Batteroo testing

[FrankBuss]

MOD NOTE: This is thread is ONLY to discuss Batteroo test results. If you want to discuss Batteriser/Batteroo in general, use the main thread:
https://www.eevblog.com/forum/blog/eevblog-751-how-to-debunk-a-product-(the-batteriser)/


This new thread is dedicated to test procedures and results of the Batteroo sleeve. There is a spreadsheet which lists the performance gain you can get from Batteroo for different products, see here:

https://docs.google.com/spreadsheets/d/18K9c2YAT0d0QABGYGpzItbvDcgfAQCRUtloEzzfXADU/edit?usp=sharing

If you have a Batteroo, please test your products with it, instructions are at the top of the spreadsheet (no test gear needed) and then send Dave the results to dave@eevblog.com and he'll add them to the spreadsheet.

Some of the results you can find in this thread:

My train test:



Diagram:



Dave's train test:



Diagram:



Youtube reviews by EC-Projects, he measured a quiescent current of 14.4 uA





Measurements by dexters_lab:





Closeup images of the circuit: https://www.eevblog.com/forum/projects/batteroo-testing/msg1107692/#msg1107692

Output ripple measurements:

by Dave:

150ohm load (resistor box), and a 1.4V Rigol DP832 PSU input.



And with no load @ 1.4V PSU input


By me:

With 1 k load:



With 100 ohm load:



By Ysjoelfir:

First, no load connected. The Batterizer is switching between battery voltage (~1.48V) and around ~1.64V with around 10 kHz.


another one with bandwidth limitation


Switched on, measured directly at the load, connected with short cables: around 200mV of "noise"


Let's zoom in a bit: switching between battery voltage, which breaks down to 1.39V under load and 1.61V with a frequency of around 40 kHz.


there are tiny spikes reaching up to 1.67V


A bit closer to measure the frequency: 40.48 kHz.


A last one without the cursors, just for curiosity. Interesting waveform with nice tiny spikes at the rising edge.

[bills]

Great will be following.

[jeroen79]

First thing I thought of was batteriser.
Guess what? The video on the batteroo site is about the batteriser.
Just a rebranding due to bad press?

Just an idea:
You are putting current measuring resistors in the supply and load circuits.
But the power supply and the load would have their own internal shunts.
Couldn't you measure the voltage across those and leave out the extra shunts?
That way you wouldn't have to worry about extra voltage drops.

[Wytnucls]

More data about the converter may be necessary:

Efficiency
Input voltage range
Input start-up voltage
Quiescent current (uA)
Auto PWM/PFM modes?
Operating frequency (1 or 2MHz?)
Current limit
Transient response
Thermal shutdown (TSD)?
Output voltage ripple
Inductor size and package
Output/Input cap size and type (Ceramic?)
ESD susceptibility

Industry standard graphs:

[wraper]

First thing I thought of was batteriser.
Guess what? The video on the batteroo site is about the batteriser.
Just a rebranding due to bad press?

They renamed themselves because Energizer was after their ass.

[Cerebus]

A snippet to share with respect to "real world testing".

One of the products that's been mentioned several times is the Apple Wireless Keyboard and also similar devices. During the discussion of this in the other thread, I was prompted to measure the current consumption of mine. I ended up having to measure the current on the mA range on my (Fluke 25) multimeter when the uA range would have been more appropriate. Why? The keyboard just wouldn't turn on with the voltage drop across the shunt on the uA range.

So the lesson to be learned from this is that current measurements on some devices are going to require very low voltage drops across any current shunts. One, because they may already be operating on quite small voltage margins they may not operate normally, or at all. Two, any additional voltage drops are going to pollute the measurements in a test setup as compared to normal usage. It might be time to break out some very low resistance shunts and some high gain instrumentation amplifiers. I've got a spare 4 wire Vishay VCS201 5 mR (yes, 5 milli-ohm) shunt or two about that I can donate if it becomes necessary.

