DealExtreme Universal 5V 3A 3-Port USB AC Charger w/ EU Plug – White (100~240V) SKU: 378490 2016.01.27 at 16:26

This post is going to be in English, as, I suppose, it will be interesting for international readers. The first thing what made me write this article was a strange behavior oh my Prestigio Multipad tablet and HTC Sensation XE phone, when connected to this charger. The link to the charger is here and lets put it to the test.

IMAG0291

IMAG0292

By the way, I couldn’t find it in my purchase history :D, strange, as I really purchased it some time ago. Ok, so lets start. Now, why could it make my device go crazy when charging (touchscreen senses touch in wrong position)? First, lets take a look how the voltage looks like on the outputs, no load:

Newfile1

Uuuuu… No load and we see that we have quite bad voltage on output. Bad thing is not even those 5,56V, but the pulses ! Shouldn’t be so ! Lets filter the DC and see how it looks like:

Newfile2

Wow ! Its a saw, not nice, really not nice. This might be interfering with tablets power circuit and as a result I have the incorrect touchscreen work. Now with 15Ω load, so it should have about 0,333A running through:

Newfile3

You got to be kidding me… Its even worse ! And this is at about 1/10 of declared load that this charger can hold (0,333A instead of 3A). And DC filtered:

Newfile4

Sorry for the double graph, just caught it that way, just imagine its single. Still, the saw tooth is a bad thing. Well… Sensitive devices will have some strange interference due to this. Chm, lets try to make it a working charger :).

First – disassemble the charger. Its easy, pop up the top cover and the internal PCB slides out.

IMAG0293

The PCB from both sides:

IMAG0294

IMAG0295

Next, understanding the schematics. The main part of this power supply is DK112 chip, though it seems to be too Chinese, not even a proper datasheet. Anyway, found a picture, with the schematics for similar power supply.

1088_192815426_repaste-detail

Its very basic, mains power goes through diode bridge MB10F (0,5A, 1000V) and from AC becomes DC voltage. Then couple of electrolytic capacitors, to smooth the DC even more. At this point it should be around 325VDC. The caps are 6,8 μF x 400V, so not even near to recommended 33 μF. Then the working horse of this schematics, DK112 chip by „Dongke Chip” (its different there, but pins are the same). Basically it incorporates the control circuit and a transistor. The pinout:

1 – GND
2 – GND
3 – Feedback
4 – Vcc
5, 6, 7 and 8 – Collector

The chip, according to feedback (PC817 optocoupler), through transistor, controls the transformer, supplying power to the power capacitor. In our case its two capacitors in parallel, each 470 μF x 10V, so total 940 μF (on schematics above its EC4 capacitor). Then it should have the filtering coil and output capacitor. But it does not. And this should be the reason of that ugly output! Ok, what else is on the board… I can see the feedback circuit, also two (!) diodes on output, each can hold 3A, connected in parallel, so 6A in total. Not bad :). Well, now lets try to improve the charger. Soldered in the couple 47 μF x 10 V tantalum capacitor on the output (watch the polarity). Why 47 μF ? Because I had one and size B fits perfectly.

IMAG0296

Tantalum caps has a very low ESR, which will help at high frequencies. But this has very small impact on the output. Still, it need whole output circuit. Problem is, that this additional circuit cannot fit inside the box :). The easiest way to fix the charger – use the LDO (Low DropOut Regulator). For example – NCP59300 by ON Semiconductors (would be great in D²PAK3 case). Small, stuffed with protections, can hold up to 3A and the voltage drop would be only 300 mV. This is good for us, as the output of charger now is 5,560 V it will be OK for regutator to make 5V. If it would be 5,250 V on the output – the regulator wouldn’t be able to supply 5V, as the drop out voltage makes it lower (5,250V – 0,300V = 4,95V). Now we are in the safe zone for charging (5V) and still have the space to dropout (5,560V – 0,300V = 5,26 V). And at 1,5A load (I gues 90 % of work time) only 175 mV dropout.

2016.02.04

Today I received the NCP59301-50. After some thinking and trying to fit the NCP inside the charger housing I decided to make some changes to the original schematics. Logic is simple – NCP requires only 1 μF capacitor on input and 47 μF capacitor on output. This mean, that even without those two big capacitors it should regulate the output voltage. I removes those two big caps (they will be replaced by several 47 μF tantalum caps, which also will be useful for the feedback circuit of DK112) and now I have the space for regulator. In case if regulator will work at maximum amperage it should heat. Naturally, that will not be all the time, but still, we have to remember that. One of the cooling ways would be attaching the regulator to some big metal part, which would take the heat. The only part inside, that could do the job is a transformer. Well, why not, cut a bit of the yellow tape out and now have a magnet core, which basically is made of metal powder. A little drop of heat conductive glue – and regulator stuck to the transformer.

