Batteroo testing

[amspire]

At least 2 uA quiescent draw are not so bad, I think Jay_Diddy_B's batteriser prototypes was about 20-30uA.

2 uA is pretty amazing. That is low enough to have no effect on the life of a battery. For the analog circuitry of the regulator, there is a current consumption versus speed trade-off, and so using a very low power analog circuit may mean really bad transient performance.

However, they could be doing another trick. They could switch the converter off totally at low currents, except for a very low powered comparator. When the comparator detects the 200mV drop, the converter circuit wakes up until the voltage is correct again, and then switches off.

It would be a way to get extremely low quiescent current at the expense of the 200mV p-p waveform at low loads. The big negative would be that if a 1A load is suddenly applied, the output voltage could drop down to almost zero before the converter starts up.

[razvanme]

Don't know how to measure the switching frequency
...............
Who wants the two sleeves next? One will be shipped with the construction with the battery holder, for easy measuring. Would be cool if someone with ESD/EMV equipment could take a look at it, or verify the efficiency results with a non-bricked eload, or test the current limiter.

If nobody wants the sleeves you should sacrifice one for science and find the component values. Also please test with higher currents / oscilloscope attached so we see how it behaves when a load is applied / detached and if the switching frequency and p-p ripple changes. Based on the data so far, it seems like the efficiency has not quite reached it's maximum so higher currents might improve it by a few percent.
You can measure the switching frequency at the inductor, you will have to solder the power in this case. Probe the two sides and see what you get.

Please test the flashlight just so we have more data, even if they said you should not use it in such a device. No need for a movie just a photodector / arduino to track the time.
I have a feeling that we will be ruling out a lot of devices from use.
They might say it's not meant for mp3 players, it's not meant for trains , it's not meant for toy cars, it's not meant for gauged devices, it's not meant for flashlights. And we will be left with the package they came in.

[PeterL]

With 100 ohm load:

Looks like regulator is in sleep mode most of the time, it makes one or two cycle at minimum on-time, overshoots, and goes to sleep. I have seen this behaviour also on convertors from TI and LT. It's not the switching frequency your looking at here, it's an active/sleep rhythm: during the rising edge the converter is active, during the falling edge it's asleep.

Since this is most certainly a boost converter it needs to operate at a duty cycle of (1-Vin/Vout). So if you feed this thing with 1.5V, it basically needs a dutycycle of zero, but these converters use a minimum on-time, which results in a minimum dutycylce if the frequency is fixed.

Things will probably get better once you lower the input voltage, and the converter actually starts boosting.

[LabSpokane]

So far, the efficiency is about what I expected, but that sawtooth output is awful. It's hard to believe that isn't going cause issues.

Custom silicon?!  While you were "customizing," Bobby, while the hell didn't you stuff in a bypass FET?

[FrankBuss]

So far, the efficiency is about what I expected, but that sawtooth output is awful. It's hard to believe that isn't going cause issues.

It is no problem with the MP3 player, can't hear any distortions or noise. Maybe it has good power supply filtering.

[EEVblog]

So Frank has entered the first test in the spreadsheet (the toy train):
https://docs.google.com/spreadsheets/d/18K9c2YAT0d0QABGYGpzItbvDcgfAQCRUtloEzzfXADU/
And it was nothing short of a horrible result. -53% (minus!) Batteriser run time with a new battery, and a paltry 3.1% extra run time form the "flat" battery. Not 30%, not 300%, 3%.

[CJay]

So Frank has entered the first test in the spreadsheet (the toy train):
https://docs.google.com/spreadsheets/d/18K9c2YAT0d0QABGYGpzItbvDcgfAQCRUtloEzzfXADU/
And it was nothing short of a horrible result. -53% (minus!) Batteriser run time with a new battery, and a paltry 3.1% extra run time form the "flat" battery. Not 30%, not 300%, 3%.

It's a bloody awful result and I'd not expected it to be anywhere near that bad but can we be sure the original cell wasn't faulty* from the packet or was it repeated with other cells?

Or am I misunderstanding how the test was applied?

*(it's happened to BigClive with Duracell Simply AA cells, one was around 400mAH and the other from the same pack was 1700mAH)

[FrankBuss]

It's a bloody awful result and I'd not expected it to be anywhere near that bad but can we be sure the original cell wasn't faulty* from the packet or was it repeated with other cells?

You are right, I should have measured the voltage. But the speed of the train was as fast as with the other cell, at least in the beginning, and the 4 minute result is not affected, because this was done with the other battery which was used in the first test.

