Building a casing for a small autotransformer (Variac)

[Vtile]

I have spend a two evenings to build a case and switches for a small autotransformer. I thought I share. The autotransformer itself is 3 amp unit I bought as a new. All other parts, except the 2.5A automatic fuse, are old 2nd hand stuff I had in my junk boxes.

Contains dangerous voltages and practices.

I did start by cutting a nice piece of 2.5mm white polycarbonate sheet to serve as a front face of the box.


One might wonder why, well this box have seen 'some' PSU projects along the years.


The autotransformer itself is a IP00 unit so it does need a box. Nice feature is the spare carbon, doesn't cost anything for the manufacturer, but certainly is nice bonus for the user.


JOKARI is a nice tool for cutting power cables. Once you have used something similar the sidecutter method makes you. 


The wiring is still not as It will be, even the circuit will be a bit different. I firts though I put the phase first through classic fuse and then to switch and there to automatic fuse, but I forgot to drill the hole for the fuse holder. I did want the unit to be usable next day (today) to test the indication light for the unit as it were old 220V and as a neon bulb I did want to know if the shunt resistor is on the right size for the specific bulb as the current mains voltage is now 230. What I did test it day before quickly with 230 volts it did seem so bright that I did wonder if it were already on the high edge in 220V setup. For the fusing I think I switch the thermal fuse to output line and see if it works and leave the classic fuse for the input line. Meratester is on the background measuring a stray voltages as usual. 
Starting to look something. 



Next day.

I made a test setup for the glimmer neon light. This is the dangerous practise part. 
Autotransformer powered IP00 test setup, without even safety isolation transformer.

Here is the picture of the setup. AVO8 is first for a voltage indication and only next it serve as a voltage measurement device here. The Unigor 6e is a current meter.


220 Volts from the autotransformer. The old style screw binding posts are so nice with this type of stuff (while dangerous) and the weight of the test equipment is several kilograms so it serves as a anchor. 


The old glimmer neon is running at 975 uA, isn't that nice scale in this.  I think I replace the old 12k shunt with 24k shunt for compensate the nominal voltage increase an drive it a bit more low current.



More picures and progress becoming days. Thx.

[Alex Eisenhut]

Nice job! I need motivation for my own Variac project!

[Vtile]

Hopefully you get it done. 

I soldered a new 24k (2x12k) resistor in place powered it through variac just to see that the current draw were substantially more. At a moment I were like  , then took the old resistor which were still inside that discolored silicon tube and attached it to resistance meter and value were 97k (120k nominal) !!! did look closer the color bands and indeed the 3rd is orange not red when I slightly move the silicon tubing.  So need to make it 2nd time and calculate a new value for the shunt resistor.  On the bright side I were almost hooking it up to place in 230 line, but something inside me said measure it before placing it in place, I must have been consiously been aware that I goofed up.

[Ian.M]

It looks like you've got space in that box to add an Ammeter^* - always useful to have on the output of a Variac so you get some warning when you are winding it up before the breaker trips or fuse blows.   A small voltage drop is tolerable so you can use a DC meter and shunt resistor wrapped in a bridge rectifier.  Use a slightly larger resistor than you would for DC and add a trimmer in series with the movement so you can trim the reading 11% high to correspond to the RMS current into a resistive load (or just put the output in series with a true RMS meter and adjust for the same reading when resistively loaded).   Use a really beefy bridge rectifier and put one or two (depending on its burden voltage) silicon diodes directly across the meter movement so it doesn't get destroyed before the breaker pops if the Variac output is shorted.

* Most of the time you need an Ammeter far more than a Voltmeter, because, as long as you know the supply voltage, the Variac's dial provides an adequate indication of the approximate output voltage.

[Vtile]

I might actually do that, but with an peaktech 4½ panel mV meter I have, unfortunately it will need separate isolated transformer and all sorts of signal conversion circuitry around it. Anyhow If I build a ammeter to it I could make it show also voltages with a flip of an switch.  My old panel meter movements are a bit too big to fit nicely on the front as it will have atleast two more indication lights. One for polarity and one for active PE. The full bridge rectified signal to DC ammeter shows IIRC aprox. 65% of actual RMS or were it from peak value.

I need to rethink the feature set. Maybe I just hook a switch and output terminals for ammeter.

[Ian.M]

A dirt-cheap 1mA movement small analog panel meter is plenty good enough, even the crappy edge reading variety if you are *really* stuck for space.  Stick a sliver of red tape on the faceplate in line with the Variac's max rated current, or if your mad skillz with fine mechanisms are good enough and the panel meter isn't ultrasonically welded shut, mark the scale itself or even replace it with a custom one that brings the Max Amps redline closer to FSD (80% FSD is good ).

