sample Rate Sample rate: 30KHZ at maximum
Sample Count: Sample count: 600 points at maximum
So what happens after the 600 samples? Anyway, the soft specs seems to be quite lackluster.

That being said, I was designing similar equipment in the past, it is not easy to do high accuracy high and sample rate at the same time.

[jeremy]

Best one I’ve found is Keysight N6781A. 20V@1A (or 6V@3A), 200kHz, unlimited sample streaming. Only downside is the price but I have one and it does the job.

I have an older Keithley as well, but it’s very limited in sample rate.

[mliberty]

I know the perfect test instrument for measuring energy consumption, Joulescope, and I just posted about it.  (disclosure: I am the creator of Joulescope).


Joulescope does not do battery simulation (it's not a power supply), but you can use it with a real battery or battery simulator.  I created Joulescope because I faced this same problem!  Existing gear is too expensive or not good enough (usually too low bandwidth and too high burden voltage).



What do you think?

[tszaboo]

I know the perfect test instrument for measuring energy consumption, Joulescope, and I just posted about it.  (disclosure: I am the creator of Joulescope).


Joulescope does not do battery simulation (it's not a power supply), but you can use it with a real battery or battery simulator.  I created Joulescope because I faced this same problem!  Existing gear is too expensive or not good enough (usually too low bandwidth and too high burden voltage).



What do you think?

It is a smart device. Looks to be perfectly tailor made for such projects.
Are you planning on operating a webstore after the kickstarter? I might purchase one or two, because this would suit our needs perfectly.

[mliberty]

I am planning on having preorders and then a full webstore after general availability.  The Kickstarter is a great discount off the retail price, so it's worth considering!  If you want to wait, I recommend that you visit the Joulescope website and sign up for the newsletter.  As details become available, I'll send ordering information through the newsletter.

[BitWrangler1001]

I am planning on having preorders and then a full webstore after general availability.  The Kickstarter is a great discount off the retail price, so it's worth considering!  If you want to wait, I recommend that you visit the Joulescope website and sign up for the newsletter.  As details become available, I'll send ordering information through the newsletter.

Does the joulescope do real-time 2Msample/s to RAM or HDD?

Thanks

[mliberty]

Does the joulescope do real-time 2Msample/s to RAM or HDD?  Thanks

Yes!    Joulescope captures to a 30 second RAM buffer by default.  In the user interface, you can press the "record" button to simultaneously record to HDD/SSD.  The recording size is only limited by your storage.  Full rate captures are about 8.5 MB/s, which gets big pretty quick.  We will add a downsampling feature (for when you don't need full rate data) before July to enable longer captures that take less storage.

[tszaboo]

I am planning on having preorders and then a full webstore after general availability.  The Kickstarter is a great discount off the retail price, so it's worth considering!  If you want to wait, I recommend that you visit the Joulescope website and sign up for the newsletter.  As details become available, I'll send ordering information through the newsletter.

Thanks. I wish you success, but I hope you understand that I am getting this for a company.
So invoice, delivery and payment terms, all the boring things are necessary.

[BitWrangler1001]

Does the joulescope do real-time 2Msample/s to RAM or HDD?  Thanks

Yes!    Joulescope captures to a 30 second RAM buffer by default.  In the user interface, you can press the "record" button to simultaneously record to HDD/SSD.  The recording size is only limited by your storage.  Full rate captures are about 8.5 MB/s, which gets big pretty quick.  We will add a downsampling feature (for when you don't need full rate data) before July to enable longer captures that take less storage.

That's great, also, is there some interface so I can stream data into a python program?  Or will the program have to read the data from a file instead?

Thanks

[mliberty]

That's great, also, is there some interface so I can stream data into a python program?  Or will the program have to read the data from a file instead?

Thanks

Joulescope is made to be integrated into your own python program!  The Joulescope "driver" in python is open source on GitHub, and you can install it using pypi:

Code: [Select]

pip install joulescope
Look at this example script which finds a Joulescope, opens it, and records to a file.  If you want to deal with streaming data, like the UI does, you can do that, too.  You have a few options, but the easiest is to periodically poll the stream buffer for new data.

[BitWrangler1001]

That's great, also, is there some interface so I can stream data into a python program?  Or will the program have to read the data from a file instead?

Thanks

Joulescope is made to be integrated into your own python program!  The Joulescope "driver" in python is open source on GitHub, and you can install it using pypi:

Code: [Select]

pip install joulescope
Look at this example script which finds a Joulescope, opens it, and records to a file.  If you want to deal with streaming data, like the UI does, you can do that, too.  You have a few options, but the easiest is to periodically poll the stream buffer for new data.

Wow!  1us range change time, seems to good to be true, makes me suspicious  I look forward to trying it

[mliberty]

Wow!  1us range change time, seems to good to be true, makes me suspicious  I look forward to trying it

You are right to be suspicious, and it's great to hear!  Makes me feel better about taking 2 years to develop Joulescope

I have a custom load board that has a 1 mega-ohm resistor in parallel with a 1 Ohm resistor.  The 1 ohm resistor can be switched on/off by a fast MOSFET and gate driver in under 200 ns.  I set my power supply to 1V with a 2A limit and captured the attached plots going from 1 microamp to 1 amp.  Notice that the total duration between the markers is the two images is 4 microseconds.  Joulescope goes from the 18 microamp range with a 1000 Ohm shunt to the 2 amp range with 0.01 Ohm shunt about 1 microseconds (it's actually very hard to measure on the Joulescope output - but it's more fun to see here!).  Note that the voltage drop is between IN+ and IN- and has nothing to do with the Joulescope shunt resistor - this is cable inductance!