[amspire]

More data about the converter may be necessary:

Efficiency
Input voltage range
Input start-up voltage
Quiescent current (uA)
Auto PWM/PFM modes?
Operating frequency (1 or 2MHz?)
Current limit
Transient response
Thermal shutdown (TSD)?
Output voltage ripple
Inductor size and package
Output/Input cap size and type (Ceramic?)
ESD susceptibility

Also need:

How does Batteriser behave after shutdown: Open circuit, forward diode junction from battery, reverse diode junction across output?
If it goes open circuit, what is the maximum reverse voltage it can handle?
How down Batteriser behave with a reverse voltage across output ?
How does the Batteriser behave if the connection with battery is intermittent while under load - does it latch up?
If it does still conduct after shutdown, how much current can it handle? The losses may be much higher in this mode then when the converter is running so the current capacity may be much less.

These cases are only relevant for the situation when there are several batteries in series.

[IanB]

One of the products that's been mentioned several times is the Apple Wireless Keyboard and also similar devices. During the discussion of this in the other thread, I was prompted to measure the current consumption of mine. I ended up having to measure the current on the mA range on my (Fluke 25) multimeter when the uA range would have been more appropriate. Why? The keyboard just wouldn't turn on with the voltage drop across the shunt on the uA range.

This is puzzling. What was the actual current?

I have measured the input resistance of the µA range on three of my meters and in each case it was 100 \$ \Omega \$. Therefore if the current was 1 mA the voltage drop would be 0.1 V. That should be well within the tolerance of the keyboard with fresh batteries.

[ataradov]

Therefore if the current was 1 mA the voltage drop would be 0.1 V.

Wireless keyboard will consume way more than 1 mA at startup.

This is a common problem with measuring power consumption of sleeping devices. Some meters handle switch between uA and mA ranges just fines. Flukes (at least some) don't, if I remember correctly.

[djos]

Therefore if the current was 1 mA the voltage drop would be 0.1 V.

Wireless keyboard will consume way more than 1 mA at startup.

This is a common problem with measuring power consumption of sleeping devices. Some meters handle switch between uA and mA ranges just fines. Flukes (at least some) don't, if I remember correctly.

That's cool, Dave will test with something reliable... Like a Brymen.

[ez24]

This is the test I want to see



https://www.amazon.com/Westminster-Toys-Magic-Toy-Monkey/dp/B0000V4H4O/ref=sr_1_2?ie=UTF8&qid=1481957889&sr=8-2&keywords=toy+monkey+battery

And the counts of the cymbals be put in spreadsheet and a statistical analysis be done on them.

[EEVblog]

Industry standard graphs:

Which chip is that from?

The efficiency vs load graph will be the first thing I do.
BTW I shot a video today on testing performance curves.

[FrankBuss]

You are putting current measuring resistors in the supply and load circuits.
But the power supply and the load would have their own internal shunts.
Couldn't you measure the voltage across those and leave out the extra shunts?
That way you wouldn't have to worry about extra voltage drops.

I don't have access to the internal shunts and don't want to solder anything in the power supply or the electronic load. But the voltage drop is no problem with the uCurrent, because up to an amp it is well below 0.1 V, and I can compensate this in my testing script and incrementally increase the output voltage in 10 mV steps, so that the max error will be 10 mV, e.g. for the efficiency vs. load graph which requires a fixed input voltage (unfortunately my power supply doesn't have extra sense inputs).

I like the efficiency vs. load current diagrams for a given input voltage, will do this, too. And the tests Wytnucls suggested are very useful, too. Output ripple would be interesting, because one application to use a battery is for low noise things like microphone amplifiers, and they might be built in a way that they expect a very clean supply.

I don't have the right equipment or knowledge to do the ESD tests, but if someone can do this, I can send the sleeves to him after my tests.

[Wytnucls]

Industry standard graphs:

Which chip is that from?

The efficiency vs load graph will be the first thing I do.

Analog Devices ADP1607
http://www.analog.com/media/en/technical-documentation/data-sheets/ADP1607.pdf

[Cerebus]

This is puzzling. What was the actual current?

The simple answer: 200 uA at idle, 3-4 mA peaks while transmitting. The more rigorous answer: the Fluke 25 has a longish integration time and a slow response (2 readings per second*) so was not possible with that setup to see the shape of the current curve as the thing was starting up. I suspect there's an interaction between an inrush current, the meter's shunt resistance (500R on uA but only 5R on mA) and undervoltage lockout on the keyboard's power supply.

