Soshine T2 USB charger for NiMH, LiIon and LiFePO - Short Review / Test

(First Blog post after 4 years of absence - wow! Let's see if I fill it with more electronics and µC stuff in the near future! :) )

Usage

First tests show that the charger works well for NiMH and LiIon cells. For a few seconds after plugging in new batteries, the chemistry is checked based upon the cell voltage.

The display shows LiIon, which is a bit terrifying - will it blow up the 1.2V battery with 4.2V? But after short time, no chemistry is shown on the LC display which means the NiMH charging program with -dV detection is used. During this detection phase you can push the buttons below the LCD to switch between LiIon and LiFePO charging parameters.

The charger monitors the cell voltage which it shows on the display during charging. And of course the current: It can display the mAh charged into the battery, also the time it needed and a fourth position which just shows "----" for me; during charging LiIons or LiFePO, the percentage is shown here.

A LiIon cell gets detected correctly, too.









Technical details

The build quality is actually quite nice, I don't see bad soldering, everything seems clean. The negative contacts for the batteries are sliding smooth. The display is easy to read.

Flat top cells can make contact with the positive tabs due to a small bump. The charger delivers 500mA to each cell when charged in pairs; 1A for a single cell. Which seems to be one of the major differences to a Soshine H2 charger. The NiMH charging program uses a constant current and switches off either with a -dV detection or a timer based time-out.

For LiIon I currently can only measure the load on the USB port. It is 0.63A at 4.8V. A few millivolts later, the load decreases to 0.55A at 4.8V, the battery has 3.9V and the charger shows 70 percent. It seems that the current is jumping around between 2.25W and 2.75W on the USB port most of the times. Overall, slowly decreasing until dropping fast in the CV phase. (This seems to be usual for USB chargers, by the way.)

This is no real problem, but it unnecessarily prolongs the charging process. The charger uses a 'good enough' approximation of a CC-CV charging scheme. After the charger shows 4.2V /FULL for the LiIon battery, the input current drops down to 0.7-1.0W. Removing the battery then lets the current drop down to 30mW. The Soshine T2 silently continues with the CV charging without increasing the counters. It silently charges a little more; this is also the behaviour of the charger when it restarts due to the cell dropping below the threshold voltage for recharging. After a short time, the current decreases to 30mW on the USB port with the LiIon battery still inserted in the Soshine T2. This will be the idle power from the microcontroller and additional circuits.

Further measurements: A 14500 LiIon cell gets charged up to 4.192V (cell voltage without load). At that time I could measure -5mA load while the charger shows "full". It switches off charging and immediately draws 5mA from the cell afterwards. As it will restart charging if the voltage drops below a certain threshold the cell doesn't get drained empty, thus no problem here. Just remember to take out the batteries when unplugging the Soshine T2. In conclusion, it works properly for LiIons. :)

Update: Here are results from a deeper look at this charger with real current and voltage plots.



Insights


The used Holtek µC in a SSOP16 housing offers plenty of IOs, 12 bit ADC, LCD driver, an internal 12 MHz oscillator, 128 Bytes of RAM; in short it is cheap and powerful enough for controlling a Buck topology for constant current and constant voltage. Though as can be seen, only one inductor is used for two otherwise independant channels and there are two 56 Ohm high-power resistors in place. It'll step down by limiting the current and then burn energy with resistors and transistors.

The LC display has its own controller or even just a shift register (sealed with a tar blob), by the way. It is connected via 6 lines, from which 3 seem to be power supply as they lead to thicker lanes on the PCB.

Bottom line

LiIon tests are still going on, but so far they are as well as the NiMH tests very promising. The charger does what I need it for - it properly fast charges my AAA batteries for the mouse at the computer at work.

Prominent parts on the Soshine T2 PCB:
- Holtek HT46R066 8-Bit MCU w/ enhanced AD
- Holtek 46R066 Datasheet
(Holtek also has an AppNote for a Holtek µC based charger - the Soshine T2 seems like an improved version of this.)
- APM9435 SOIC8 P-Channel MOSFET: -30V, -5.3A, less than 90mOhm @ Vgs=4.5V
- Diodes, SOT23 transistors, ceramic capacitors, heaps of SMD resistors.

Backside of the PCB:
- Display, Inductor/Coil, electrolyte capacitors, MicroUSB connector.


So far for my fast first-look at this charger.

Worst blog ever

This is somewhat awkward - I never find the time to post something here. I'm still around, but there aren't many news anyways. Too busy with the job... I hope you all have a nice holiday season soon!

Cheers
Koepi

Longer time without update again

Again two months passed by without real news from me. On the Xvid-side, nothing happened for some time - so there is no need for a new build. The work is taking up plenty of time of course, as usual. So there is no time for playing around with code.

Anyhow, I'm still alive. Should there something happen like a new Xvid release, I'll be right there. Since I don't have a 64 Bit Windows and according build system around, I still can't provide a 64 Bits build. It's nice that you ask for it though!

Oopsi - long time without news

Erm *caugh* Well... Hm. I don't know what to say, but I really more or less forgot about my blog here for a while. Sorry!

I have no real news anyhow right now. The Xvid-release 1.2.1 is out - you probably realized that without me writing about that. If you want to, you can try the build with VAQ-patch applied as it gave a great result in my limited tests.

Job-wise, I am lucky here in the south of germany. The whole area (not just my colleagues) is nice. It's a tourism region, so it has to be. :-)

Ubuntu 8.04 on the EeePC

After a kernel-update for Ubuntu 7.10 yesterday all my EeePC-specific modifications didn't work anymore. Things like overclocking to 900 MHz on-demand were lost, wifi as well as the control of the sound volume via function keys. After fiddling around with the init-scripts, ACPI-events and so on I decided it's not worth the effort. So this morning I finally started a fresh installation of Ubuntu 8.04, that version with long-term-support.

Since 8.04 came out, a lot of people came up with very useful tools and scripts to make it easier to customize the Ubuntu-installation for the EeePC. First, I had to create a bootable USB-Stick with the installation-CD-image. I found the project UNetbootin on Sourceforge. The developers created a tool which takes an image of an installation-CD for e.g. Ubuntu, Fedora, Mandriva or FreeBSD and creates a readily bootable USB-Stick from that. It is available as sourcecode, but packages/binaries for Linux and Windows are downlodable on the project-site as well.

The installation from that USB-Stick went straight-forward. I connected my EeePC via cable to the net so the installer could download language-packs and other things. After the first reboot I got offered more than 180 updated packages, resulting in another 200 MB which got downloaded and installed. You might want to delete the contents of the cache-folder in /var/apt to save some space on the tiny solid-state-disk.

The network connection is also useful for the RiceeeyTweak.sh-script. It downloads all the necessary sources, scripts and packages for overclocking the processor on demand, for the wireless-driver and it also adopts the font sizes, makes changes for reducing write-operations to the SSD and more.

The customization for the EeePC for Ubuntu 8.04 is as simple as this: Start the terminal. Type

mkdir temp
cd temp
wget http://eee.ricey.co.uk/files/eee/RiceeeyTweak.sh
chmod +x ./RiceeeyTweak.sh
./RiceeeyTweak.sh

Watch the script running, asking you for the superuser-password every now and then. It only takes a few minutes and you're done.

I did another check with that Pinnacle 72e USB-DVB-dongle. It didn't work out of the box. I had to fetch the updated v4l-sources and compile and install the new driver from linuxtv-org. I used the SVN-snapshot. Simply unpack the tarball and compile and install the drivers. Keep the file v4l/Module.symvers, you'll need it to build a working webcam-driver again later. The firmware which is already installed by the default Ubuntu-kernel doesn't work with the dongle though. I used the firmware from this old package to replace the non-working installed firmware by typing in sudo cp firmware/dvb-usb-dib0700-1.10.fw /lib/firmware/linux-2.6.24-19-generic.

Then I had to install subversion: sudo apt-get install subversion . The webcam-driver-sources can then be fetched via svn checkout svn://svn.berlios.de/linux-uvc/linux-uvc/trunk. Copy the file Modules.symvers to the trunk-directory. A simple make and make install then lead to the installation of a working driver for the webcam.