My trusty GW Instek PSP-603 which I procured from the Farnell tradecounter when it was open in Leeds, has died.
There was no warranty on this product as it was a customer return and I got it for £80 so I cannot complain, it has given me almost 4 years of reliable service so far. I think the issue is with the onboard power supply which generates the regulated +5V and unregulated positive and negative rails. Those rails are low when the power supply is cold which leads to a dim display with all segments lit. I suspect that, despite the 5V rail being correct (all of the digital logic operates on 5V) the malfunction presents itself because a rail comparator is indicating a fault condition or there is some self-test performed which indicates a fault state and the processor watchdogs until this disappears.
This is actually a fairly interesting design. It's a 60V, 3.5A power supply for lab use, with an SMPS at heart to reduce the weight and probably the cost too. A pre-switching regulator generates a voltage a few volts above the output and a linear pass transistor produces a stable, regulated output. The pass transistor seems to remove most of the switching noise although I don't tend to use this PSU for sensitive applications. (It feeds the scope prototype, but that is mostly buck converters anyway.)
The other particularly curious thing about this design is it contains not one but two 8-bit 8051 microcontrollers (an W78C032C40DL and MG87FE52AE), one with an external flash memory and one with internal flash. Instead of an ADC, a DAC is used as some kind of successive approximation slope converter. The DAC also generates the Vset and Iset signals for the PSU board by precisely timed output signals from MCU #2, which are latched into sample and hold circuits which generate the required signals for the PSU board. The power converter itself is a fairly standard two-transistor half bridge. Passive PFC is accomplished with a large choke in line with the AC power feed.
I'm not exactly clear on why there are two MCUs, other than it simply being too difficult to fit into one MCU. As far as I can see, the bulk of the work is done by MCU #2 - it drives the keypad, DAC, reads voltage outputs and communicates with MCU #1. The purpose of MCU #1 appears to be to handle the external RS232 port, drive the display, read the EEPROM and picks up pulses from the rotary encoder. This looks like a very old design that's probably been only upgraded slightly over the years and must be expensive to make if it is still in production (Farnell website shows it available.) The date codes on my units suggest a production date of around 2012, but the parts date from the late 90's. I've confidence that almost everything on this board could be replaced with a single ARM or PIC microcontroller nowadays.
Anyway, I found two bulging capacitors on the PSU. I tested with a scope, and indeed when the PSU is not working the +/- rails are at considerably different output voltages to what they are when the PSU is working. Indeed, freezing the caps briefly stops the PSU powering up.
More to follow ... hopefully I can fix this thing.