If I can sum up the energy and find the time in all the Christmas preparations (including the requirement to get my scope, breadboard and bench power supply off the living room coffee table**) I'll try and rig up something to characterise what's going on. My curiosity is piqued too. Moreover, I'd quite like to characterise the Apple keyboard's battery gauge against actual in circuit voltages.

*Checking the specs, it actually does 25 readings per second on the analogue bargraph; but I wasn't watching that, just the main display.

**Add to that, it's also almost time for me to go and do the spud bashing for the Saturday evening roast (Chicken and mushroom pie, roast potatoes, roast parsnips, carrots and peas plus extra gravy, if anyone's interested). By my reckoning it'll be Tuesday before I have any time for electronics.

[FrankBuss]

I'm preparing the tests, cleaning and ordering the gear on my workbench a bit. And I bought some fresh AAA batteries, from different vendors, because the typical end user test would be important as well. Here is Dave's spreadsheet for entering test results:

https://docs.google.com/spreadsheets/d/18K9c2YAT0d0QABGYGpzItbvDcgfAQCRUtloEzzfXADU/edit?usp=sharing

I'll do some of these simple tests as well, as described in the spreadsheet. But it is difficult to find gadgets which run on 1.5 V AAA batteries. Most modern devices have accus, with USB charging, and many other devices I have use AA batteries. But I found a crusty old USB/MP3 player, you can see it in the image below. I tested it with a power supply and it turns off at about 1.12 V, when you lower the voltage slowly (works fine indefinitely at 1.12 V, but turns off randomly after some time at 1.11 V). Between 1.12 and 1.14 V it says "low battery when you try to turn it on at this level. At 1.14 V it turns on normally. Sound volume at the headphones is the same at 1.5 V and 1.12 V (but I'll verify this by measuring the voltage, will solder some audio rectifier and very low low pass filter to get an average of a few minutes for an automated test). With a typical application (playing "AC/DC - Highway to Hell" with max volume in an infinite loop), it needs about 70 mA. So a battery should be drained in less than a day.



In the photo you can see the electronic load, the power supply, another power supply for -5 V (for the analog switch and if I need symmetrical voltage for op-amp circuits, like the audio rectifier), the benchtop multimeter and the Raspberry Pi, controlling it all, connected by ethernet so that I can ssh to it, and it has mounted a folder on my NAS (with RAID0 configured harddisks), to avoid wearing out the SD-card when logging for a long time. A RS232 TTL to +/-9V converter I built some time ago with a nice box connects the RS232 of the multimeter to the 3.3V TTL level pins of the first serial port on the Raspberry Pi GPIO headers. The second RS232 connection to the electronic load is implemented with a RS232 USB adapter. And finally the power supply is connected by USB, too.

[razvanme]

I'm preparing the tests, cleaning and ordering the gear on my workbench a bit. And I bought some fresh AAA batteries, from different vendors, because the typical end user test would be important as well. Here is Dave's spreadsheet for entering test results:

https://docs.google.com/spreadsheets/d/18K9c2YAT0d0QABGYGpzItbvDcgfAQCRUtloEzzfXADU/edit?usp=sharing

I think step 2 and 7 combined might mess with the results, the dead batteries regenerate (I understand they don't recharge), even 1 minute of on time with a low load (1-2ma) might give batteriser a lead. I don't think it's fair to wait for them to regenerate. You should have another step
3.1 - Run the dead batteries from step 2 with batteriser and note the "extra" time if any.

Although doing this, I don't know how you will calculate the final results ... hmmm  . I think you are already taking this into account in step 8-10.

This depends on how long they will wait to be used.

PS. Regarding the current reading, and not soldering to the sleeve. You can take any 9V battery, open it up, you will have a bit of metal strips that you can use, you don't have be a mechanic, you just have to take two, isolate them and slip them between sleeve and the battery. You can even make a smal indent in one to make good contact.

[Towger]

But it is difficult to find gadgets which run on 1.5 V AAA batteries.

I am sure Dave can DHL you a couple complementary EEVblog multimeters for side by side testing. One with and one without AAA size Batteroos.[emoji57]

[CrashO]

It'l be interesting to see the results when they come in.. that's for sure 

[Mr.B]

Very interested in the results, so, bookmarked...

[razvanme]

@FrankBuss

Part of the reason I'm sending in pairs, is to test them combined, I'm curios how it behaves when you have two batteries in series / parallel, do you get 3V constantly until both are dead, do they share the current in parallel.
Also you have an oscilloscope, so you can check the switching frequency. Check the standby current before you start pushing them to the limits .

After you test them (and possibly blow one up), check the other components, I guess the big component is an inductor, what are the others and what values.

[EEVblog]

I'm preparing the tests, cleaning and ordering the gear on my workbench a bit. And I bought some fresh AAA batteries, from different vendors, because the typical end user test would be important as well. Here is Dave's spreadsheet for entering test results:
https://docs.google.com/spreadsheets/d/18K9c2YAT0d0QABGYGpzItbvDcgfAQCRUtloEzzfXADU/edit?usp=sharing

Yes, I encourage everyone to follow that procedure for product testing (no test gear needed) and send me the results to dave@eevblog.com and I'll add them to the spreadsheet.

[FrankBuss]

I'm preparing the tests, cleaning and ordering the gear on my workbench a bit. And I bought some fresh AAA batteries, from different vendors, because the typical end user test would be important as well. Here is Dave's spreadsheet for entering test results:

https://docs.google.com/spreadsheets/d/18K9c2YAT0d0QABGYGpzItbvDcgfAQCRUtloEzzfXADU/edit?usp=sharing

I think step 2 and 7 combined might mess with the results, the dead batteries regenerate (I understand they don't recharge), even 1 minute of on time with a low load (1-2ma) might give batteriser a lead. I don't think it's fair to wait for them to regenerate.

I don't think this will result in much difference. And it is always good to use best-case scenarios, because if they performs badly even then, you can be sure that Batteroo  is not good for the tested product.

To check exactly when the MP3 player I'm using for the first test stops playing, I created a simple peak detector with very slow fallback:



I'm not good with analog electronics, so first I forgot R5 and R6 and the voltage at the plus input was slowly running towards -5 V, but I saw this in the logging and could fix it. This is how the peak output looks like, compared to the amplified input after the first op-amp:



This is the script I'm using to record the data (it is really easy with the RPi_LogNut framework), and here are the first minutes of the log (both voltages are measured every 10 seconds precisely). As you can see, the peak level slowly faded away, until I hot-fixed it by adding the two resistors But looks good so far:



The battery voltage is measured with two wires in the battery compartment, between the battery and the contacts. This should not influence Batteroo later in any way. Thanks Dave for your latest video, I might need to replace the ADG608 with an ADG609, so that I can switch two signals. In this setup the battery minus is directly connected to the multimeter minus, but later when I switch to the shunt for the characteristics curve etc., I need to switch both multimeter inputs, to avoid the voltage drop on the wires for higher currents.

I hope logging works for hours, never tried this before. When it is done, the battery will have some time until the Batteroo arrives to regenerate.

[ez24]

FrankBuss can you post a picture of your MP3 player, it looks like it could be different than 

https://www.aliexpress.com/item/High-Quality-8GB-USB-2-0-Flash-Drive-LCD-Mini-MP3-Music-Player-with-FM-Radio/32644511571.html?spm=2114.01010208.3.1.ulkwFJ&ws_ab_test=searchweb0_0,searchweb201602_2_10065_10068_10084_10083_10080_10082_10081_10060_10061_10062_10056_10055_10054_10033_10059_10032_10099_10078_10079_10077_427_10103_10073_10102_10096_10052_10050_10051-10033_10077,searchweb201603_3&btsid=46859591-b34f-4d2a-9a43-551af6a52f12

I have used these for years and have collected a dozen so far.  It was the first thing I thought of when I learned about the Bat

(sorry about the link, I do not know how to shorten it)

[FrankBuss]

FrankBuss can you post a picture of your MP3 player

It is a really old one, from 2004, didn't use it anymore since the iPhone came out. I think it is this:

http://www.ciao.de/Pro2_Mp3_Player_256_MB__1205082

But not the first picture, the other 3 pictures. Can't remember the price, but some people in the comments wrote that it was 80 EUR, with 256 MB flash and not so good sound quality, but this was normal for this time.

I think there is a firmware message for a short time when I turn it on, will try to take a picture when the battery is empty.