Comments: CKCS PB0A

[utterances-bot]

CKCS PB0A - DONE.LAND

I2C-Enabled Programmable 1S LiIon Charger and Discharger

https://done.land/components/power/powersupplies/battery/chargers/charge-discharge/ip5306/pb0a/

[314rs]

I had some trouble with the i2c. I got very flaky results. Then i removed the smaller IC (on the left, 6 pins). Now it works like a charm.
Do you know, what this chip actually is? My guess is, its some controller for the onboard LED. Mine is permanently off now.

[TobiasPSP]

Great find - yes I believe this is the led controller that enables the rgb effects.

I had flaky results at first, too, but shorter i2c wiring and stronger pull resistors fixed it.

I did notice that some dev boards worked more reliably than others. Maybe you found the culprit. Could be this IC messes up the i2c signals since the led signals share the same pins.

In other designs, for i2c to work, you must remove the led. With this charger, removing the rgb controller ic is probably the safest way for reliable i2c.

May I ask how you run the setup (which mcu, with level shifter or how powered)? Thx!

[314rs]

PB0A Test

I did some further testing in different configurations. I used a esp32c3 supermicro board. Connections to the PB0A are as shown in your pictures on the PB0A Test Setup and PB0A Production Setup pages.

Tests

Variables:

• power: The PB0A Board is powered either from the esp32c3s 3v3 pin or with a battery
• i2c pullup: The esp32c3s internal pullups are enabled allways, in some test i added external 2k2 pullup resistors
• LED IC: The LED IC is either connected (stock) or removed
• logic level shifter: with or without logic level shifter. connected as shown on the PB0A Production Setup page

With LED IC

Test 1.1

feature	value
power	3v3 from esp
i2c pullup	2k2 in hardware
LED IC	yes
logic level shifter	no

flaky results:

=== I2C Bus Scan ===
Scanning addresses 0x03 to 0x77...

Device found at 0x17
Device found at 0x1B
Device found at 0x26
Device found at 0x2F
Device found at 0x45
Device found at 0x46
Device found at 0x47
Device found at 0x49
Device found at 0x4A
Device found at 0x4B
Device found at 0x4D
Device found at 0x4E
Device found at 0x4F
Device found at 0x55
Device found at 0x56
Device found at 0x57
Device found at 0x59
Device found at 0x5A
Device found at 0x5B
Device found at 0x5D
Device found at 0x5E
Device found at 0x5F
Device found at 0x65
Device found at 0x66
Device found at 0x67
Device found at 0x6A
Device found at 0x6B
Device found at 0x6D
Device found at 0x6E
Device found at 0x6F
Device found at 0x75
Device found at 0x76
Device found at 0x77

Scan complete. Found 33 device(s)
===================

IP5306 at 0x75: ONLINE

=== IP5306 Register Dump ===
0x00 SYS_0   : 0x00 (0000 0000)
0x01 SYS_1   : 0x3F (0011 1111)
0x02 SYS_2   : 0x3F (0011 1111)
0x20 CHG_0   : 0x96 (1001 0110)
0x21 CHG_1   : 0xAA (1010 1010)
0x22 CHG_2   : 0xAA (1010 1010)
0x23 CHG_3   : 0xAA (1010 1010)
0x24 CHG_4   : FAIL
0x70 READ_0  : FAIL
0x71 READ_1  : 0x00 (0000 0000)
0x72 READ_2  : FAIL
0x77 READ_3  : FAIL
0x78 READ_4  : 0x12 (0001 0010)
===========================

this is pretty representative of the flaky results I've seen with this setup.

Test 1.2

feature	value
power	battery
i2c pullup	2k2 in hardware
LED IC	yes
logic level shifter	no

=== I2C Bus Scan ===
Scanning addresses 0x03 to 0x77...

