How to spec a mains transformer

[nukie]

AC is one of the things I rarely get involved, I am not good at it at all.

Often, I would just hack a pre-made 5V switchmode phone charger as a powersupply for my projects. It works damn well because most of my micro powered designs are 5V and below. Then come another 0-30v power supply project which specifies 24V 3A 72VA mains transformer. that's all good and nice.

So what do you do when you are given the input AC voltage as the only requirement?

I have a device which specifies 15VAC input. I don't know how much current it will consume so I wired up the DC circuit and found it consuming 40mA. So I sourced a 2.8VA 2x15VAC transformer. I thought if I screwup I could wire the two secondary winding in parallel to increase the current. A single secondary would supply 15VAC 1.4VA 93mA.

Unfortunately for me, I didn't take in account unloaded voltage. When I wired up the transformer to the mains, I am getting about 26VAC unloaded. I connected the device and measured the dc voltage after rectifier diodes it's 30VDC. You might think this is all cool and normal but there's a 78L15 regulator on the board and it's dissipating some amount of power, it's quite warm actually.

So I should find a transformer with a lower VA? How do I spec a transformer for situation like this?

[IanB]

Something is wrong here. If a transformer winding is specified at 15 V RMS then an AC measurement should show close to 15 V, not 26 V. Well designed transformers have a degree of self regulation so the voltage doesn't jump about wildly. (Just imagine if your mains voltage jumped up from 240 V to 416 V when your neighbours are not using any power!)

So you need to double check your 26 V AC measurement. Either your voltmeter is crap or your transformer is crap.

If you rectify the 15 V AC and feed it to a smoothing capacitor then the DC voltage on the capacitor without load should be 15 x sqroot(2) = 21.2 V.

Nothing is cool and normal about any of this. It is all wrong, terribly wrong...

[jahonen]

I think that your measurements are nothing out of ordinary. Small transformers have quite large voltage swing between full load and idle (tens of %'s). For larger transformers, situation improves significantly (we get around 5% or so). One way to solve this is to use transformer with a lower voltage secondary, but it might take some trial and error to find the right one. And of course if the current consumption of your device will vary too much, there might be trouble at times when the current consumption is high.

Regards,
Janne

[Bloch]

I think that your measurements are nothing out of ordinary. Small transformers have quite large voltage swing between full load and idle (tens of %'s).

True

Here are an example 6VA http://produkt.dantrafo.dk/product.aspx?m=prodlist&viewtype=4&catid=135&lang=uk

15VAC = 100% load
21VAC = 0% load

DC = SQR(2) * 21 = 29.69V at 0% load

Try add a small permanent load at maybe 2-3mA And then Measure the Voltage

[nukie]

I hooked up a 265ohm resistor(translate to about 80mAC load current) across the output, the AC voltage drop to about 20.5v. The transformer is rated 93mA.

Maybe this is mislabeled, I have 4 of these transformer, I will test them later.

[Bloch]

Maybe this is mislabeled, I have 4 more I will test them later.

No it works fine 

Then try at 2*90mA load or the or 2 *  a 265ohm resistor

I am 99.99% sure that it will give 15VAC



Remember also to measure the Pri. 230V

[nukie]

I hook up resistors to both the secondary winding to almost full load, lowest voltage I saw was 20.1VAC. I tried a normal 9V e core transformer, unload voltage is around 10.9v, 50% load is slighty above 10v.

It seems the German like their voltage high? I'll email the manufacturer and see what they say.

[Rufus]

The data sheet specifies a no-load factor of 1.31. Your mains is probably higher than 230 - measure it.

Something isn't right. With 250v mains and no load you shouldn't see much more than 21v from a 15v winding.

[nukie]

I'm beginning to suspect these transformer are out of spec. Unfortunately all 4 of them are out of the same batch so there's no way to tell if they are manufacturing problem.

My voltage is 246vac. Calculated ratio is about 1:15.3 it's still way off from the 230VAC->15VAC when loaded.

Like I say before, I tried a cheap 9V E core transformer it performs as it should up to specification. It certainly does not behave like the PCB transformers. I will update again if I get any news from the manufacturer.

[nukie]

Hello guys,

I received a reply! hmmmmm.....

We refer to your below mail and would like to inform you after discussion with our R&D that your measurements do not show any obstacle to use the transformer.

 -         unloaded and with primary 246 Vac a secondary of 26.9 Vac should be o.k. as nominal input is 230 Vac and secondary should be between 24 and 25 Vac

-         and loaded with 80mAC although the transformer can be loaded with almost 95mAC on each of the 15 Vac secondary the voltage of 20.9 Vac should also be no reason to worry

 

[Rufus]

I received a reply! hmmmmm.....

