Batteroo testing

[IanB]

Batteroo have already admitted on their product sheet that it's not recommended with torches with DC-DC converters, why?, because they are active loads!

Not all LED lamps have good regulation, for example, this one:

Ultimate Survival Technologies 10-Day 6-AA Lantern

It takes 3 or 6 AAs in a 3s2p arrangement. I bought it for blackouts/emergencies based on the good reviews, but when I tested it I found it has absolutely no regulation at all. The current consumption increases exponentially with increasing battery voltage, about twice as much at 1.5 V as at 1.1 V, and rising ever more rapidly if you go above 1.5 V.

I guess this works OK with alkaline batteries, but I think the design would be much better if it had regulation to maintain a constant, stable brightness when the voltage increases above something like 1.2 V.

Anyway, to the point: I would guess the Batteroo would work against this lamp as boosting the input voltage will drain the batteries much quicker due to the voltage/current load characteristic.

[samgab]

I know there are dozens of ways to do a reed switch counter circuit, using 4026 IC, ATmega8 circuit, PIC16F84 circuit, some kind of Arduino circuit, using 7 segment displays etc, but for anyone out there who is thinking of setting up a simple counter circuit for any kind of testing, where you want to keep it as simple as possible and don't want to make a little circuit for it, here's a nice cheap and easy solution using a cheap-as-chips pedometer and a reed switch:
http://www.trainelectronics.com/artcles/pedometer/index.htm

https://goo.gl/juvaWu

The advantage of using a reed switch and magnet is that there is no physical contact to trigger the switch, so it doesn't affect the device under test. Go nuts people!

[razvanme]

Keeping a constant voltage to a passive load should be the same thing? If I keep a constant 1.5V over a 15 load, it means I am keeping a constant 100mA over the load. So the batteroo is a lot worse at doing that, in any passive load. Am I missing something? It seems it's the exact thing they warn about just worse.


Also, they might get away with "murder" since they do not mention anything about extra battery life, they were very careful with the wording.

"Don't waste time on new batteries to get great performance, just slip on a Batteroo from the start and your device will perform at it's best until all the energy is used up"

It's what I expected, the train will run full speed until the end, it seems even that is a fail.

[samgab]

Does anyone else here remember user 5ky here on the forum https://www.eevblog.com/forum/profile/?u=110170 , he has already got the same Garmin unit that was used in batteroo's infamous rigged test, and has done some pretty good initial testing. He has also ordered batteroos, but he hasn't been active on the forum here for a year or so. I hope we see some more from him when his batteroos arrive!

[6581]

The advantage of using a reed switch and magnet is that there is no physical contact to trigger the switch, so it doesn't affect the device under test. Go nuts people!

As long as the magnet on the DUT is kept away from motor and batteroos inductor - could affect results if too near. (I guess the effect to motor would cancel out though.)

[Brumby]

It hasn't been clear to me - but how many times have each of these tests been run?

A single test is a very dangerous sample on which to base conclusions.  I would suggest getting a pack of 6 batteries and running the train test 6 times - 3 times with a sleeve and 3 times without.

[Brumby]

The advantage of using a reed switch and magnet is that there is no physical contact to trigger the switch, so it doesn't affect the device under test. Go nuts people!

As long as the magnet on the DUT is kept away from motor and batteroos inductor - could affect results if too near. (I guess the effect to motor would cancel out though.)

Simple answer - just use a light source and a photosensitive device where the train interrupts the beam.

[f4eru]

Thought about getting two and running side-by-side but then people might complain the trains aren't identical. So will run two tests and then edit side-by-side footage with timer. Maybe add a lap counter?

That's not a problem : color code the locomotives, then exchange them and re-run the test, all in one shot.

For the lap counting, Dave, you could use two of your huge collection of 4-bangers, and rig the "=" button to reed switches
Then type "0+1", and if it repeats, it counts the laps.

[Wytnucls]

Here is an attempt to identify the DC to DC converter IC (I find it unlikely that they would have developed their own IC at that price point).
There are no resistors to set the output voltage, so it is probably a fixed voltage output IC (~1.5V), which usually comes in a 6 pin package. This one has 8 pins.
A 2.2uF inductor usually means a switching frequency of 2MHz for similar ICs.

The picture is a bit blurry, but here my best guess for the PCB topology and pin functionality:
Feel free to chime in if you spotted other connections

[IanB]

A single test is a very dangerous sample on which to base conclusions.  I would suggest getting a pack of 6 batteries and running the train test 6 times - 3 times with a sleeve and 3 times without.

The number of tests for a significant result depends on the magnitude of the effect being measured. If the difference is something striking like 2:1, then only a couple of tests with a similar outcome is enough to start establishing confidence in the conclusion.

[Brumby]

That's why I mentioned three.  Three very similar results will give a very high degree of confidence.

[FrankBuss]

Frank,

Some tests you could do.

Good ideas. But I shouldn't do it all by myself, then there is nothing left for Dave and others to test  Dave could measure the temperature with his IR camera, and/or tape a thermocouple on the case with kapton tape. I have only one sleeve left, don't want to break this. The packet is now shipped to the next tester. I have included another cute little toy I ordered from eBay and just got today and didn't test so far, would be interesting to see it in a video.

Thought about getting two and running side-by-side but then people might complain the trains aren't identical. So will run two tests and then edit side-by-side footage with timer. Maybe add a lap counter?

A lap counter would be nice. I thought about this, too, but didn't have much time, always crazy time at the end of the year with client projects. An easy way would be a photodiode, and then an Arduino which measures it on an analog input, and a simple software high pass filter and hold-off, no need for calibration. Outputs each lap time on the serial port.

Running two side-by-side is possible, just do two tests with four batteries, first test train A without sleeves, train B with sleeves, second test train A with sleeves and train B without sleeves, all in one shot. This should proof that there is not much difference. Then wait for a day so that the batteries can recover. Then a third test, using the two batteries from the tests without the sleeve: one in train A without the sleeve, one in train B with the sleeve. I guess the results will be really devastating for Batteroo. Don't forget to measure the open loop voltage of each battery at the beginning and the end of every test.

The number of tests for a significant result depends on the magnitude of the effect being measured. If the difference is something striking like 2:1, then only a couple of tests with a similar outcome is enough to start establishing confidence in the conclusion.

I'm pretty confident that the train tests are significant. I did now four tests, two with the sleeve and two without the sleeve, with two different battery brands. What I was thinking, after the second test you could argue that the train might have wear off, or getting dusty or something like this. But this is not valid anymore after the 4th test, and all tests are showing that with the Batteroo sleeves the time was significantly reduced, and the 4th test suggesting, that even the additional time for dead batteries with the sleeve is lower than without it (but maybe the train is running a bit faster for a very short time). Needs 6 tests as described above to compare all scenarios.

[rollatorwieltje]

Here is an attempt to identify the DC to DC converter IC (I find it unlikely that they would have developed their own IC at that price point).
There are no resistors to set the output voltage, so it is probably a fixed voltage output IC (~1.5V), which usually comes in a 6 pin package. This one has 8 pins.
A 2.2uF inductor usually means a switching frequency of 2MHz for similar ICs.

The picture is a bit blurry, but here my best guess for the PCB topology and pin functionality:
Feel free to chime in if you spotted other connections

Not quite sure about the green Vout. If that was actually ground, the pinout matches the Monolithic Power MP3414 closely, except for the feedback pin of course.

[conspiracy hat]
Monolithic Power is about a 20 minute drive from Batteroo HQ
[/conspiracy hat]

[EEVblog]

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 %.

And there goes the ball game.
Typical passive device and typical active device FAIL

And who was it that predicted maybe a 10-20% typical increase from a "dead" cell in their original video?
And also the Oz university professor interviewed by the newspaper predicted the same thing.
The odd device might be higher, but I'm pretty sure the results are not going to be pretty...

[bktemp]

Here is an attempt to identify the DC to DC converter IC (I find it unlikely that they would have developed their own IC at that price point).
There are no resistors to set the output voltage, so it is probably a fixed voltage output IC (~1.5V), which usually comes in a 6 pin package. This one has 8 pins.
A 2.2uF inductor usually means a switching frequency of 2MHz for similar ICs.

Are there any boost converter ICs preset to 1.5V output voltage?
I couldn't find one. 1.5V is an unusual output voltage for a boost conveter. Typically you need 3.3V or at least 1.8V for powering a microcontroller from a single battery.
Therefore my guess would be it is a customized version, so even the pinout could be customized.
Guessing from the high output ripple and the bad load regulation, it also doesn't seem to have a fixed switching frequency.
Many ultra low power regulators turn the switch on until a peak current is detected, then turn it off until the output voltage is below a fixed theshold voltage. The output regulation will be bad, but the control circuit will be much simpler, consuming less power.

[razvanme]

Here is an attempt to identify the DC to DC converter IC (I find it unlikely that they would have developed their own IC at that price point).
There are no resistors to set the output voltage, so it is probably a fixed voltage output IC (~1.5V), which usually comes in a 6 pin package. This one has 8 pins.
A 2.2uF inductor usually means a switching frequency of 2MHz for similar ICs.

The picture is a bit blurry, but here my best guess for the PCB topology and pin functionality:
Feel free to chime in if you spotted other connections

The Red Vout in your image is actually Ground.



The IC is mirrored. VIN is on the Left.

Therefore my guess would be it is a customized version, so even the pinout could be customized.