Device found at 0x0C
Device found at 0x0D
Device found at 0x0E
Device found at 0x0F
Device found at 0x10
Device found at 0x11
Device found at 0x12
Device found at 0x13
Device found at 0x14
Device found at 0x15
Device found at 0x16
Device found at 0x17
Device found at 0x18
Device found at 0x19
Device found at 0x1A
Device found at 0x1B
Device found at 0x1C
Device found at 0x1D
Device found at 0x1E
Device found at 0x1F
Device found at 0x20
Device found at 0x21
Device found at 0x22
Device found at 0x23
Device found at 0x24
Device found at 0x25
Device found at 0x26
Device found at 0x27
Device found at 0x28
Device found at 0x29
Device found at 0x2A
Device found at 0x2B
Device found at 0x2C
Device found at 0x2D
Device found at 0x2E
Device found at 0x2F
Device found at 0x30
Device found at 0x31
Device found at 0x32
Device found at 0x33
Device found at 0x34
Device found at 0x35
Device found at 0x36
Device found at 0x37
Device found at 0x38
Device found at 0x39
Device found at 0x3A
Device found at 0x3B
Device found at 0x3C
Device found at 0x3D

Scan complete. Found 50 device(s)
===================

IP5306 at 0x75: OFFLINE
Device not responding

Test 1.3

feature	value
power	battery
i2c pullup	2k2 in hardware on hv side
LED IC	yes
logic level shifter	yes

Device found at 0x15
Device found at 0x16

Scan complete. Found 2 device(s)
===================

IP5306 at 0x75: OFFLINE
Device not responding

flaky but far less false positive addresses. no false positives most of the time

Without LED IC

Test 2.1

feature	value
power	battery
i2c pullup	2k2 in hardware on hv side
LED IC	no
logic level shifter	yes

=== I2C Bus Scan ===
Scanning addresses 0x03 to 0x77...

Device found at 0x75

Scan complete. Found 1 device(s)
===================

IP5306 at 0x75: ONLINE

=== IP5306 Register Dump ===
0x00 SYS_0   : 0x35 (0011 0101)
0x01 SYS_1   : 0x1D (0001 1101)
0x02 SYS_2   : 0x64 (0110 0100)
0x20 CHG_0   : 0x01 (0000 0001)
0x21 CHG_1   : 0x89 (1000 1001)
0x22 CHG_2   : 0x02 (0000 0010)
0x23 CHG_3   : 0xBF (1011 1111)
0x24 CHG_4   : 0xD6 (1101 0110)
0x70 READ_0  : 0x17 (0001 0111)
0x71 READ_1  : 0x00 (0000 0000)
0x72 READ_2  : 0x04 (0000 0100)
0x77 READ_3  : 0x80 (1000 0000)
0x78 READ_4  : 0xC0 (1100 0000)
===========================

Test 2.2

feature	value
power	battery
i2c pullup	esp32 internal in software
LED IC	no
logic level shifter	no

=== I2C Bus Scan ===
Scanning addresses 0x03 to 0x77...

Device found at 0x75

Scan complete. Found 1 device(s)
===================

IP5306 at 0x75: ONLINE

=== IP5306 Register Dump ===
0x00 SYS_0   : 0x35 (0011 0101)
0x01 SYS_1   : 0x1D (0001 1101)
0x02 SYS_2   : 0x64 (0110 0100)
0x20 CHG_0   : 0x01 (0000 0001)
0x21 CHG_1   : 0x89 (1000 1001)
0x22 CHG_2   : 0x02 (0000 0010)
0x23 CHG_3   : 0xBF (1011 1111)
0x24 CHG_4   : 0xD6 (1101 0110)
0x70 READ_0  : 0x17 (0001 0111)
0x71 READ_1  : 0x00 (0000 0000)
0x72 READ_2  : 0x04 (0000 0100)
0x77 READ_3  : 0x81 (1000 0001)
0x78 READ_4  : 0xC0 (1100 0000)
===========================

Test 2.3

feature	value
power	esp32 3v3
i2c pullup	esp32 internal in software
LED IC	no
logic level shifter	no

=== I2C Bus Scan ===
Scanning addresses 0x03 to 0x77...

