The deadline for my PSU project was end of 2016. It's now past mid january 2017 and I am still finetuning things. In all of my projects this is the most stressful part, when the main stuff is there, and I try to push the project to its limits and see how it reacts. Then I always find things to improve, small things, but can be annoying in certain situations if left unsolved. Thermal issues is now a specific area of improvement for the C128+1541-II version of the PSU. C128+1541-II+cartridge draws about 2.8A on the 5V line, so I am testing the DC-DC converter based on XL4005 with 3.1A load. XL4005 temperature. I blew one after several hours with 3.3A. My declared limit for a time unrestricted operation will be 3A. I added a better heatsink to the XL4005 module which brings XL4005 temperature with 3.1A load to about 80°c (room temp 20°C). Chip max temp according to datasheet http://www.xlsemi.com/datasheet/XL4005%20datasheet.pdf is 125°C, so 80°C max is fine. I measured this temperature with the thermocouple extension of my multimeter on top of the heatsink where is it attached to the chip. Now the PSU has run at least 12 hours with 3.1A without issues, 8h continuously. High load test goes on.Transformer temperature. Transformer thermal class is F, so its max temperature can be 140°C with 25°C ambient temperature. At full load (C128+1541-II+cartridge) the transformer reaches a temperature of about 70°C measured inside a hole of the laminated pack. Since the enclosure is made of ABS plastic and its max temp should be 60°C (according to datasheet http://www.scame.com/doc/ZP00604-I-1.pdf ), I evaluated several solutions to cool down the transformer which is adjacent to the enclosure top cover. Adding a noisy fan was not an option, since it also adds the trouble of dust in the long run. Adding holes with vents was very difficult, because finding proper vents is not so easy and cheap, they must be made of plastic otherwise I should ground them (PSU is not grounded, since it has no metal part which can be touched from the outside). The solution I preferred was to use the (limited) ABS plastic box cover thermal conductivity to get rid of heat, I only needed to distribute the transformer heat more evenly over the cover. So I added a metal plate on the whole inner surface of the cover, which is perfectly adjacent to the transformer top when closed. And voilà, the transformer inner temperature dropped down to 63°C at full load. Now that also the metal plate is in between the transformer and the plastic, plastic temperature will for sure be below 60°C in all its points.
Other issues: - fuses, I still need to properly test fuses, especially self resettable ones. The 3A fuse I used on the 5V line triggered too soon (around 2.8A) due to derating with temperature. - I ordered fuses from China but they take a long time to arrive. I should check if I can find them in my local store. - I checked 5V ripple at 3.1A, and it's around 70mV. Voltage drops to around 4.8V with 3.1A, which is acceptable (recommened voltage range for C64/128 components is 4.5-5.5V) - voltage on the C64 with cartridge (approx 1.5A load) is 4.98V and ripple 60mV, so very good. Transformer temp is 50°C - added LC filter on mains side (ferrite + capacitor) to reduce spikes during ON/OFF. - added small capacitor 150 pF in parallel to R2 to filter spikes which triggered the overvoltage protection when touching the PSU case with high electrostatic charge on the human body.
5V line ripple and voltage with 3.1A load:
Metal plate needed to distribute transformer temperature on the cover surface. Note the temperature probe in the transformer hole (red arrow):
Note the heatsink for XL4005 module:
C128+1541-II+REU 256kb:
Current consumption on 5V line is 2.80 A:
XL4005 temperature is 83°C with heatsink:
