Restoration and Repair is a Project14 theme I've been looking forward to. Rarely is there nothing in need of a bit of a fix, which often means saving some money, preventing something from going into landfill and learning more about a given item through the process of repair. This was, and continues to be, something I encourage people to do regularly and has given me a lot of joy especially when salvaging things off the side of the road and bringing it back to life.
I prepared a plan for a rather significant project before I left for my overseas holiday to Japan in March 2020. Needless to say, as COVID-19 transmission ramped up, my holiday was cut severely short and I returned home to self-isolation for a while. This would have been an ideal case to catch up on some repair and restoration work, if it were not for COVID-19 throwing a "spanner into the works."
My Original Plan
Back a while ago, I reviewed the Rohde & Schwarz HMP4040.04 programmable power supply. Unfortunately, it suffered rather severe shipping damage, but despite chasing it up, it was never fully resolved and its fourth channel decided to fail during review. After holding off for a while and ultimately deciding that it was unlikely that any resolution were to happen, I decided to power a teardown of the supply to see what was wrong and whether I could fix it. Ultimately, this first explorative teardown didn't discover the conclusive cause of the failure, but did note that the channel module input fuses and Schottky rectifier diodes were shorted. A downstream short was still detected.
Luckily for me, someone commented on that post which led me to investigate the two N-channel MOSFETs on the top side, discovering they were damaged. This became a promising lead, as perhaps replacing these in addition to the other components could revive the channel. Having won a shopping cart from "Giving Gifts on Valentines Day" back in February this year, I decided to spend the funds on some supplies to make this repair happen - hopefully being met with success. While I waited for the components, and before my overseas trip, I started the process of dismantling the supply.
The HMP4040 is quite a hefty monster, and as a result, taking it apart was not easy. I had no service manual to go on, so I relied on my best judgement along the way, undoing screws as I saw fit, removing the front panel, cross-bar, I/O card cage and ribbon. Eventually, I had to desolder the sense wires from the front panel connections and clip the high-current connections to both the front panel, rear panel and transformer windings, as they were nigh impossible to desolder with my iron due to their thermal mass and lead-free solder. I figured there was enough slack in the wire for them to be stripped and re-soldered after the repair - of course, if it still fails, then I'll have to think carefully about how to proceed afterward.
This also proved to be a great opportunity to get some close-up shots of the PCB. It was around this time that a reader of my review, interested in the HMP4040, asked me more about its design. While I initially gave some vague and generalised answers, the reader pushed for specific detail which made me re-examine the design with more care. It's quite nice to see the channel module like this with its large heatsinking area in the middle. The culprit MOSFETs can be seen on the left image - desoldering them with the hot air gun went successfully, but it did reveal that one of the plastic packages went a bit "gooey" inside, suggesting it did overheat. This led me to wonder whether the nearby Linear Technology LTC3703 Synchronous Switching Regulator Controller might also be busted, but since this part is a bit small, fiddly and expensive, I decided I would not go down that rabbit-hole just yet.
The MOSFETs were removed, but the flux residue looks a bit ugly. Shouldn't hurt anything though. I took the chance to tour other parts of the PCB as well.
A current-shunt resistance in a TO-247 transistor package? I've not seen such a thing before. This looks to be an Isabellenhutte 0.1 ohm 3W ISA-PLAN precision resistor. Power filtering caps are Nichicon GU-series, 105C rated, 3000h, with a slight scratch on the heatshrink which was probably my fault.
The underside of the module, after it had been liberated from the cable harness. It's missing the fuses, bad Schottky diodes as well as an electrolytic capacitor I removed on suspicion it may have been bad. I looked around my room and managed to find the bag of components I removed - as it turns out, there are some SMD resistors as well. A key danger of such repair work is doing things in bits, separated by long periods of time. It's been over two months since I took it apart - I'm not sure I even remember how to put it all back together and where all the screws go. I have some help thanks to a limited cache of photos and the fact that products are usually built with symmetry, so we can refer to the adjacent channel modules for some hints.
Regardless, this project had to be shelved, as a shipping discrepancy for the shopping cart meant that I was missing the fuses. We all know how important those are for safety - preventing fires and other nasty side effects. I wasn't going to substitute anything else for the original Littelfuse Nano^2 fuses, so I guess I'll have to wait.
Restoring the Variac
Instead, it's over to Plan B - looking at this lovely Japanese-made Variac I purchased for AU$50 from the recent Central Coast Amateur Radio Club Annual Wyong Field Day (aka Hamfest).
