.:madworm:.'s photos with the keyword: rework

I wouldn't know what to do without that stuff ;-)

The DIM input of the PT4115 is very susceptible to over-voltage - even from a high-impedance source (skin resistance, bridging DIM to VCC).

The scope was behaving strangely, so I replaced the cpu. I removed the old one with ChipQuik and kept it for future projects. Maybe it is not entirely bad, but only can't handle being overclocked to 20MHz. I marked it with an X. Let's see if it did any good. Please note that a new hardware version of this device has been released, therefore these images will most likely not apply to it.

I accidentally killed these by applying 26.0V instead of 2.60V ;-( Instead of typing 026 on my power supply's keyboard I typed 260 and pressed the ON key immediately. The current limiter kicked in immediately (was set to 40mA), but that was too late already. It seems the regulation isn't as fast as expected. Nothing good can come from freak ideas after midnight.

Moving on to new adventures: SMD desoldering of big fat chips. Getting a hot air rework station is too expensive now, so I'll try this first. The first victim will be an ATmega64 sitting on my Jyetech oscilloscope DIY KIT. www.chipquik.com

After hours of searching with the PCB against a strong lamp and not finding anything, I had a look at this via here. After desoldering the header it looked ok at first glance.

The paste flux is really sticky and smells absolutely weird. Not like any flux I've smelled in 20 years. Use sparingly.

You don't need much of the alloy. It's also very brittle. Bending it like normal solder doesn't work. I just cut it to size.

You don't need much of the alloy. It's also very brittle. Bending it like normal solder doesn't work. I just cut it to size.

I had to find a short between the SPI SCK line and ground. I just couldn't find it for hours :-(

After assembly the yellow LED blinked as expected, but the LED matrix stayed dark. I checked the SPI lines (clock, data, latch) and the clock line was pulled to GND for some reason. Luckily I could remove 90% of the circuit from my search due to an intelligently placed jumper. Nevertheless I removed one SO16 shift register after enother and the short still was there :-(

The culprit is on the other side of the PCB, just barely visible with a strong magnifier.

Hidden under the solder mask, there was a faulty connection between the via and the GND line. It was maybe 50µm wide, just barely visible. It looked like dirt or just a figment of my imagination, but I just cut it and the short was gone. Interesting that such a tiny piece of copper showed a resistance of less than 0.4 Ohms.