Chinese Power Bank How to Open an Account
Last year I did a review of Power Bank Tomo T4, where I talked about its significant shortcomings, such as a fairly fast self-discharge, the inability to fully use the battery energy and a large spurious voltage drop on the circuit elements, which reduces the efficiency of the device. I managed to finalize Power Bank and completely eliminate the described problems. If you are interested in the details of my revision, welcome to cat. Let’s start with the resulting device characteristics:
1. The consumption current in the off state is 2 μA (it was 3.5 mA).
2. The device shutdown voltage: 2.65 V. 2.8 V, depending on the load (it was 3.1 V without load and 3.4 V at 1 A load).
3. A parasitic drop on the circuit elements when operating from one battery to a maximum output current of 0.15 V (it was 1 V).
In addition, the following modifications were made:
1. During operation, Power Bank displays the voltage on the batteries instead of the output.
2. The output voltage was raised from 5.1 V to 5.25 V, which allows to obtain closer values to 5 V under load.
But, unfortunately, the laws of physics cannot be fooled, so I had to sacrifice the following functionality:
1. Separate channels for batteries. now they are all connected in parallel. Those. if you insert charged and discharged batteries into the PB immediately, the former will charge the latter until the voltage across them is equalized.
2. Protection against inserting batteries in reverse polarity. Now it is, but if the battery is incorrectly turned on, a very large current flows through it (and the circuit), which leads to heating (and possibly failure) of the battery and / or failure of the 0 Ohm protective resistor on the board.
So, consider the modifications made.
The first thing I did was solder the copper wire parallel to the springs, as they had a maximum voltage drop. If you absolutely do not want to upgrade the electronics of the device, but you want to make it better, follow this refinement, it will give a significant result. It looks like this:
To the second coil, starting from the narrow side of the spring, a copper wire of suitable section is neatly soldered from the inside, which is passed through the entire spring and soldered to its output on the board. It’s best to use the wire that connects the speaker coils to the contacts on their case, but I couldn’t find one at hand, so I took a regular multicore wire. most likely, with frequent use it will break quickly and will have to be replaced. For tinning a steel spring, the LTI-120 flux is great, just remember that it is advisable to wash it off later. Soldering to the second, and not the first turn allows you to not make changes to the contact between the spring and the negative terminal of the battery.
To get rid of high consumption in the off state, it is necessary to somehow disconnect the power of the microcontroller and the output converter. The easiest option is to install a mechanical switch. However, the entire discharge current (and this is up to 4 A) will pass through it, our precious voltage will drop on it (and the connecting wires), well, a window will have to be cut under it in the device case, which is unlikely to be done beautifully. Also, if you leave four separate channels in Power Bank, this switch must be quadrupled.
Therefore, it was decided to abandon the individual channels for the batteries and connect them in parallel, which will allow switching only one wire. Also, 4 separate charging microcircuits TP4055 are now connected in parallel and can even charge one battery with a “quadruple” current of 1600 mA. But when you insert charged and discharged batteries at the same time, the former will now charge the latter until the voltage on them is equalized. Personally, this does not cause discomfort to me, because it will not affect everyday work at all (li-ion is always simply connected in parallel where necessary).
It was decided to switch the power wire on a powerful P-channel MOSFET’s, and control it using a separate circuit using the existing power button of the device and some signal from the microcontroller (so as not to modify the device’s case). After a more detailed analysis, it was found that it is convenient to take the on / off signal of the output converter supplied to its output 4 as such a signal. When the converter should be turned on, there is a high logic level (2.8 V), when the converter should be turned off, the microcontroller feeds a logical zero there. Unfortunately, to control the P-channel fieldman, reverse voltages are needed (a low level opens it, and a high level closes it), so an inverter on a transistor will have to be included in the control circuit. The task is also complicated by the fact that it is rather difficult to find a P-channel field worker with a low channel resistance in the open state when controlling a small (2.5 V) voltage. I had Si4931DY assemblies, including two P-channel transistors with a resistance of 22 mOhm with a gate voltage of 2.5 V, which I was going to use.
And then in time I once again looked into the datasheet of the G5177C converter, where I found that in Stand-by mode it consumes less than one microampere current! Those. you can turn off only the microcontroller itself and not turn off the output converter. This means that you can use any low-power P-channel fieldman with almost any channel resistance, because microcontroller consumption current. 10 mA. As a result, the choice fell on an unknown smd transistor, labeled WA8A and soldered in due time from some kind of motherboard. The fact that this is a P-channel MOSFET was shown by a Chinese transistor tester, but it was not possible to find any documentation on this designation on the Internet. Looking ahead, I’ll say that a practical test showed that this transistor is ideally suited for its role. with a gate voltage of 2.5 V, it only drops about 2 mV at a current of 10 mA.