Switching the same DC meter movement (or Digital panel meter)  to read current and voltage on a mains AC supply can be a bit of a PITA.   You have to use a DPDT switch so you can swap the movement from across the shunt inside the current bridge rectifier to in series with a dropper resistor on the output of the voltage bridge rectifier, and the switch and the voltage bridge rectifier have to be rated for sqrt(2)*max_variac_boost with a large safety factor for inductive spikes.  >500V rated DPDT switches aren't small or cheap. You'll probably also need a bidirectional TVS diode across the bridge's ac terminals.   IMHO its not worth it , , ,

[Vtile]

True too much work for no gain.. It would cost 30€ for a readily build meters for a current and voltage in correct rating. The cost of DIY quickly rises to that level.

You are correct for the switches I have old Bulgins DPDTs, but they are only good for 250Vrms (isolation about to 2.5kV), while I'm driving the variable autotransformer in overdrive with max of aprox. 260 Vrms.

[Vtile]

Maybe I'll just build (note wording = single thought of theory is not given  ) an LED indicator that lit when the output is on 2.9 amps or so.. I need to think about it. I have atleast 7 multimeters (true number unknown ... TEA group therapy is calling) around my workbench so I'm not too much on fuzz about the meter in the variac box.. While the amp warning could be definedly handy.

Maybe through an current transformer, but there is always the EHV problem if the secondary gets open for a reason or another.

[Ian.M]

That's why you solder a permanent burden resistor across the CT secondary for direct readout of current in volts.    It could be quite neat - just a pair of 2mm sockets to take DMM probes and a label:

100mV per A
3A Max.

OTOH it would probably make more sense to get a suitable small cheap panel meter. e.g.

http://www.ebay.co.uk/itm/Class-2-5-Accuracy-AC-0-5A-Analog-Panel-Round-Meter-Ammeter-Meter-Black-D1U4-/201952695750

0-5A AC,  1.5" square

[Vtile]

I wonder what that "Test 1000V NO" does mean. Atleast it doesn't have propriate star with kV marking in it (VDE 0410), but the orientation and movement type symbols (VDE 0410) are still there. 

https://www.eevblog.com/forum/testgear/exploring-the-symbols-on-a-analog-multimeter-quiz/?action=dlattach;attach=12140;image
https://www.eevblog.com/forum/testgear/exploring-the-symbols-on-a-analog-multimeter-quiz/

Well, I haven't made my mind, but I might actually go with the small current transformer route. In a local shed they sell these https://www.digikey.com/product-detail/en/talema-group-llc/AC1005/1295-1101-ND/3881332 for 3.8€ a pop. The smallest I have in my junk box is rated to 100A (electrical lab cleanout trash from a few years back) and is as big as the variac already in the box. 

[floobydust]

I made a variac box and marked the main knob with voltages.
Mine is input 120VAC and output can be 0-120VAC or 0-140VAC depending on how you wire the transformer.
So 100%=? I'm not sure what terminal "D" is for you.
Also, the circuit breaker is really important so you don't cook the transformer.

[Vtile]

Comparison to yours, in my unit those numbers are these letters.

1: C ("normal" input)
2: B (N.A)
3: E (Wiper)
4. A (common / neutral)
5: D ("Overdrive" input)

Yep. The circuit braker indeed should be on the output, I weren't actually aware that the nomencalure for the amp rating is for output when I ordered this unit, but it makes sense. I just wonder how close the braker works, what I have a small exposure of them they "usually" have been pretty nonsensitive for the input voltage. I guess these thermal ones just have bimetal spring that pops when the burden voltage and current heats it up, but I really don't have any black on paper from the subject.

[floobydust]

OK it looks very good  As long as you know if "100" is 240VAC or 280VAC.

I would consider adding an outlet, as I like to simply plug in equipment. It's less mains wiring on my bench.

[Vtile]

OK it looks very good  As long as you know if "100" is 240VAC or 280VAC.

I would consider adding an outlet, as I like to simply plug in equipment. It's less mains wiring on my bench.

I thought also the outlet route, but this is more of an "do it all" unit build with as low budget I can afford (in safety aspects), so I did end up on the unorthodox solution I wire an banana plug (shrouded) mains outlets. One with PE pins and one without for use with isolation transformer.  That is the reason it actually have the PE at all. Then I can also easily wire it to my shielded test box.

I also will add a indication lights for the polarity (as we have here the idiotic schuko) and Protective Earth. Which obviously both are not lit when there is no PE connection. (Needs some experiments and measurements before I drill the holes, mainly for the parasitic coubling aspects and how reliable the indication will be. I'm going to try with glimmer neons, but parasitics might be enough to lit them, so in that case the LED solution might be better as they typically need mA not uA for illumination)

[Vtile]

Hmm.. I wonder if there is an LEDs that have very low forward voltage in 0.8 to 1.2 volts. Then I could keep the shunt resistor below 0.4 ohms. Might need some trickery still to get a sharp enough cut out and over voltage protection when keeping everything passive. Edit. Ah, the current transformer will solve the small power, small current, but 1.8 volt burden shunt problem obviously.. I need to make some calculations andput it on the paper to see if I can get everything from one transformer or do I need two.