We refer to your below mail and would like to inform you after discussion with our R&D that your measurements do not show any obstacle to use the transformer.

 -         unloaded and with primary 246 Vac a secondary of 26.9 Vac should be o.k. as nominal input is 230 Vac and secondary should be between 24 and 25 Vac

So either they are talking crap or their specifications are crap. Being a 'short-circuit proof' transformer could account for abysmal load regulation but the specification should reflect that in the no-load voltage factor.

I just checked a 240v 1.2VA transformer from my junk box, the measured no-load voltage factor was 1.24 which I suspect is typical for normal transformers of that size.

[Zero999]

Lots of small transformer (<3VA) are inherently short circuit proof because they have a high DC resistance and a crappy efficiently.

[nukie]

I won't trust the datasheet from this manufacturer in the future. Safe for use yes but no on my equipment, too close to maximum input of 78L15.

[ciccio]

I've used similar transformers,  because of the small size and lots of approvals, but their low efficiency was annoying.
They are inherently short-circuit proof, and that means that the winding impedance is high.
This will save you a fuse on the secondary side, but the transformer will, in some applications, dissipate more heat than the voltage regulator. 
I had a small company custom manufacture for me a pi-to-pin replacement, not short-proof (I actually did not need this feature) but with really better characteristics.
When they disassembled the original transformer (just because they were curious) they told me that the wires were too thin for their standards. Thin wire > high resistance > low efficiency > high no-load voltage.

Lets do some simple (too simplified, in effect) math (I hope that nobody will be annoyed):
Have a look at the specs on the manufacturer's site: they state a 57% efficiency. 
That means that 100 - 57 = 43 % of input power will be dissipated inside the transformer.
If we model the secondary winding as a "perfect" ac generator with a series resistance, a full load of 2.8 VA at 15V (190 mA = 0.19 A) will result in
2.8*43/57 = 2.11  VA dissipated by the internal resistance.
These 2.11 VA result in 2.11 VA /0.19 A = 11.12 V   that is the voltage loss from no load to full load.
The no load voltage will thus be 15 + 11.12 = 26.12 V.
Increase this value to account for the higher mains voltage, this will result in 26.12 * 246/230 = 27.94 V, that is the value that nukie has measured (27.22).
I think that the data sheet is correct in terms of efficiency, but the no-load voltage is wrong.
In any case, I've learned by experience that a 5 volt linear regulator must be supplied by a 9 volt transformer. 15 V is too much, in any case.

[Zero999]

IT's possible to make a transformer which is both short circuit proof and thermally protected of any size by integrating PTC resistor into the winding. I wish most manufactures would do this, it's not expensive or difficult to do.

[ciccio]

IT's possible to make a transformer which is both short circuit proof and thermally protected of any size by integrating PTC resistor into the winding. I wish most manufactures would do this, it's not expensive or difficult to do.

No, from my experience this it is not practically feasible in such small units: the transformer that nukie was testing is 22.7 mm x 22.0 mm x 15.0 mm
That's less than 7.5 cm3, or about 2/3 of a cubic inch!
There is no place for a thermal fuse: it will need some pins inside for soldering it . Really not practical.
Thermal protection, in such small sizes, is entrusted by it being intrinsically short-circuit proof.

[Zero999]

I'm talking about a PTC resistor, not a thermal fuse.

PTC resistors are available in SMT packages which are tiny compared to the dimensions you've given.
http://docs-europe.electrocomponents.com/webdocs/060d/0900766b8060d31d.pdf

[ciccio]

I'm talking about a PTC resistor, not a thermal fuse.

PTC resistors are available in SMT packages which are tiny compared to the dimensions you've given.
http://docs-europe.electrocomponents.com/webdocs/060d/0900766b8060d31d.pdf

Yes, there is the space for such a small component, but it will be a really hard task to solder it to the pins and the wires.
Do you know some transformers manufacturer that uses them?
In effect, the Raychem data sheet  does not mention transformers in the applications list  (only a mention to chargers, that have a PCB inside).

Regards

[Zero999]

I've seen a bi-metal strip fuse inside a wall wart transformer before but I admit it wasn't as small as that, it was around 4VA from memory.

And so what? Just because no one's done it before, it doesn't mean it isn't possible.

[nukie]

In any case, I've learned by experience that a 5 volt linear regulator must be supplied by a 9 volt transformer. 15 V is too much, in any case.

If you mean my case, I said 78L15.
Protection fuse in small transformers is not uncommon there are readily available.