I must have designed and built hundreds of AC and DC power supplies over time.  Everything from unregulated extra safe low voltage through variable switch mode designs to high voltage but each was usually specific to an application.  I also have a couple of home made fixed voltage supplies providing regulated +/- 5 and +/- 12 V.  But from time-to-time having a variable 30 V 10 A supply is handy so six years ago I purchased two low-cost mains powered Longwei LW-K3010D variable Switch Mode Power Supplies (SMPS) with adjustable current limit and digital display of voltage and current.  The output voltage is optionally isolated from ground, and is stable with minimal high frequency noise (a common issue with SMPS).

Figure 1.  Longwei 30 V 10 A Bench Power Supply

These power supplies have become my work-horses and all was well - until it wasn’t.  While gradually increasing the voltage on one of the power supplies via the five turn front panel control the output voltage suddenly jumped from about 6 V to over 11 V.  At the time the supply was powering a pump with a motor rated at 24 V to adjust the pump rate.  The pump motor accelerated which was of no consequence but imagine the disaster if the powered equipment was rated below 11 V without over-voltage protection.

The problem appears to be the multi turn voltage adjustment potentiometer (pot).  This assumption is hardly systematic trouble shooting which can lead to all manner of problems but with the power supply disconnected from the mains I can feel that the pot is wire wound and that the action isn’t smooth at around one turn from fully anticlockwise.  Replacing a defective pot hardly warrants a write-up here but you might be interested in why the pot failed.

So, with the power off, and the power supply left to stand for 30 minutes, I removed the five case screws and  the steel back-shell.  Mains powered SMPS often use rectified mains as the input.  The stand time and a quick voltage check of the filter capacitors is a jolly good idea to avoid inadvertent sparks, damage and/or a nasty electric shock.

The construction looks to be a sound build.  The main board is a double sided PCB with no soldering issues apparent and sensible component clearances.  Pretty much everything, including the control pots and display module are connect to the main board with polarized connectors.

Figure 2.  Main Printed Circuit Board

The voltage adjustment knob is a press fit and simply pulls off.  After disconnecting the connector and removing the hex mounting nut and washer and the potentiometer was on the service bench.

The first thing I noticed was that the one of the connector crimp terminals was not inserted fully in the plug body.  An easy fix but not the cause of the problem.  A quick Ohmmeter test confirmed that the pot was defective.  The total resistance between terminals 1 and 3 was a solid 10 K Ohms, but the terminal 2 wiper resistance to 1 or 3 didn’t change smoothly from 0 to 10 K with rotation of the shaft.  At several points in the movement it was jumping to high resistance or open circuit.  So the pot is truly buggered.

Figure 3.  Voltage Control Potentiometer

I have just a few multi turn pots in stock but these are not 10K and are physically too large for the mounting space.  So I ordered two ten turn 10K miniature pots for $10 that look almost identical to the OEM part.  As it turned out I got this slightly wrong.  The OEM parts were five turn with a slightly shorter body and shaft but this was of no consequence.  I carefully cut the shaft to length and the replacement fitted perfectly in the case with adequate clearances from the main board.

Figure 4.  OEM Pot (Right) and Replacement

I removed the OEM connector and re-soldered it to the replacement pot.  Unlike the original soldered connections I moved the wires to the eyelets to reduce the possibility of inadvertently overheating the pot body and internal helical coil resistance winding during soldering.  With the replacement in the case and cover re-fitted the power supply voltage adjustment was back to working perfectly, albeit at 120 degrees per volt as opposed to 60.  And now I have a spare pot.

I was interested to see what had gone wrong with the OEM pot.  So I carefully disassembled it by removing the three small screws on the top plate and turning the plate until the wiper came free from the body.  The wiper looked to be in good order with adequate spring compression.  But I couldn’t see the condition of the helical resistance winding.

Figure 5.  OEM Pot Disassembly

I carefully cut the case open to remove the winding without disturbing it.  It‘s a helical coil of 0.04 mm diameter Nichrome resistance wire (approximately 1 K Ohm per metre) wound over a 1.05 mm diameter enameled copper wire former.  The helical coil sits in a spiral groove in the pot body, fixed at the ends with epoxy .

Terminals 1 and 3 were connected to the coil by direct soldering.  If you’ve ever tried soldering to Nichrome or aluminium then you’ll know that this can be tricky but these connections were sound.  The trick with Nichrome is a high temperature iron,  oxide removal and cleaning, and an appropriate flux.  Solder clearly isn’t appropriate where the wire is a heating element in which case crimping, welding, or mechanical terminals are used.  Aluminium oxidizes so quickly that soldering is best carried out under oil after cleaning under oil.

Figure 6.  Terminal Solder Connection to Nichrome Resistance Winding

A microscope revealed the problem with the winding.   The Nichrome had slid on the copper wire former to form bunches at about one turn of the wiper travel.  Other bunches were beginning to form throughout the length of the Nichrome resistance winding.  And some of the turns were loose on the former.  This was not a consequence of the disassembly.  The bunches formed intermittent shorts between adjacent turns while the gaps and loose turns resulted in poor wiper contact.

Figure 7.  Displaced Resistance Winding on Copper Former

How did this happen?  In the first instance these are cheap and cheerful pots.  The Nichrome helical coil is only held in place on the copper wire former by winding tension, soldered joins to the terminals, and the epoxy at the ends.  Even assuming negligible electrical heating of the Nichrome, changes in ambient temperature result in a significantly greater change in the length of the 10 m long Nichrome winding compared with the circumference of the 1 mm diameter copper wire former.  This differential thermal expansion is estimated to be about 0.05 mm / turn °C.  On cooling the Nichrome stretches on the former.  And on heating the Nichrome winding tension on the former reduces.  Now combine this with wiper action over time and the Nichrome begins to move into bunches.  It is telling that the significant bunching on this pot was at about the 6 V position where the failure was observed.  In recent times this power supply has been routinely adjusted between about 3 and 6 V.

The solution is either a higher quality 10K five turn miniature pot with the Nichrome resistance coil fixed on the former, or a digital input modification to the SMPS.  Bourns make an appropriate pot rated for a million shaft rotations over an operating temperature range of -40 to 125°C but I don’t want to be spending more than the cost of a new SMPS on a replacement pot.

So fingers crossed I’ll get another six years of use out of the new pot.  And I’ll be taking care to ensure that I don’t connect a load susceptible to over-voltage failure without first establishing the output voltage.