Broken microcontroller: what to do next

[maxmalkav]

I'm opening this thread to get some advice (or at least receive some comfort).

>1 year ago I bought a cheap DIY keyboard kit but I only assembled it recently. The PCB had the microcontroller, resistors and so already soldered, so I just threw some switches and LEDs at it. The resulting board was working beautifully .. for 2 days. Out of the blue the number row got stuck: all LEDs on and keys not working at all.

After resetting several times the keyboard (the firmware support a combination of keys to do so) I was not getting any progress. LEDs were working "fine" (in some the backlight modes, such as "wave", they were working fine), but keys were pretty much dead, so I proceeded to check the internals.

After tinkering a bit with my multimeter I saw that the broken row was receiving roughly 0.7V instead of the 4.9 ~ 5V that other functional rows were getting. The microcontroller is kind of small but I managed to check the continuity between the pin and the attached row (continuity seemed alright), and the voltage at the MCU pin was also 0.7 sad Volts.

My knowledge about microcontrollers is pretty rudimentary, but I'd say the pin attached to the number row just died and I cannot do much about it. The microcontroller is a fairly mysterious Vision VS11K06A. There are several cheap Chinese mounting it, but the technical info about it on the Internet is virtually 0. Moreover is quite small, so even if I got a spare MCU, it would be tricky for me to replace it.

Can I declare the keyboard as "dead" or am I missing something I could still test? Thanks

[kakan]

Take some high quality pictures of the pcb and we may figure out whats wrong with the board

[vvp]

Try to measure the resistance between the row and the ground when the keyboard is not connected (powered off). Make sure you place the negative probe of the ohm-meter on the ground (notice that many ohm-meters have negative pole on the red probe). You should measure values in the range of 10 MΩ or bigger. If the value is lower then check whether you do not see some weak connection between the row and the ground. You can disconnect the row wire at some places to better localize the wrong part. Maybe you made a week connection there in the range of few hundreds of ohms and the microcontroller cannot power it (it cannot supply enough current). If you find the wrong connection and fix it then you have a chance it will still work. Often MCU pins survive some over-current. Do not run the keyboard powered unnecesarily till you do not fix it.

[maxmalkav]

Hi! I really appreciate the advice

Before reading your replies I tried to resolder the troublesome pins in the MCU (time ago I practiced this with small SMD stuff) Funny fact: after this the troublesome row started to work for a minute and then started to degrade and die again after this.

Now I have another small problem: the usb cable to the PCB broke and I need to resolver it. In the meantime I’ll check the impedance between the tow and ground.

I’ll try to post some pictures of the PCB, I’ll aim for the parts that are potentially related to this problem.

Thanks again guys !

[vvp]

If soldering MCU pins fixed the problem for a while then it is likely you have badly soldered MCU to PCB or the MCU is broken. The thermal stress may have temporarily (dis)connect something in the MCU package. The chance that it is somewhere else around the broken row traces on the PCB is lower now. But still possible.

If it is a QFP package then try to reflow all the MCU pins using drag soldering.

If it is a QFN package without a lot of thermal vias on the PCB then you need a hot air station. If the QFN footprint has a lot of thermal vias then you can do this step with an iron as well. Try to reflow the MCU. If it does not help then you can try to de-solder it, clean up and solder it again. If that does not help it is time to replace the MCU (only an option when you can buy it and get the firmware and the programmer for the MCU).

Search youtube as for as how to reflow an MCU (drag soldering, hot air, soldering QFN through thermal vias). There is a ton of quite a good videos there about it.

[maxmalkav]

The MCU is QFP, I resoldered one side of the chip using the drag soldering technique, I will go back to it when I have the USB connector working again and see what happens.

The MCU seems pretty obscure, so the chances to find a replacement are almost 0, I fear.

Thanks again for the advice!

[abrahamstechnology]

Make sure you are using leaded solder from a reputable brand. The politically-correct whacko lead-free solder barely works at all.
Also, some flux may be conductive, so thoroughly clean the area with alcohol. Flux residue was causing some issues with my Alps64 until I cleaned it off.

[DMA]

drag-soldering QFN? That's hardcore.
It _could_ be a dead GPIO, but it's likely a short somewhere in the PCB. Dead GPIO is usually one of the rails (usually ground because ground it easier to short to). 0.7V is usually a short to the ground somewhere far away from the MCU with a GPIO struggling but still alive.