Recommended load resistor for small transformers

In my small project to measure a 3-phase system (voltage only), I’m testing a 400V/24V 2VA miniature PCB transformer to hook on single phases (so no direct 400V connection on them) in an attempt to minimize core saturation problems to acquire a better voltage waveform. The transformer is this one. PDF spec sheet.

Now, I tested it with a divider net similar to the one used in EmonTX and experiencinh higher than anticipated secondary voltages. Running it from ~240V instead of 400V should result in about 14.4V secondary voltage, but with a divider net of about 120k I read about 21V. It appears the transformer secondary voltage is specified at rated load. And indeed, lowering the load resistor lowers the voltage, no surprise here.

My question is, if my goal is getting a clean, unistorted secondary voltage waveform, should I match the rated load (or near it) with smaller resistors, or should I adjust the divider and stay with as high resistor values as possible? Was trying to use such a small transformer a bad idea? I don’t have a scope to test for obvious distortions or other parameters, unfortunately.

That is indeed the way that all small transformers are rated. Our standard “EU” a.c. adapter is rated at 230 V in, 9.0 V ±5% out, the open-circuit voltage is 11.5 V ±5%.

That should result in less distortion, because it will reduce the flux in the transformer core. I believe we don’t do it because it’s perceived as a waste of energy.

I did, quite a long while ago, test a similar 400 V transformer on 240 V, and it was not noticeably better in the degree of waveform distortion compared to our standard a.c. adapter, which surprised me. I suspect the reason is such small transformers are designed for the minimum possible size, and waveform purity and losses are not a consideration.

If you have a computer and sound card, you could use the method described in the a.c. adapter test report in the ‘Learn’ section. The spreadsheet (or rather, instructions to make it) is here: ZMPT101B, powerfactor and current shown - #52 by Robert.Wall
It is set up for checking the phase error. If you do that, please note the warnings in that and subsequent posts, and be VERY, VERY careful if you connect to the mains side. If you’re only interested in the distortion components, then you need only use one channel on the low voltage side, and that should be relatively safe. But you still need to make sure do don’t destroy your sound card with too much voltage applied to its inputs.

You change the harmonic number in cell K3 N2 and compare the amplitude and phase of each harmonic with the fundamental (1).

Thanks for the insight. I’m not very concerned about a few extra watts, though I planned the device in a project box with confined space, so heat dissipation might be a problem near the rated load. In any case I’ll try to minimize the secondary current as much as possible. For that I’ll take on the suggestion using the sound card method and see what the small transformer does for me. Maybe test results will be useful for others planning to make a similar project.

I’ve made a few samples, please see the attached images. I don’t see the pattern that usually signifies core saturation issues, though my eyeballing might not mean much. I do see the usual distortion around the peaks that’s prevalent on nowadays’ electric systems.

I used a 100R + 10k divider, input is around 20.5V, output around 200mV RMS, 282 mV peak. Is it worth to make any more tests or calculations, or this output can be good enough? I only need about 1% accuracy (with simple calibration) but worse is not a blocking factor.

Vertical resolution: 100 mV/div



That doesn’t look too bad at all. The harmonics (separately) are about 3% for 3 & 5, 7 & 9 are about 1%, and if your supply is like mine, most of that is likely to be incoming waveform itself, not added by the transformer.

I don’t think there’s much to worry about there, it certainly looks no worse than our Ideal adapter.

Thanks. I forgot to add the THD info the audio scope calculated, it was around 4.5% if that says something. I think I’ll use this setup, but will need a somewhat larger “low” resistor to accommodate the 3.3V max ADC input better.

This is what a 400 V transformer looks like on my mains.

Green is the mains input, red is the transformer output.

That also looks good. How did you add the mains to the audio channel?

As per the report in “Learn” on a.c. adapters. Bear in mind I’m an electrical engineer by profession, so I’m aware of the dangers and take all necessary precautions. If you’re not absolutely certain that you have done everything correctly, DON’T TAKE THE RISK.

1 Like

Yes, thanks for the warnings, I’m aware of the risks but right now I don’t think this kind of test is necessary. But as for your answer, it’s a simple resistor divider as I thought.

Now something occurred to me. I’ll eventually need to put 3 of these small transformers near each other in a small box. Should I worry about electromagnetic or magnetic coupling/interference between them that can adversely affect the secondary side signal? If so, is there anything that I can do about this besides trying to put them as far from each other as possible (I’m not sure if this is possible in that box)?

The whole point of the iron is to contain the flux, so very little will leak out. But some will. All I can suggest is you energise one and try to measure the voltage out of a second that’s not energised. their relative orientation will make a difference. If it’s in the low hundreds of millivolts, i.e. a few percent or less, I wouldn’t worry.

And it isn’t as if it will change significantly, so it will appear as a small amplitude and phase error, that’s more or less constant so will disappear when you calibrate.

I see, thanks.

I’ve made a simple test: while one VT is powered, I put another one right besides it, plastic housings touching. I measured about 20mV AC on my DMM in mV range on the unpowered VT (primary unconnected). Transformers are aligned WRT their orientation. Moving away decreases the voltage. About 5 cm away the coupling disappears. Placing the unenergized one perpendicularly induces about 3-4mV. Moving away a few cm and coupling disappears.

Even 20mV is only about 0.1% of the total 21V secondary voltage so I’m not really concerned. Leaving a ~1 cm gap and placing the middle one rotated 90° and we’re far below noise floor.

1 Like

That’s more or less as I expected.