Questions?  Still suspicious? 

[BitWrangler1001]

Wow!  1us range change time, seems to good to be true, makes me suspicious  I look forward to trying it

You are right to be suspicious, and it's great to hear!  Makes me feel better about taking 2 years to develop Joulescope

I have a custom load board that has a 1 mega-ohm resistor in parallel with a 1 Ohm resistor.  The 1 ohm resistor can be switched on/off by a fast MOSFET and gate driver in under 200 ns.  I set my power supply to 1V with a 2A limit and captured the attached plots going from 1 microamp to 1 amp.  Notice that the total duration between the markers is the two images is 4 microseconds.  Joulescope goes from the 18 microamp range with a 1000 Ohm shunt to the 2 amp range with 0.01 Ohm shunt about 1 microseconds (it's actually very hard to measure on the Joulescope output - but it's more fun to see here!).  Note that the voltage drop is between IN+ and IN- and has nothing to do with the Joulescope shunt resistor - this is cable inductance!

Questions?  Still suspicious?

It's very impressive, and certainly fills the void that has existed for a while in the IoT space. 

Can the UI zoom into that current edge?  Perhaps just view 10 pts so we can look at the glitches  ?

Have any more kind of real-world examples, like a SoC BLE chip profile example?  The 18mA is the sweet spot when RF circuitry activated for many of the BLE devices I've used.

Would be nice to see a shot of a BLE SoC in polling advertise mode, then to an active data send mode, and then back to idle advertisement mode.

Always suspicious 

[mliberty]

Thanks!  It's been great to hear stories from lots of other people that have struggled with energy measurement & optimization!

The UI zoom is currently limited to 1 sample per pixel.  If you want to really see the glitches, you really need multiple captures since timing varies.  I'll see what I can do...

I don't have a BLE example, and I haven't worked on BLE recently.  Do you have a specific dev board + example code + system setup that you would like to see?

[mliberty]

Can the UI zoom into that current edge?  Perhaps just view 10 pts so we can look at the glitches  ?

I measured the Joulescope step response in a way that can be replicated with reasonable ease, and I posted the results.  Comments?

[tszaboo]

Great, please come back here and let us know what you discover - good and bad.
I'd also like more than 5V capability to cover my various projects (in particular 2S LiFePO4 and maybe 4S LiIon).
I've been considering the ITech IT6411 which is similar to the Keithley 2281S.

OK, so I visited them and spent like 20 minutes talking to the guys, and checking out the equipment.
It seems to be quite OK for the price, like it has the same capture window for the UART as the power measurement, so you can just send log messages on UART, synchronize the power measurement with it, or capture GPIO. It is a powerful unit, even without the battery capabilities. I think we will buy one to test, because honestly, the price is really good. I mean we pay something like 350 EUR for the TTI power supplies, for about 100 EUR extra, you get UART, ADC and a very easy to use software. OF course the power supply has more voltage and current, but I dont think it matters if your device is a battery powered microcontroller project. And then see how it goes from there, probably I have to convince the software guys that this is useful, use it.

[supriyala]

Excellent topic! I got lots of valuable information from here. Thank you.

[Kean]

Great, please come back here and let us know what you discover - good and bad.
I'd also like more than 5V capability to cover my various projects (in particular 2S LiFePO4 and maybe 4S LiIon).
I've been considering the ITech IT6411 which is similar to the Keithley 2281S.

OK, so I visited them and spent like 20 minutes talking to the guys, and checking out the equipment.
It seems to be quite OK for the price, like it has the same capture window for the UART as the power measurement, so you can just send log messages on UART, synchronize the power measurement with it, or capture GPIO. It is a powerful unit, even without the battery capabilities. I think we will buy one to test, because honestly, the price is really good. I mean we pay something like 350 EUR for the TTI power supplies, for about 100 EUR extra, you get UART, ADC and a very easy to use software. OF course the power supply has more voltage and current, but I dont think it matters if your device is a battery powered microcontroller project. And then see how it goes from there, probably I have to convince the software guys that this is useful, use it.

Thanks.  Just to clarify... you are talking about the Otti here?

[tszaboo]

Thanks.  Just to clarify... you are talking about the Otti here?

yes

[kaz911]

Great, please come back here and let us know what you discover - good and bad.
I'd also like more than 5V capability to cover my various projects (in particular 2S LiFePO4 and maybe 4S LiIon).
I've been considering the ITech IT6411 which is similar to the Keithley 2281S.

OK, so I visited them and spent like 20 minutes talking to the guys, and checking out the equipment.
It seems to be quite OK for the price, like it has the same capture window for the UART as the power measurement, so you can just send log messages on UART, synchronize the power measurement with it, or capture GPIO. It is a powerful unit, even without the battery capabilities. I think we will buy one to test, because honestly, the price is really good. I mean we pay something like 350 EUR for the TTI power supplies, for about 100 EUR extra, you get UART, ADC and a very easy to use software. OF course the power supply has more voltage and current, but I dont think it matters if your device is a battery powered microcontroller project. And then see how it goes from there, probably I have to convince the software guys that this is useful, use it.

I have the Otti and the weakness are
1. Sample rate
2. The annual charge for the premium "features"  (Battery profiling; Battery simulation; Scripting & Automation; Importing debug logs. ) (Euro 400/year + VAT)

Which is why I have pledged for Joulescope as I will never subscribe to a set of product features on an annual basis.

But apart from that the Otti is good even without the premium license but spec wise Joulescope is better. And I like the re-replaceable front panels. I just hope for USB-C connector vs the old type A. (Not USB-C PD but just normal "usb" power delivery just via the USB-C connector. I like them better than USB-A/B/Micro.