As a result, we got such a control scheme: There are two ways to open the “power” transistor Q2: using the Power Button, which closes the transistor’s gate to ground, and using the transistor Q1, which opens with a positive voltage that allows the output converter to work.
Thus, the general logic for turning on the device becomes as follows: the power button is pressed, the key transistor opens, power is supplied to the microcontroller. After a short time interval (while the microcontroller is initializing. the button must be pressed at this time), the microcontroller gives an enable signal to turn on the output converter, which opens transistor Q1 and ensures the operation of the device after releasing the button. When the microcontroller decides to turn off the device (and this happens if it does not detect any load on the output), it removes the signal allowing the output converter to work, Q1 and Q2 are closed and the microcontroller is de-energized. A very elegant solution that allows you to not add any additional controls or modify the device case.
As Q1, a transistor soldered from the motherboard was also used. It turned out to be PMSS3904 (marking W04). As a diode assembly D1, which protects the gate of the Q2 transistor from possible electrostatics (which can get into it through the power button), BAV99 (marking A7) was also used, which was also soldered from the motherboard. there it was used to protect the data outputs of USB ports. Resistors R5 and R3 are taken in size 1206 (since one was used during prototyping, and the track goes through the second on the printed circuit board), the rest. 0805.
For these components, a DipTrace program (which is free for ham radio enthusiasts) developed a printed circuit board measuring 17.6 x 12.7 mm (track thickness: 0.4 / 0.4 mm): Then the board was made by photoresist. For me personally, the film photoresist method turned out to be much simpler and more repeatable than LUT. The process consists of the following steps. I print the negative of the board on a transparent Lomond film for laser printers (since the printer was originally bought for LUT, so the laser), I put the Ordyl Alpha 350 photoresist on a pre-prepared (cleaned) board, apply a printed drawing and expose UV 30 W CFL Black Light lamp from a distance of 15. 20 cm for 2 minutes. The advantage of negative photoresist is that the UV-protected areas of the circuit board are washed off with an alkaline solution, while the unprotected ones remain. Thus, if the printer leaves small unprinted dots on the fill between the tracks, they will most likely be washed off and the board will not even have to be adjusted before etching. And with such relatively “thick” tracks, there are no problems at all. The result can be seen in the following photo: The assembled board looks like this: The photo shows how the board has curved edges. with a size of 17.6 x 12.7 mm, sawing it straight out with a jigsaw blade from a piece of getinax is a very difficult task, especially if you want to do it quickly, because at 23:00 hours. It is good that this will not affect its performance 🙂
All further modifications are made on the Tomo T4 board. The details on the board are numbered, so I will refer to their indicated numbers. Appearance of the original board: First of all, transistor assemblies Q5, Q6, operational amplifiers U9, U10, resistors R13, R15. R20 (we leave R14), R53, R54 are soldered. Resistors R13. R20 are voltage dividers by 2, from which the microcontroller receives voltage information on each battery. It is due to them that he determines that the batteries are dead and turns off the device. Since it was decided to combine the channels into one, these inputs also need to be connected, and the voltage divider should be changed so that the MK switches off the device at a voltage lower than 3.1 V to provide a more complete discharge of the batteries. Fortunately, MK uses these inputs only to determine the degree of battery discharge, and it gives charge control to the TP4055.
Leaving the lower resistor of the divider unchanged (33 KΩ), I selected that the resistance of the upper resistor should be 24 KΩ, so that the MK turned off the device at 2.7 V. And so that the current through the divider did not discharge the batteries when the device was turned off, the divider must also be connected after the key transistor (i.e. to the place where the voltage stabilizer power supply of the microcontroller comes from). Therefore, it is advisable to place this resistor on an additional board. according to the scheme, it is R5 connected to the “Measure” contact.
Next, it is necessary to remove the fuse F1 (since a sufficiently large voltage also drops on it) and replace it with a jumper. After that, the role of fuses will be played by resistors R61. R64 with a resistance of 0 Ohm. By the way, I saw a photo of previous versions of PB from Tomo, this fuse is not there at all.
It is also necessary to drop out the diode D2, through which the input voltage is supplied to the output converter. The general idea of this circuit solution is this: if external power is supplied to the Power Bank, then it will go through the D2 diode to the common battery connection point (after the output of the channel balancing circuits) and, if it is higher than the battery voltage, the balancing circuits will close and the output converter will be powered will only be from the input voltage. This allows you to use Power Bank as an “uninterrupted”. if there is an input voltage, the converter runs on it, the batteries are not discharged. If the input voltage disappears, the inverter continues to run on batteries. In practice, the Tomo implementation has two significant drawbacks. Firstly, a R66 resistor with a resistance of 0.5 Ohm is in front of the diode, so the voltage after the diode cannot reach any serious value even with a small load. Suppose the converter consumes 1 A, then 0.8 V will fall on the diode, 0.5 V on the resistor and somewhere else 0.5 V on the input circuit and cable. Those. the output will be only 3.6 V. This means that until the voltage on the batteries drops to 3.6 0.15 (drop in the channel balancing circuit) = 3.75 V, the converter will continue to be powered by them. And what is 3.75 V? This is the voltage of batteries that are already half discharged. But the funniest thing is that the manufacturer chose an R66 resistor of such a small size (and, accordingly, power) that at a current of just over 1 A, it heats up so that it itself is soldered out of the board.