For details see the video, it is uploaded now (needs still some time until HD is processed for better quality) :



Video description: Testing a Brio Battery Powered Engine, Art# 33595, with the Batteroo sleeve. The tracks were glued with double-sided tape to the pertinax board, for same test conditions. Result: without the sleeve it ran for 127 minutes, with the sleeves for 59 minutes and using the sleeves on the dead battery of the first test, it ran for 4 minutes. The video shows all 3 tests in one shot, no editing, only time lapse when it is not interesting. In the small overlay at bottom right in the first half of the video, you can see the test with the sleeve, to compare it without the sleeve.

At 6:28 you can see in the overlay how it dies with the sleeve, while it is still running without the sleeve at this time, and after this you can see the 4 minute run, and at the end of the video at 24:20. You can also see that the train runs faster without the sleeve.

Conclusion: The Batteroo sleeve has no advantage for this toy.

I published the video under CC license, so anyone can use it for own videos, for voice overs etc. (I'm not much of a talker, as you can hear in the video).

[onlooker]

I guess the "dead" battery was allowed to recover for about one hour before test 3. One more interesting test  as others have already mentioned is to measure a "dead" battery's extra time without the sleeve and after one hour recovery.

[FrankBuss]

I guess the "dead" battery was allowed to recover for about one hour before test 3. One more interesting test  as others have already mentioned is to measure a "dead" battery's extra time without the sleeve and after one hour recovery.

Right, I guess it might be longer, too. But doesn't matter much, the interesting result is that the advantage is 3% instead of 300%. Maybe a problem while they calculated it? "0.03, hmm, times 100, percent. Oh, did I already pressed enter?"

But I just did the train test with the sleeves again, this time checking the battery (was from the same pack of 4) and open loop voltage was 1.6 V. It is important to verify experiments. The Batteroo sleeve was the other one I have, so no faulty sleeve, no faulty battery. Time confirmed, 61 minutes this time. The train is available everywhere to buy, e.g. Amazon, everyone can do the same test. I didn't specifically chose the train because of its bad performance, it was just one of the few items I could find at all which has a motor and uses 1.5 V AAA batteries. I expect lots of other similar toys will have the same problem, like probes the monkey.

While testing the MP3 player, the clips at the end broke off of one of the sleeves I have when I removed it from the player. I was careful, but the MP3 player has little springs at the positive battery terminal, too, and it seems it got stuck with it. Can't find the clips, they are in Batteroo nirvana now.



Trying to photograph through the lens of my microscope (I'm sure Dave's awesome Tagarno will produce much better images soon). Looks like they were soldered:



And the inside:

[CJay]

It's a bloody awful result and I'd not expected it to be anywhere near that bad but can we be sure the original cell wasn't faulty* from the packet or was it repeated with other cells?

You are right, I should have measured the voltage. But the speed of the train was as fast as with the other cell, at least in the beginning, and the 4 minute result is not affected, because this was done with the other battery which was used in the first test.

I'm not sure the terminal voltage would have been different if a cell was low capacity, after all a D cell has much higher capacity than an AA but the same terminal voltage from new.

 I'd just be interested in seeing the test repeated a few times with same brand/type cells to eliminate the possibility of a low capacity cell.

I'm almost curious enough to try and lay hands on some sleeves to play with but, damnit, if this carries on Batteroo will make a living selling sleeves to curious engineers

[PA0PBZ]

Time confirmed, 61 minutes this time.

Anyone care to guess why the result is so bad? Simply looking at the efficiency does not explain it.
Frank, did you measure the current consumption of the train? Maybe I missed it...

[dcac]

I’ve seen some tests of AA batteries where the capacity variation in a 10 cell pack could be as much as 70% difference between the best and the worst cell from the same package. But this was the extreme case in the 9 brand test, other brands varied about 3-20%, so still something that really should be taken into consideration.

What are the capacity tolerances on a typical battery anyway? are they ever specified. I believe output voltage and internal resistance are monitored at manufacturing, but how about the capacity the battery can deliver when its energy is drawn over hours/days/months or even years.

Though I’m not saying this explains the test results we seen so far.

[FrankBuss]

Anyone care to guess why the result is so bad? Simply looking at the efficiency does not explain it.
Frank, did you measure the current consumption of the train? Maybe I missed it...

Current is about 250 mA at 1.5 V and 210 mA at 1 V.

[razvanme]

Anyone care to guess why the result is so bad? Simply looking at the efficiency does not explain it.
Frank, did you measure the current consumption of the train? Maybe I missed it...

Current is about 250 mA at 1.5 V and 210 mA at 1 V.

That might explain it, we don't have any data for higher current draw, if it takes 250mA at 1.5V and the batteriser keeps that, it will draw more and more current.
We need some data at higher current draw.

[onlooker]

Right, I guess it might be longer, too. But doesn't matter much, the interesting result is that the advantage is 3% instead of 300%.

I understand your argument. I am just curious whether that 3% extra turns out to be 0% in real term.

I think your test results should be indicative for all direct battery powered motorized devices and "toys",  including the monkey ...

