Three years ago, I purchased a pair of Manson HCS-3102 programmable switch-mode power supplies. These were not particularly expensive supplies, and judging by some of the noises they made and the granularity of the programming and read-back, weren’t particularly high precision either. They were mainly useful as a “bulk” power supply with some level of current limiting – not suitable for highly current-sensitive work, but good enough to prevent cable-melting disasters. The 1-36v/0-5A range covered most needs, and the remote programming could be leveraged to integrate it into some automated test workflows in a “crude” manner. For the price, it was quite acceptable and they have served me admirably for the past three years. They were part of a (yet to be published) experiment that involved one of them running continuously for over half a year. Now that my PhD experiments are over, they have finally been “released” from duty for regular bench-top use.
As you will see from my original article, one of my biggest annoyances with the supply was the lack of a front panel output enable/disable switch. Many professional and higher-grade supplies have at least a button on the front panel for this feature, and some even boot up with the output off and let you configure everything first (or, let you pre-configure what to power-up in). I was alerted to the fact (and I was somewhat cognizant of it anyway) that the remote control terminal on the rear had a output control line that was active in all modes and could be used to toggle the output even when used from the front panel.
At the time I published the review, the supplies were new and under warranty, and were being heavily used for my research, so instead we worked around the issues with external switching and worked around the 10-second self-test on power-up by being just a little patient. Now that I’m on my own, I’ve decided to take matters into my own hands.
I decided to attempt a modification on one supply just to see whether it worked and whether it was worth my time. I really didn’t want to have anything occupying the remote control terminal if possible, so that I could reuse the port in the future. I also didn’t want to waste the plug and have a dangling cable (I have enough of those already).
I looked up the manual online, which showed that the output control line was a “pull-down” line, with ground on the adjacent pin. This meant that there would be no problems “adding” a switch in parallel to the line, as long as the switch was open circuit when external control was desired.
That’s when I went switch hunting. Because it will basically be shorting a control signal to ground (<10mA current), the demands on the switch were not great. This is a good thing, because it meant that the switch could be small and “nestle” inside the casing in some way. Because I wanted a switch that could be panel mounted, the best type of switch I could think of was a miniature toggle switch. They come in a very large number of varieties – different stem heights, styles, switching patterns, momentary contacts, poles, threaded/non-threaded, with different contact materials. In the end, I settled for the Multicomp 1MS1T1B5M1QE from element14, a miniature single pole, dual throw (of which only one would be used) toggle switch.
The next step was to disassemble the supply and find some place to mount the switch, as well as trace the wires from the control port on the rear to find a way to “tap” them nicely. The front plastic panel seemed to imply there was some space for installing a switch, but because the main controller is housed there with a metal front panel as well, it seemed a bit too intimidating for me.
When looking at the front, on the right side of the supply, there seems to be quite a bit of empty space and clearance, which makes it a good place to mount a switch. It’s also reasonably close to the front, and well within easy reach, but does mean that it’s not possible to stick power supplies side-by-side with no gap … not that you’d really want to be doing it.
You can also see the bulk capacitance on the front and rear terminals, that makes this supply somewhat “sluggish” in voltage changes and “spiky” when it comes to current limiting.
A look at the rear connector shows that the control wire is purple, and the ground wire is red. Way to go in keeping colour convention *cough cough*. If you’ve been astutely following, you would have seen that I had traced the wires through the loom and plucked them out slightly at the front panel end. I decided to cut the wire with some clearance, and unthread the wire out of the loom to add a little length. I stripped the ends, and twisted them together.
Then, the ends were tinned, and placed into a two-way terminal block (in this case, an Elkay 1276-18) for easy solder-less connection and disconnection of the control switch. It also made for a much neater insulation than soldering wires and wrapping tape around it.
Preparation on the lid involved measuring the distance to drill a hole for the switch – you need to provide enough clearance for the upright edge. The hole drilled was initially a little smaller than needed, which was then de-burred with a larger drill bit, and then drilled out to the right diameter to try and make it neat. This is what it looked like after de-burring and before the final drill.
A set of wires with appropriate insulation rating was stolen from an old ATX power supply wiring loom, and soldered to the switch.
