Percul gokart laptimer repair and teardown

[adinsen]

My son recently bought an old Birel MX32 kart. It came with a Percul laptimer which I'm trying to resurrect. The timer is based on an H8/538 processor with an external EEPROM and RAM. It has a four cell NiMH battery managed by an onboard MAX713 battery charging IC. A 7W74 single flip flop is controlled by a power switch. This flip flop works, but there's no power on the rest of the circuit when it toggles. I'm trying to trace the circuit from there. There's a separate MAX708 reset generator on the board, which is not recieving any power.

The board has some corrosion from water which has ingressed through the charging port. The battery was more or less dead so I've cut the leads and is powering the thing with 4.8V from my lab supply via them. I have attempted ot remove as much grit as possible.

I haven't been able to obtain a schematic or other service information from Percul, which today, is a contract manufactorer near Matsuyama, Japan, but they're not responding to my e-mail. Their contact form on their website does not work. They do show a picture and a brief description of their laptimers as discontinued products on their website https://percul.co.jp/service/discontinued/

I've resurrected lots of circuits before even without schematics, but I've always tried to avoid compact SMD boards like this as tracing and identifying connections is difficult. I feel stuck now. I'm poking around looking for dead capacitors and short circuits, but as I don't know what I'm looking for, it's hard. Should I give up?

/Anders

[Gyro]

Looking at the power supply end of the board, I'm not surprised that you're not getting power to the MAX708. There is corrosion on both sides of the board so I assume it will have attacked the vias too. If there is any suggestion of leakage from the NiMH battery, you should wash the affected end of the board with a dilute acid - citric acid, white vinegar, lemon juice etc. to neutralize the alkaline residue. Then scrub the board in water until you have removed as much of the corrosion as possible (from the photos, you have a long way to go) followed by an IPA wash. You need to get the board clean (back to green solder resist) before you can start following tracks (including the ones under components), checking continuity of vias, re-making solder joints etc. Finish off with another IPA wash and dry. Edit: the buttons may also need replacement if they have become contaminated.

Sorry, there is no short cut. At least the damage seems to be restricted to the lower tech end of the board, so you have a reasonable chance without schematics.

[jchw4]

Quoting myself:

That's some serious corrosion here and it looks like an exact reason for the fault.

So it could be either a leakage (i.e. a new current path) or a corroded trace or via (i.e. a missing path).

I would first clean up all the corrosion making sure there is none left under the ICs.

Personally I would submerge the PCB in vinegar (regular 5% or so) and leave it there for a 2-3 minutes. It's risky, though, but the corrosion is also pretty bad.

Then blow the liquid away with compressed air. Wash in tap water, blow with compressed air again. An then leave fully submerged in pure 99% alcogol for a few days. I personally found kitchen ziplock-style bags to be very useful here.

If it does not help, I would start hunting for some coroded trace. I.e. taking photos, tracing as usual...

In this particular case you have this huge PLCC socket that may have damage underneath. The chip was glued that does not help either.

I would suggest $20 ultrasonic cleaner with vinegar, then soapy water, then clean water. Take it out of the cleaner, blow out all water with compressed air. Dry with a hairdryer  and submerge in pure IPA as described above. You may want to refresh IPA after a day. Blow with compressed air, use hairdryer again....
(See https://www.eevblog.com/forum/repair/advice-on-aperture-repair-on-camera-lens/msg5478658/#msg5478658 too.)

Then you can start debugging.

Hopefully it's a two layer board. You can take PCB photos with your phone, align both sides on a computer and start tracing.
You probably don't need to trace too much, but having both sides on a PC screen helps a lot.

[adinsen]

I made a dilution of citric acid and submerged the pcb. The visibly corroded areas and the switches, reacted very clearly as can be seen in the first picture. I then cleaned it with tap water and IPA from a spray can, and dried everything with compressed air. I found that the glue on the H8 PLCC does not seal in the corners, so I blew extra air in the corners. It now looks better although there is still residue on the RAM-side near the battery connector. And I have seen the first signs of life as there is light in the LED's, the display is showing something, and the CPU clock oscillator is running. Unfortunately the unit is unresponsive, but at least the power supply problem was solved just by this cleaning.

Thanks a lot both! 

[Gyro]

That sounds more hopeful 

Be careful to make sure that you have dried the board completely. Electrolytic corrosion can still occur in tiny gaps like the underside of SMD passives. You can't beat a decent spell in the sunshine and breeze.

[adinsen]

Be careful to make sure that you have dried the board completely. Electrolytic corrosion can still occur in tiny gaps like the underside of SMD passives. You can't beat a decent spell in the sunshine and breeze.

Thanks - it's now enjoying the late afternoon sun here! 

[adinsen]

Unfortunately it stopped working again, but with a different failure mode now as the VCC only climbs to 1.2V when the power is turned on. I'm trying to find out how the circuit switches VCC on. It's clearly controlled by the flip-flop, and there is a six pin component (slightly larger package than a SOT23-6) labeled K1, which I think does the job as it's connected to the battery through a fuse and to VCC as well (although I can't trace the via to the VCC layer, but it's probably hidden in the pad itself). The device is unlcearly marked something like F01ADF but googling that doesn't provide anything meaningful. I'll try to take a picture later today with a macro lens, but if someone has an idea what this device is, it will be helpful.

[adinsen]

Ok, so I just realized my six pin device is just a P-channel power MOSFET, see attached screen shot from a random datasheet and the device marked on the board. I've traced its gate to be driven directly from the flip-flop, source is connected to the battery, and the drain pins are connected to the VCC layer.

Now, I just have to find out why there's only 1.2V on the drain side of it. It seems like a strange failure mode for the MOSFET to start acting like a resistor, so I'll be looking for a short circuit.

[Gyro]

If you can trace back the components on the gate drive to the MOSFET, you might find that there is some board leakage that is dragging down the signal and causing it to not turn fully on. The resistors below your circle in the image really don't look great. You might want to re-solder them or even remove them and clean the PCB underneath. Check that the resistor values are as expected too.

[adinsen]

You're right about the reistors, they don't look good, but they do not seem to be associated with the power supply, at least not the VCC. There are other components on the board that look similar, however, and I'm considering resoldering.

The signal on gate looks clean, no noise. It is very close to 0V when the power should be on, but the MOSFET is not condicting, 0V is measured on drain. This contradicts my theory that drain was directly con I think the MOSFET is blown for some reason, perhaps I short circuited something while working with it. Something else than the MOSFET must be dragging VCC up, and I can measure a diode voltage drop of about 600mV between the MOSFET output and VCC of the HC08 nearby. What is then pulling VCC to 1.2V? I don't know but I obviosly need to replace the MOSFET.

[adinsen]

What I thought was a P-channel MOSFET turned out to be more. I removed the component from the board as it was clearly not providing power. Underneath it was the number "5520". It took me some research, but I eventually found the device: R5520 are power switches for USB applications.

There are two different versions of it, R5520H001A and R5520H001B the only difference being the polarity of the EN-signal that turns on the switch. The one fitted by Percul is the A-version that switches on with a 0V input signal. For further testing, I've replaced it with a wire link as shown in the picture. Unfortunately Mouser only stocks the B-version and I haven't been able to find other sources of the A-version. I've not yet searched eBay.

As can be seen there are now signs of life again. It doesn't react on button presses, though. Pressing the buttons convert to voltage between 0 and 1V depending on what button is pressed. I've traced this signal to an analogue input on the MCU. With the clock generator running at the right frequency the CPU must be running, so why is it not reacting in the display? 

I'll have to do some more tracing...