Device found at 0x75

Scan complete. Found 1 device(s)
===================

IP5306 at 0x75: ONLINE

=== IP5306 Register Dump ===
0x00 SYS_0   : 0x35 (0011 0101)
0x01 SYS_1   : 0x1D (0001 1101)
0x02 SYS_2   : 0x64 (0110 0100)
0x20 CHG_0   : 0x01 (0000 0001)
0x21 CHG_1   : 0x89 (1000 1001)
0x22 CHG_2   : 0x02 (0000 0010)
0x23 CHG_3   : 0xBF (1011 1111)
0x24 CHG_4   : 0xD6 (1101 0110)
0x70 READ_0  : 0x17 (0001 0111)
0x71 READ_1  : 0x00 (0000 0000)
0x72 READ_2  : 0x04 (0000 0100)
0x77 READ_3  : 0x80 (1000 0000)
0x78 READ_4  : 0xF0 (1111 0000)
===========================

result

All test setups without the led ic are reliable.

[TobiasPSP]

Awesome - very helpful results! Thank you so much for your efforts.

In my tests, I got reliable results with a classic esp32, powering the board from the esp32 3.3v pin using 2k pullups. With other mcus (including esp32 c3 supermini), results were flaky when my i2c lines were long (simple dupont wires), improving with short wires.

So I‘d conclude - with your results in mind - that the led ic is distorting the i2c signals somehow. It was evident that - with the led chip in place - there are significant differences depending on the mcu type used. Devkit V2 (classic esp32 board) seems much more robustly interpreting i2c signals than c3 or s2.

If only someone with an oscilloscope could shed more light on the nature of distortion.

That said, I don’t see an issue removing the led chip since in most i2c cases, the onboard led isn’t useful anyway. How did you remove the chip?

[TobiasPSP]

One of the things on my agenda btw is trying an ATTiny since that’s pretty much the only mcu I know that supports the entire LiIon voltage range and can be used directly without logic level shifter regardless of battery soc.

Another useful idea would be building an adapter using pogo pins that can be used to program the board on the fly. If used in a powerbank-like scenario where there is always power, it could be easily programmed to appropriate charging currents, and if programmed to be always-on, it keeps the new settings even when removing the adapter.

The risk in this scenario of course is that if for any reason the powerbank ever gets completely empty, the charger forgets its settings and returns to defaults since it has no flash.

[314rs]

I briefly checked the stock PB0A with an oscilloscope (i. e. no esp connected).
The LED controller talks I2C to the IP5306.
Interestingly, the connection to the LED also shows some serial data. Maybe the LED is adressable (like neopixel/ws2812).

Will post more detailed results when I have time.

[TobiasPSP]

Great finds - please keep us updated. Investigating this board apparently yields a lot of surprising insights. If you have a chance, maybe you can compare signals on a board with and without the led chip to find out how it interferes. If it is, too, communicating via I2C, then most likely we‘ll see conflicting communication?

[cuvoodoo]

There is an alternative (newer?) version of the CKCS PB0A. You can recognize it with the smaller SOT-23-6 chip in the same direction as the IP5306. The layout is also different. I reversed it, and L2/LED2 is tied to BAT+. The IP5306 also does not seem to be the I2C version anymore. Even after lifting the L2 pad and connecting to it, the I2C scan (using an ESP32 with level shifter) does not detect it.

[TobiasPSP]

Good catch - I just discovered this version, too. It is offered cheaper.

Do you own one and have investigated I2C?

i wonder if it is a fake clone since it doesn’t make any sense to use a non-I2C-enabled ip5306 on this board while still exposing a „data“ pin, and there are already plenty of simple ip5306-boards w/o I2C.

Fortunately the new version is easily recognizable, and I have stocked 50 old ones to play with. That said, the old I2C-enabled version is still for sale, so no worries.

[cuvoodoo]

Yes, I do own these boards and did investigate I2C. I also desoldered the chips to test them outside the board. They don't react to I2C requested or scans.
The data pins just correspond to the LED light status output (charging/charged). There are not I2C data lines anymore.