The second significant drawback of this solution is that if you insert discharged batteries into the device and simultaneously connect the load, the device will try to charge the batteries (current up to 1.6 A) and power the output converter from an external source (1 A), which will give a total consumption current of 2.6 And more. Such a current can withstand far from every USB port. Someone may call this an advantage, but I tend to consider it a disadvantage, because Almost no USB sources with a current of more than 2A.
But the D2 diode must not be soldered because of the described problems. Since I refused the channel balancing scheme, if I leave this diode, when external power is supplied, the batteries will be charged uncontrollably through it. By the way, without a diode, we still do not lose the “uninterrupted” function. because four TP4055 will try to charge the batteries with a total current of up to 1.6 A. This current can go both to charge the batteries and to power the output converter, depending on the current consumption and degree of battery discharge. Only now the device will not be able to consume more than 1.6 A under any conditions (well, okay, I observed 450 mA consumption on the TP4055, which gives 1.8 A in total), which guarantees better compatibility with USB power supplies.
Now you need to connect all 4 batteries in parallel. This is done after the resistors R61. R64 with a resistance of 0 Ohms (which serve as fuses in this circuit) and connect the resulting point to the “output” of the soldered key Q5 (pin 1).
The next change I made was to disconnect the battery charge management from the MK, providing this process completely TP4055. To do this, the resistors R3. R6 are soldered from the board and soldered on top on the U1. U4 microcircuits between terminals 2 and 5. If this is not done, at the end of the battery charge (when TP4055 signals this), the MK turns off the charging, and after a few seconds turns it on again. TP4055 again indicates that the batteries are charged, the MK turns off charging and the process repeats. Now only TP4055 will be responsible for the charge, which quite well copes with the full charge cycle itself.
Next, I decided to make the Power Bank not display the output voltage of the converter (which is very good at 5.1 V) when working on the load, but the voltage on the batteries, because it is much more informative. To do this, I tore the track from the right (on the photo of the printed circuit board) output of the resistor R25 and connected the resistor to the input of the power stabilizer MK U5. Fortunately, MK uses this information only for display on the screen, and the modification did not affect the operation of the device.
After that I soldered the missing capacitors C23. C25 (I took 10 microfarads of frame size 1206), parallel to C23 I soldered the Schottky protection diode (in the opposite direction, cathode to the plus of the power supply), and I corrected the 1.2 and 2.4 Ohm correction resistors on top of R41 and R42 so that the device It showed the output current more accurately (the ratings were selected empirically). Also, a resistor with a resistance of 75 KOhm was soldered on top of R52 to raise the output voltage of the converter by 0.15 V. This allows us to obtain output values closer to 5 V even at high load currents. To increase the stability of operation, the MK also soldered a 10 μF capacitor of size 0805 parallel to its power supply (parallel to capacitor C15).
Now it’s time to install and connect an additional board. There is a place for it only on the back side of the device board, in a small compartment between the springs and the rear wall of the housing. To fix the board, I bent a single-core copper wire of a suitable diameter in the shape of the letter T, then soldered a flat face to the lower track (common wire) of the developed board. I soldered the remaining protruding wire to the contact of the extreme spring (in the case I had to cut an opening for this wire). To connect the wires to the board, I drilled a small hole near the USB connectors to a suitable diameter so that it could lead wires to the other side of the board through it. In the photo it looks like this: Also here you can see the wire that goes from the additional board to the control button (the only wire that connects from this side of the board). Now two more explanatory photos. Details to be removed from the main board: Points that need to be connected (blue lines), tracks that need to be cut (red crosses) and connection points of the additional board (yellow arrows): The result should be something similar to this: We check. Powered by an external unit, the load current is 2A. At 2.8 V. reliable operation over time. Not bad: We assemble the device and check again. Now from the batteries:
For me, the completion goal has been fully achieved. the device provides a decent battery discharge (up to 2.7 V) and an excellent quiescent current when turned off. 2 μA (by the way, half goes through the Schottky protection diode). Also, the indication has become more informative, and the charge current has become independent of the number of batteries.
Would I buy this device now if I knew all this? Most likely, still no. Quite a lot of time was spent on the revision, in my opinion, it is better to look for a better finished solution, even if it will be somewhat more expensive. Nevertheless, now I have a Power Bank ready to give my old 18650 a “second life”. according to tests they have about 1 Ah at a discharge with a current of 1 A, i.e. Using these 4 pieces, you can charge the phone once.