Don't think it's a custom chip, what's the point of having 3 grounds and 2 unconnected pins. They could have done a 4 pin chip.

[MatthewEveritt]

I have one of these single AAA toy trains.
http://www.toysrus.com/buy/preschool-trains/imaginarium-power-steam-engine-train-set-5f5eedb-12595303
Will do the same test as Frank did.
Thought about getting two and running side-by-side but then people might complain the trains aren't identical. So will run two tests and then edit side-by-side footage with timer. Maybe add a lap counter?

If you record time per lap you can look at how the train slows down -  might be interesting. Toyboy discharge curves.

[bktemp]

Don't think it's a custom chip, what's the point of having 3 grounds and 2 unconnected pins. They could have done a 4 pin chip.

GND is the most critical pin: During switching the current ramps up slowly and then abruptly drops to zero when the switch turns off -> You need a good connection to GND with low parasitic inductance.
In your drawing the two left GND pins are probably power GND (short connection to input capacitors), while the right GND pin could be signal GND for sensing the output voltage (short connection to output capacitors).

It would be interesting to see if both NC pins are really not connected or if they are unused functions like undervoltage detection, power good, enable or similar usual functions.

I don't think it is a completely custom silicon, but maybe a customized version (output voltage set to 1.5V and maybe custom pinout).
Many manufacturers offer customized voltage options in addition to the usual output voltage version.

[mikeselectricstuff]

A crude motor load like this may well cause confusion to the regulator - might be interesting to look at the voltage waveform on the output & input current. With such a difference in lifetime, a question is where is the energy actually going - batteriser heating? motor heating?
Or maybe just crapping out early - might be interesting to take the battery out of the sleeve and see how long it will then run the train by itself.
 

[amspire]

The IC is not a long way off the Linear Technology LTC3539/LTC3539-2

http://cds.linear.com/docs/en/datasheet/35392fc.pdf

If it was an IC with the same pinout, then pins 8, 7, and 6 would be connected together on the output.
Pin 4 would connect to the input and the indictor. Pin 1 to the other side of the inductor.
Pins 2 and 3 -ve (that is the one that looks wrong).

It is possible they have gone to LT, or even a chinese company who makes LT copies and have changed the packaging/pinout slightly for some reason.

There are differences. The ripple in burstmode from the LT device is 20mV whereas the Batteriser is 200mV but has a lower quiescent current.

[FrankBuss]

And there goes the ball game.
Typical passive device and typical active device FAIL

Right, not much left of what they claimed.

Another useful test would be testing the current limiter. I found this diagram: An AAA battery can provide 12 A peak short circuit current, and 2 A with no problem for several minutes. Another claim for their product was that it helps providing stable voltage for devices with current spikes. If this doesn't work, it would be the final nail in the coffin for the Batteroo sleeve.

[Jay_Diddy_B]

The IC is not a long way off the Linear Technology LTC3539/LTC3539-2

http://cds.linear.com/docs/en/datasheet/35392fc.pdf

snip ....

In January 2016 I built a 'fake' batteroo using the LTC3539. There are pictures here:

Link: https://www.eevblog.com/forum/blog/eevblog-751-how-to-debunk-a-product-(the-batteriser)/msg849800/#msg849800





I believe that this is a custom ic.
I would like to see it de-caped and photographed.

I used some expensive parts, the inductor I used was from Wurth MAPI series.

I think the Batteroo inductor is something like CKP20162R2M-T from Taiyo Yuden This has about 140m of DCR.

Link: http://www.digikey.ca/product-detail/en/taiyo-yuden/CKP20162R2M-T/587-2606-2-ND/2349935






@FrankBuss

Great work !! I appreciate your efforts.

Regards,

Jay_Diddy_B

[amspire]

 I agree that it looks like a custom IC, but probably to the extent of simple changes. Such as hardwiring shutdown to always on, pwm/boost mode to boost, internal feedback divider to a fixed 1.5v, and possibly lowered the current in sleep mode at the expense of increasing the low current waveform to 200mV pp.

The inductor is more likely to be something like a 2.2A Murata.

[mikeselectricstuff]

I'd be surprised if it was custom, considering that it's hardly a new requirement.
My guess is it's something from an obscure Chinese company. 

[amspire]

I'd be surprised if it was custom, considering that it's hardly a new requirement.
My guess is it's something from an obscure Chinese company.

By custom, I did not necessarily mean a new die fabricated. Many of the Asian converters are made with preset output voltages, and perhaps that is something they can preset  with lasers on the die. The other changes may involve wire bonding changes and perhaps the choice of package. It did seem to take them ages to get hold of the final converter ICs, so I don't think it was an existing part.

The slow delivery is another issue, and my suspicion there is they are doing the potting by hand in the US for the initial devices.