The switch was then bolted onto the casing. The wire ends were stripped and tinned. The enclosure was then partially-assembled, the cables screwed into the terminal block and then screwed tightly together.
So now, it’s a power supply with a small toggle switch sticking out of the side.
Now that the modification is done, the question is – does it work, and was it worth it?
The first part of the question is easiest to answer. Yes. It does work. With the switch in the down position shorting the two lines together, the display shows the set-point and allows for adjusment, while the CC/CV LEDs flash quickly in turn. Bringing the switch to the up position opens the lines, and the supply activates with the metered output displayed.
Unfortunately, that’s really where the good news ends. The first caveat I noticed was that the switch response time is ~0.5 to 1 second. It doesn’t sample the line that often, so it’s not suitable for immediate power-down for sensitive devices.
The other thing I noticed was that once the supply decides to power down, it sort of stops regulating entirely, so the voltage floats up slightly before very slowly falling as the capacitors discharge. With a 12v set-point, the voltage jumps to 12.6v when it decides to turn off. That’s not very nice … but for bulk power applications, I suppose it’s not lethal.
With the output set much closer to the 20v limit of the oscilloscope, when unloaded, the voltage takes more than 18 seconds to really reach “zero”. It seems there really isn’t much bleed resistance on the output – maybe programming speed can be improved by putting a fixed load internally to drain the caps. Maybe even reducing the value of the capacitors might work and improve the current limiter behaviour, but at the cost of load ripple.
Powering up was much quicker, but still in the ~80ms ballpark unloaded to 19.5v, which isn’t quite as fast as some other supplies.
All of this was working up to something – namely this. When you power up the supply from cold, and have the switch for the output to off, the supply still decides to power on its output for about 400ms, then turns off with the rising transient and slow decay. This does, to some extent, defeat the niceness of having an output enable/disable which is the ability to have a load connected while the power supply is powered up with no chance of damaging the load. As a result, the load should still be disconnected during power-up regardless of output setting.
It was only then I discovered that the output does have some voltage on it even during the self-test sequence of around 2V. This confirms the advice above – do not connect anything to the output during power-up.
The main benefit of having the switch seems to be the ability to turn the output on and off after the supply has stabilized without waiting for the 10 second self-test of switching the mains power and avoiding the self-test output. But for timely control of the output, and for transient-free control, it doesn’t seem to be the ideal solution.
The modification was successful in adding an output enable/disable switch to the HCS-3102 by leveraging the remote control capability, however, because of the design of the supply, there are several caveats that need to be kept in mind. This includes the output “transient” during power-up of the power supply, the limited response rate to the switch toggle, and the power-off transient/slow discharge of capacitors.
Because of these caveats, I’m not sure if I’m bothered to modify the other “paired” supply. The benefits it confers is limited, and the labour required does take some time. Maybe I will do it at a later date, but at this stage, it seems unlikely I will bother.
In light of the caveats, actually having a physical toggle switch or relay as per older supplies that breaks the connection to the banana jacks on the panel might be more effective as that guarantees a voltage-free output immediately, along with modifications such as adding internal load and reducing the capacitance slightly. Unfortunately, a panel switch that can handle 5A at 36VDC is fairly hard to come by, and is likely to be quite large necessitating significant wiring and panel work to accommodate.
Maybe I will have to instead build an external banana plug/jack board with external switching that will be “universal”.
An Aside: An Update
It’s been a relatively long time since I’ve posted anything up, and not surprisingly, that’s been because I’ve been busier than ever. The list of things includes getting involved in a few freelance jobs on the side which are still ongoing, being part-time employed by the university to write a few publications (which seemingly never ends) and indulging in some Korean TV. But by far and away, the biggest change in my life has been Pokémon Go, which has consumed a decent slice of my day (Go Team Instinct!) and been the greatest motivator for actually getting away from the computer. Frankly, it did something that six fitness trackers couldn’t. After almost a month, I’m now finally Level 22 with 97 filled in my Pokédex, and have walked about 180km. Soon (tomorrow maybe), my thesis assessment results will be out, so it seems posting will become relatively infrequent because of the added workload. There’s always so much to post, but so little time.