The other device types of interest are those with built-in DC-DC convertors and those with pulsed power consumption.  The interest  is mainly on checking all different predictions made throughout the 300+ pages.

[LabSpokane]

I’ve seen some tests of AA batteries where the capacity variation in a 10 cell pack could be as much as 70% difference between the best and the worst cell from the same package. But this was the extreme case in the 9 brand test, other brands varied about 3-20%, so still something that really should be taken into consideration.

What are the capacity tolerances on a typical battery anyway? are they ever specified. I believe output voltage and internal resistance are monitored at manufacturing, but how about the capacity the battery can deliver when its energy is drawn over hours/days/months or even years.

Though I’m not saying this explains the test results we seen so far.

With alkaline batteries, I think it's very safe to say that the error bars are at least 10%. A 3% difference is no difference at all.

[FrankBuss]

The MP3 player test is done. I used the method Dave described, just filming it, no measuring, to actually see when it turns off (looks like there was no low battery warning, maybe voltage jumped a bit). With sleeve a fresh battery worked 17.9% less long than without the sleeve and the additional time after using the sleeve on the dead battery from the first test with the MP3 player (which had a few days to recover) was 13.4 %.

[LabSpokane]

And holy cold solder joints!  That's terrible. You'd think someone from Flextronics would know those terminals need to be preheated before reflow. 

[Hensingler]

Anyone care to guess why the result is so bad? Simply looking at the efficiency does not explain it.
Frank, did you measure the current consumption of the train? Maybe I missed it...

Because the train is a non-switching regulated load and increasing the feed voltage above the train's drop out voltage is pure waste. That and batteroo losses means twice as many batteries will end up in landfill.

This train is a good example of where batteroos are a complete fail. The only way to win is to put sleeves on already dead batteries and in this case you get another 3.1%. Given a batteroo costs about 6 times more than quality Amazon branded batteries you would have to clip onto nearly 200 dead batteries to break even which no one sane would bother to do and the batteroos would likely fall to bits long before that anyway.

[PeterL]

Anyone care to guess why the result is so bad? Simply looking at the efficiency does not explain it.

I think one thing that explains it is the way the batterisers loads the battery.

Without the batteriser there is a constant load on the train of .2 Amp.
With it say we can assume a pulsating load, judging from the voltage ripple we have seen. Let's say the load is now 2 Amp for 10%
(neglecting the losses in the batteriser)
This has a big effect on the Powerlosses in the ESR:

Without Batteriser:

P_ESR = 0.2² x R_ESR = 0.04R_ESR

With Batteriser:

P_ESR =  10% x 2² x R_ESR = 0.4R_ESR

So we have much more losses in the ESR due to the batteriser.

It would be interesting to see the current waveform on the input of the batteriser at different charge levels of a battery to check this theory.

[StillTrying]

Current is about 250 mA at 1.5 V and 210 mA at 1 V.

That's about 1/2 of what I expected. Is that with the train just held in hand, or driving itself around the track?
200mW-300mW doesn't seem enough input power to drive a wooden train around a wooden track.

data.energizer.com/PDFs/E92.pdf

[LabSpokane]

The MP3 player test is done. I used the method Dave described, just filming it, no measuring, to actually see when it turns off (looks like there was no low battery warning, maybe voltage jumped a bit). With sleeve a fresh battery worked 17.9% less long than without the sleeve and the additional time after using the sleeve on the dead battery from the first test with the MP3 player (which had a few days to recover) was 13.4 %.

A test of runtime using recovered battery with *no* sleeve is also needed.

[samgab]

Thanks for the Youtube video with the toy train Frank. Excellent real world demonstration, shows simply and beyond a doubt that there is no advantage - rather a disadvantage - to using the sleeves compared to a battery by itself. Looking at the train on the round track like a clock, you can even see that the train runs faster throughout the test with just the bare battery.
One simple real world test has utterly exposed the Batteroo as total bullshit.

[samgab]

Right, I guess it might be longer, too. But doesn't matter much, the interesting result is that the advantage is 3% instead of 300%.

I understand your argument. I am just curious whether that 3% extra turns out to be 0% in real term.

I think your test results should be indicative for all direct battery powered motorized devices and "toys",  including the monkey ...

The other device types of interest are those with built-in DC-DC convertors and those with pulsed power consumption.  The interest  is mainly on checking all different predictions made throughout the 300+ pages.

Yes, I can almost guarantee that if the bare battery from the test with bare battery had been left to rest for an hour and reinserted again bare, without the batteroo, it would have run for at least as long, probably longer again than it did with the batteroo on. As a kid that was always my method with dead batteries, swap them back and forth giving them time to rest and the toy would always run again for a good while.
You'd actually get a better increase in life from the "dead" battery by biting it hard and reinserting it than by putting a batteroo sleeve over it. Seriously.