Compact Soft-starter/Autotransformer

Some products sold in Australia with voltage specifications like “200V-240V” were really not up to it.  Manufactured for 220V markets, given 240V they would stop working the day after the warranty expired!  The very day! 😉

Apart from a locally purchased Sony TV with that “clever” labelling that’s not my problem.  I brought a stack of 220V equipment back from Hong Kong and my house’s voltage is always around 250 so with SACD players, D/A converters, valve pre-amps and power amps, disasters would be inevitable.  The TV blew its power supply three times before I put a Japanese “Matsunaga Stavol” 220V servo-controlled HK market variac in front of it about 5 years ago and it hasn’t given trouble since.  Modern equipment with switched mode power supplies have a wider input voltage range, but 250V is usually at the upper limit.  I even use one of those Matsunaga transformers in front of a brand new PC just in case.  Inside one of them:

That arm swings around depending on the input voltage (150V-250V) so that the output is always at 220V.

Well I can’t build one of those, but ever keen for a new project, I stumbled across this new article at the ESP web site which inspired me to build this compact 2.3kVA “magic box” to step 250V down to around 218V in a fixed ratio (240V in would give about 209V out).

It has a standard “Nuvotem Talema” toroidal transformer rated only to 225VA but has no trouble powering a 2400W column heater!  To do this the toroidal is connected as a bucking (with a “b” 😛 ) autotransformer in which the light primary winding carries only a small portion of the current.

The 15V secondaries are connected in series with each other for 30V and also in series with the 230V primary winding to subtract 30V (given 250V it works out at a 32.6V subtraction).  Here is how the centre taps were tied and isolated from the chassis with a bit of Teflon tube and double-wall hot-melt glue heatshrink:

The 10A fused active input goes first to the heavy gauge secondary windings then via the active pin of the IEC output socket and the light gauge primary winding to neutral (which is connected straight through).  The earth pins of the IEC sockets are connected internally via a chassis/ground lug.  Wired this way the connected equipment remains protected by the safety switch at the meter box.

It’s less expensive, more compact and is a much safer option than the heavy isolating transformer featured here in which the connected equipment is not protected by the safety switch due to the isolation between the windings.  It has better regulation, is more efficient, can deliver slightly more current and is about 5 times lighter too.  Here they are side-by-side:

↑ Ever vigilant 🙂

Although a small toroidal like this does not need a soft start device for its own benefit, I included one to limit inrush current to around 5 Amps for any connected amplifier(s) deliberately left with their front panel switches in the ON position.  Normally I would use a twin-relay ESP soft start which is the real deal and has additional features, but they require a small auxiliary transformer to power the low voltage timing circuit.  Instead, and as I really wanted to make this thing as compact as possible, I salvaged an “LC Audio” soft start from an old project and used it here.  It was a “light-weight” device with a 330R ballast resistor in circuit for about 1 (long) second doing little more than delay the inrush until after the relay clicked in – low thermal stress on the resistor for sure, but hardly useful for my purposes.  Anyway it operates straight off the mains for compact installation and I did like the way it exploited both poles of its relay to double the nominal contact current rating, so to convert it into a “proper” soft start, I swapped the ballast resistor for a bank of three 11 Watt, 150 Ω wirewound ceramics (= 33W/50Ω), replaced an odd thermistor with a 10 Amp thermal fuse and epoxied it to the top resistor just in case the relay fails to click in one day (I don’t like fires).  The resistors are rated to 250°C and the solder at the ends of the resistors would melt at around 180-190°C so I chose a 167°C fuse.  Some say that these fuses cannot be soldered in place and must be crimped or screwed down – else the wax melts from the heat of the soldering iron.  Well they can if the legs are kept long between the soldering points and the body and the body is held against a cool wet sponge during soldering! 😈

Also as far as I can tell, unlike a regular isolating transformer the series-connected primary/secondary windings do not block unwanted DC which might be present on the mains supply, so I included a DC trap circuit comprising back-to-back 4700 μF 65V electrolytic capacitors and a 35A bridge rectifier (circuit again “stolen” from the ESP website) so the toroidal and the power supply transformers of any connected amplifier should always run quietly.

As usual, the capacitor terminals went a bit close to the chassis lid (about 3 or 4 mm) so this time I used a spare wall switch cover plate as an insulator:

There is also a 10A EMI filter and double pole power switch (at bottom right):

… with a lovely red neon glow. 🙂

Output voltage tested correctly (around 218V), so as an “over the top” failure test and a measure of the integrity of the ballast resistor/thermal fuse arrangement, I switched it on and off 10 times at roughly 10 second intervals with a 2kW electric kettle attached – each time allowing the resistor bank to get hotter and hotter as 5 Amps (250V/50Ω) passed through it for the full pre-relay click duration.  In normal use it wouldn’t suffer such abuse because the inrush drawn by any connected transformer will fall off rapidly after an initial surge and I certainly wouldn’t put any amplifier of mine through rapid power cycles anyway!  It was kind of a test of what might happen if the relay failed.  The only thing that “blew” was the epoxy which cracked off the top resistor with a puff of smoke.  I kind of half expected that as I didn’t use proper thermal epoxy. 😉  I later used some proper thermal conductive epoxy called “Arctic Alumina”.

1250 Watts (417 Watts per resistor) of dissipation is probably a bit of a tall order come to think of it, and although the contraption will never be used at full load conditions, I changed the resistors for a bank of four 330Ω (= 83 Ω) to reduce the inrush limit to 3 Amps.  Each resistor would only have to dissipate 189W in the worst case.  I wasn’t really happy with the idea of that surge current passing along thin copper PCB tracks (designed for only 700 mA) so I put the resistor legs to good use:

And the steady state load could be quite high with a few Class-A power amps attached, so I beefed up the remaining high current tracks around the relay the same way.

And just in case that right hand live wire sits a bit close to the metal stand-off I insulated it with two layers of heatshrink:

I decided to use the contraption in front of the upstairs bi-amped system (all 220V spec) in which both power amps used to switch on occasionally with nasty thuds:

It replaces a Stavol unit and powers up the two power amps, a CD player, a pre-amp and an active crossover with just one switch.  Both amps switch on silently now and as an added bonus the 50 Hz mechanical vibration of the large EI transformer of the CJ MF2500 power amp seems to have reduced a bit (maybe).

Update:  The soft start PCB failed after playing just one CD.  The thermal fuse did its job and blew which means that up to 3 Amps was passing through the resistor bank for some time.  Why?  Well the relay coil seemed OK, but an under-specified surface-mounted signal diode after the active supply voltage-reduction resistor string failed causing the control circuit and relay coil to lose power.  I noticed that a photograph on their web site shows a bigger diode at that position, so I guess they had a few returns!  Anyway I fixed it with a garden variety 1N4004 power diode and it works again (for how long who knows?):

The nice “Managing Director” of LC Audio sent me an upgraded SMD diode all the way from Denmak!  I’ll keep it as a spare.

Working well after a year or so …


Here’s another one (with a 12V tranny for 225V on my mains).  This one uses a cheap pushbutton soft starter PCB from China that needed all of its solder joints redone on arrival!  You get what you pay for I guess!

Leave a Reply

Your email address will not be published.