When I got it, it was secured to an aluminium base plate and was very dirty. Attached to the input was a twin-flex (unsheathed!) connected to the three prong plug (unearthed) with the wires tinned into a ring to be wedged into the binding posts. One of the wires even snapped off as the unit was being transported home. The unearthed nature of this much metal is very concerning, but so is the twin-flex which is single-insulated. A slight nick on the bench would expose live wiring, making this a distinct danger. The panel binding posts had quite a bit of exposed metal, especially around the input, but also seem to only accept unshrouded 4mm banana plugs. The polarity is not marked at all, meaning there is the potential for incorrect wiring.
The unit comes from a Japanese company known as Yamabishi Electric Co. Ltd. This unit belonged to L.E. Harkness at one stage. This unit has a model type of S-260-1, Number 66-401 with a rating of 260VA. I love these stamped/engraved rating plates. The unit takes 240V input, with a variable secondary from 0-260V, at 1A. It's hence not a particularly powerful Variac, but having something like this is quite handy for generating unusual AC voltages, trimming the line voltage to ensure consistent experimental measurements and for testing/servicing older mains-powered equipment especially those which might need a gentle start to reform the electrolytic capacitors. Finally, it seems this unit was not intended for "portable" standalone use (which are usually fully enclosed), as it claims "for installation".
This becomes most apparent when the unit is flipped upside down, exposing the windings which could be live if the insulation is pierced. There is some discolouration, giving some indication that the unit may have been slightly abused at one stage in its past. It doesn't look bad enough to cause the unit to fail - after all, with so many windings, the adjacent windings are generally <1V in potential difference relative to one another, and as long as too many windings aren't shorted out, the unit will still work albeit less efficiently.
The extent of the winding damage seems to be localised around the 180V to 240V mark, which makes sense as that's probably where the unit was used most. It doesn't look too bad, although there is quite a bit of carbon on the top track, so best to clean that off somewhat. The winding order and polarity was confirmed - the common terminals are on the left side, which seemed a little unexpected to me.
Aside from cleaning the track, I also verified the carbon brush was in good condition - and it was! Unlike modern Variacs which may feature a "rolling" carbon brush which might produce less wear and carbon dust, this one has a spring-loaded carbon "pointer" which rides along. An older design, but it seems it has served effectively.
As a result, I cut up some new three-core sheathed flex with a plug on one end, stripping the wires and terminating an Earth lug to the chassis with crimp-ring connectors. The neutral and active were connected in such a way to ensure the output neutral is at the same potential as the input neutral - important to wire this correctly, otherwise the device could be "floating" above Earth potential in such a way to cause issues or risk of electric shock. I also fused the output with a 1A time-delay fuse - unfortunately, I didn't have enough fuse holders to also fuse the input. The output led to a piece of scrap three-core sheathed flex which I terminated in an in-line mains socket (similar to the end of an extension lead), securing the arrangement with cable ties. The wires themselves attached to the Variac using its binding posts - unfortunately, I had no suitable crimp ring connectors, so I just bent the twisted wire ends around the binding posts and secured them tightly. Not ideal, with exposed metal as well, but much safer than it was initially. The proof? See the video!
Video
Below is a video covering what I wanted to do, but failed, but also the "Plan B" and demonstrating the restored Variac.
Conclusion
Sometimes things don't go to plan, so that's why there's a "Plan B". I had also repaired a DAB+ Radio by replacing its battery, but I had no time to film a video for that one. So instead, I took a look at a vintage Variac, cleaning it up and making it safer for use. Now, I feel much more confident about using it, since it's less likely to give a shock now that I can confirm it is earthed, wired properly with regards to polarity and is not otherwise internally faulty. It's surprising how many vintage pieces of equipment are durable and well-built in ways far beyond the standards of modern products, meaning they could serve long lives, but also it's interesting to see how designs have changed to reflect more modern safety ideologies. It's unlikely to find a modern Variac with exposed binding posts that accept unshrouded banana plugs and are not clearly marked as to their polarity! It's not entirely safe yet - ideally a fuse on the inlet should also be used to guard against catastrophic failure of the Variac itself and perhaps a plastic shield should be used over the binding posts to prevent accidental contact, but as I'm aware of the dangers, I can avoid them as much as I can and enlist other safeguards - e.g. having an RCD device protecting the supply into the Variac.
Top Comments