DIY AC power measure (240 V, 50 Hz, < 10 W)

o-scope + current transformer?

yeah, I'll need something.

I found a 170 € meter that would be ideal (not so ideal - see Gough's reply).

I was trying to measure the efficiency of a flyback with a Irms and Vrms meter on the primary. But I get silly results. 

I set my load to draw 50mA (10% load)

output: 12.290 Vdc, 0.0498 Adc = 0.612W power

input: 222.1 Vrms, 0.00876 Arms = 1.946 W apparent power

That 's an apparent efficiency of 31%. While the data sheet says 75%

It would almost mean that my power factor cosφ=  0.612 W / 0.75 (expected input power based on output power and expected efficiency) / 1.964 W (apparent input power)

= 0.816 W / 1.964 W

= 0.415 

yeah, I'll need something.

I found a 170 € meter that would be ideal (not so ideal - see Gough's reply).

I was trying to measure the efficiency of a flyback with a Irms and Vrms meter on the primary. But I get silly results. 

I set my load to draw 50mA (10% load)

output: 12.290 Vdc, 0.0498 Adc = 0.612W power

input: 222.1 Vrms, 0.00876 Arms = 1.946 W apparent power

That 's an apparent efficiency of 31%. While the data sheet says 75%

It would almost mean that my power factor cosφ=  0.612 W / 0.75 (expected input power based on output power and expected efficiency) / 1.964 W (apparent input power)

= 0.816 W / 1.964 W

= 0.415 

It's not unusual for switchers to have poor power factor at low loads that improves at higher loads. This is especially true for passive PFC designs or those with no PFC considerations at all. The power factor can also depend on the quality of the mains sine wave as well. Because measurement errors are greatest when the values are smallest, significant percentage efficiency differences are expected.

If you don't have access to a power analyser (let's just say that's not always an easy win either), then I'd say an oscilloscope with careful set-up and a non-inductive resistance in series with the neutral could be sufficient to start. A CT may be good if you can afford it, but it really depends on the current ranges. AC switchers can be tricky - the switcher probably also causes high current crest factors and non sinusoidal patterns which makes it difficult. If you didn't deskew both channels first, you can't accurately get a phase relationship between the channels to get the real power and if the crest factor is high enough, you might get good current resolution on the peaks but not in the valleys.

That particular meter you linked may or may not be enough - 10mW resolution with 0.4% + 0.1% range (+3 counts - hidden away in the manual) means its accuracy could be doubtful. They didn't indicate things such as deratings for high crest factors. The manual doesn't even list the internal ranges it has, just "automatic".

- Gough

Gough Lui said:

It's not unusual for switchers to have poor power factor at low loads that improves at higher loads.

That resembles my measurements. At 100% load, I measure 72% efficiency with my naive VArms method, vs 81% spec. 
That would indicate a cosφ =  (6.129 W / 0.81) / 8.528 W = 7.567 W / 8.528 W = 0.887

Gough Lui said:

The power factor can also depend on the quality of the mains sine wave as well

It's smurf-hat shaped here, not even close to a proper sinus.

if it is badly distorted you can check your agreement with the utility - there is a possibility that above certain level of harmonics some discount is warranted 

Over here they 'd probably tell me to go buy my electricity somewhere else .

And here, they wouldn't believe you. They won't trust any measurement that they don't perform themselves : ( I couldn't even get compensation for a single light-bulb (it was a fancy £15 decorative bulb!) when I offered them good quality measurements that their voltage was out of the specified range immediately after a power failure.