[Vtile]

Yep. The passive led solution through current transformer is not my solution. Pretty annoying circuit to test and since I do not have led model and my on the fly model gives 200mV wrong voltage readings.. It would be easy if the led actually would have ideal forward voltage knee... and the safety shunt would act as a voltage source until the forward barrier brakes.. and another led for different brake through voltage level in another halfcycle.. but no. The result is some form of (poor) constant power source. phew. Interesting brain teaser that did show how much I had forgotten the circuit analysis. Interesting actually since there is not many ideal current sources around.

[floobydust]

I'm not clear about what you are trying to do with the LED.
A current-transformer can light an LED to give a load current indication.
But need many turns through CT and its DC resistance might be high. I use old GFIC CT's (1000T) to drive LED's at ~1A up.

[Vtile]

I'm not clear about what you are trying to do with the LED.
A current-transformer can light an LED to give a load current indication.
But need many turns through CT and its DC resistance might be high. I use old GFIC CT's (1000T) to drive LED's at ~1A up.

Basicly to get a led light in reasonably (highly subjective) sharp on/off curve with 1/1000 CT. I think it could be done, but uh. got irritated for the analysis, I now found one "might work" model for LED from the net. One thing is what I tried to test it (real led) with DC ideal current source ("ideal" from 2.8 to 15 mA with <30v potential) is that the turn on seems out of resistor programmability, only way I got close were to use series forward diodejunctions to lift the overall forward voltage drop. I had an 1k shunt in place. In this case it did seem that the forward drop of the LED didn't follow the voltage at all, but the constant current turned the led on (dim) and I lost any sort of ability to set the turn on current. I haven't driven leds from current domain though, voltage dominated circuits have been the norm, so I wouldn't be suprised that I try to think of it at the wrong angle.

Indeed if I divide the ratio I would have more current to play with, compared to situation where you try to shift the 3 mA to leg to another in the circuit.  Why I didn't though that and just parroted the specsheet 1:1k. 

Anyways, I think I will hook up the CT to the unit, put an 100 ohms resistor as a shunt (with parallel trimpot) and attach an BNC connector at the frontpanel. This gives the ability to attach the scope directly to the variac box and therefore look at the current trace without my flimsy current clamp or use multimeter with BNC connector or banana adaptor. Later if I feel that I still need an current level warning (if the Variac haven't let the smoke out at that point) I hook up a small powertransformer to mains and put a small voltage comparator to lit the red led at exactly in certain current level. 

One idea that come to my mind is some form of transistor switching. Ie. in ideal zener setup. ...This variac setup were supposed to be a quick project. :/ Edit. In the morning, nope I leave these and just put the plain CT with BNC and return to it with better time (it also have less influence to measured circuit). The magnitude of current flows are still the main problem (if ratio is not divided), with only a few mA (where led would take 2mA so >65% at my case) the circuit would need some form of latching transistor circuit (since low current pool, Fets.) around the LED to keep it lit once the first transistor is turned on at desired current so the first current branch is dropped out and the LED branch gets all the current.  *meh* Some 10V or so MOV might solve the EHV guarding at a background. I would also need to build an safe voltage controlled ideal AC/DC current source capable to really progress on this matter and see how different silicon components react to currents (which propably will happen since it would be eye opening).

[Vtile]

Now I rewired the whole unit again, soldered new (and this time right valued... I'm still wondering a bit why I did do it wrong at the first time as I can't find the original anymore) current limiting resistor for the neon bulb and assembled the current transformer board. I also put the fuse in place. Some tinkering still needed.

The DPDT switch have a few wires attached.  It does now have selection for 0..230V or 0..~260V ranges. There is no yellow-green tubing on that wire, it is just an optical illusion. (I had only the color in that diameter.. )


Looks pretty neat. The BNC does have now output of 100 mVac per 1A and because it is galvanically isolated I can attach my scope directly to it. 
The low voltage (260AC) output is now red as I found out that the black is too easy to mix with typical DC minus with red and black test leads. The polarity and PE error indicator lights are still missing, I must try to source something nice neon bulb based (Leds start to be the norm. Too much current on those ).

[floobydust]

It is coming along well, looks good.

If you are running a LED from a CT, you need a shunt-diode so the CT sees an AC load and core does not get magnetized.
I like the idea of knowing if some load current is flowing. Otherwise you wonder "is this thing on?". Take a GFCI current-transformer from the garbage, that's what I do.

Ensure there is a PE ground connection to the transformer/case. The BNC might be best isolated from chassis ground to prevent a ground loop, if you want to scope the signal.

[Vtile]

Thanks. 

I can't get access to old GFCIs and cheapest Wung-Ho-Low units cost 10€ a pop ( and real hardwired GFCIs cost >35€ ), eBay is option, but shipping costs makes it uneconomical. The case is all plastic so no grounding, but there is definedly some insulating/separation work to do. I need to source also an insulating bushings for the variac bolts and put some heatshrink to PE, connector. Also those corner rivets are going to be replaced with nylon bolts. I wish I would have a hipot tester (in the build/buy list) so I could see the magnitude of insulation on the variac shaft. The knob is a bakelite so the only access to the potential in case of failure is the tightening screw which I will insulate also, so the shaft insulation isn't that much a deal for me personally.