IMAG0326 IMAG0327

Regulator is protected from overheating and this way I will have even more protection, in case if transformer gets hot – it will heat the regulator and it will shut down.

IMAG0328

Here you can clearly see, the regulator. I left the pad outside, in case if it will generate a lot of heat – it will be possible to attach some aluminum radiator or something (also, glue or soldering). Mine version looks like this:

IMAG0350 IMAG0351 IMAG0352 IMAG0353

Blue wire – input, red wire output, and a little piece of paper clip – ground. Assembled:

IMAG0354

Everything fits perfectly. Next step – testing. Same conditions, no load:

Newfile1

Almost perfect straight line, as it should be. Now under same load:

Newfile3

There are little ripples, but or sure, not as big as in original version. Same, without DC:

Newfile2

Here we can see the maximum voltage is in acceptable range, but due to input fluctuations output also has a small level ripples, about 250 mV. Before it was 350 mV. This seems not too good, reason is, that when input falls below certain level – regulator is not able to maintain the output at 5V. As a result, we have the upper part of graph smooth and what is below – appears due to lack of voltage on input.

Newfile1

Input and output if regulator. And without DC voltage:

Newfile2

Here, we can clearly see, what was I talking about – when input (blue line) is higher than threshold, output is very nice and smooth (yellow line). When input drops down – at some level output is also following. Well, this upgrade can be working without any interference, due to nicer output, but still, its not 100 % complete. Pity I have an oscilloscope, with it I can see that my job is not finished :D. Simplest thing to correct this – change the feedback voltage divider, so the output of transformer (and the input of regulator) would be 0,5-1 V bigger. Here’s the closeup of the feedback part:

It should be simple, when voltage goes higher than needed, LED is ON, it turns on the optocoupler transistor, which turns off the DK112. When voltage goes low – DK112 is ON again. So we have that teeth view on the oscilloscope. Ok, the LED is turned on by a specific device in SOT23. According to my measurements it should be programmable reference, a type of diode with control electrode. Judging from the marking 431, it can be TL431. Lets try to make the voltage bigger, lets say 6V ? And why 6V ? Just so, we can go up to 18V (yeah, and the caps are only 10V), but more difference between input and output of NCP59301 – more heat it will generate. So chosen relatively save margin. Simulation of the voltage divider for 431:

schem

On the right side – original configuration, generates 2,439 V (according to datasheet reference voltage is 2,495V), on the left side one resistor is smaller, 2 kΩ, and to reach the desired 2,439V input voltage should be 5,945V – perfect, the desired 6V. Now lets try and see if something will explode :).

Newfile2

Nothing exploded, strange :D, but the view is perfect. As per calculation, the input voltage is higher, 5,960V, close enough to the calculated 5,945V.Newfile3

Closeup of the ripples, we can see the blue line (output) is flat, as it should be. At this point I could say I finished my work here, but… Output is only 4,440V, instead of 5V. Chm, voltage can go down if load is too high… Heres the view without load:Newfile4

Thought so – without load we have a 5,240V, what would be perfect for a charger. But need to make it to hold the same voltage under load.

2016.02.11

RTFM. No, seriously, RTFM, because I didn’t, ant the reason for the lower voltage is output capacitor, which should be (as mention in the datasheet) 47 μF ceramic type instead of 47 μF tantalum, that I used. Today I got the 47 μF x 6,3V ceramic capacitor in 0805 housing and here are the results, under 0,3 A load:

Newfile1

Under 0,6 A load:
Newfile2

Under 2,5 A load:Newfile3

Impressive 🙂 Perfect blue line, at maximum load holds 4,680 V, without any fluctuation. Now I can say I did everything :). Final configuration:

IMAG0368 IMAG0369

And the last remarks. On the DX page there is a notice about charger features:

  1. EU Plug Standard Quality – well, its EU plug, though how it defines quality I don’t know.
  2. Built in resistance chip – no, there  is no built in resistance chip. All the resistors are standard SMD resistors, no dedicated resistance chip. They say this chip should increase the charging stability, decrease the damage to digital device. Well, in original version the damage to digital device existed 😀 and as there is no some mystical resistance chip – maybe that is the reason ? And the output stability was awful…
  3. Protective circuit to prevent overcharging – this is not true, all the charging and protection from overcharging is done by internal circuit in the phone, tablet or other device. And inside of the charger there is no any circuit which would do the mentioned things. Well, now, with new regulator it has some protective features:

Fast Transient Response
High precision regulator, stable with Ceramic Output Capacitor
Current Limit, Reverse Current and Thermal Shutdown Protection

Leave a Reply

*