Viewer comments:

In RF land we call that a transmission line transformer. At low frequencies, the ferrite coupling does the heavy lifting, and at high frequencies, the transmission line coupling does the heavy lifting. Simple in principle, but getting the crossover really flat in amplitude and phase takes a good bit of experimentation.

this starts to make sense now

Lots of people claiming they could make a wideband injection transformer for a few bucks. Time for a competition perhaps Dave?

It is one thing to make 1 of those.
Producing lots of them is a completley different thing!

$15 for the parts, $485 for the virgins..

I actually did a “mini-internship” at Omicron in Austria a few years back. No nude virgins assembling the devices though :/
Clothed virgins, males.

Its a specific piece of test equipment. If you need one you don’t particularly worry about the cost; it’s a drop in the bucket compared to all your other test gear

Btw you can make directional couplers, hybrids ,circulators etc with good opamps at arbitrary impedances. At 10 Mhz this is no big deal!

The term you were looking for on the transformer was “bifilar windings” it is a common trick for making a broad band transformer.
As an injection transformer, it is usually better to have a 10:1 step down so that the 50 Ohm output impedance of your generator looks like 0.5 Ohms on the system under test.
I was slightly surprised by the small cross section of the core. Generally, the more massive the core the more energy it can store. On the high current side of the transformer, you can have significant DC current leading to significant stored energy in the transformer’s core.

Top tip for anyone wanting to make a transformer with better performance than this: Use binocular ferrites. Hard to get, but if you are serious about being a manufacturer of quality test gear, it is viable when you are charging hundreds of bucks for a wideband flat transformer. If you are being really snazzy, use a pair of binocular ferrites.

The transformer is bifilar wound. The capacitive coupling aids the high end of the BW.

The twisted pair is a type of bifilar winding. It insures the two transformer coils occupy on average exactly the same magnetic space as each other allowing for an almost perfect one-to-one ratio. A typical accuracy of such a winding can be better than 1 PPM.

The pictures below show various different pulse transformers that were tested. The core type and winding method are listed along with the measured inductances.

Recently, I have been designing some SMPS and required some hardware to measure the actual loop response of the complete converters. People familiar with this area will probably know this can be done breaking the feedback loop of the converter and injecting a small AC signal and measuring the loop response at the output of the converter. However, such a measurement requires a mechanism to inject this signal differential across a small resistor inline with the normal feedback network. Unfortunately, as most signal generators are ground referenced we need special hardware to convert this output to a floating output.

Commercial PSU analysers come with expensive injection transformers that allow us to do that. These transformers are designed to have a very flat response in the region of conventional loop bandwidths of SMPS converters, these range from a few Hz to upto a few MHz.

For example this transformer for the Bode-100 costs $500!

The reason these transformers are so expensive is the use of exotic cores that result in high inductance with minimum turns. This way the low frequency behaviour of the transformer is improved without degrading the high frequency. Simply speaking, we need to add more turns to the transformer to improve low frequency behaviour, but the increased parasitic capacitance then degrades the high frequency behaviour.

I experimented with a few DIY transformers by purchasing some cheap £10 cores from RS. I wound around 5m of differential cabling scavenged from an old Ethernet cable.

The results are surprisingly flat, and both transformers work well from about 10 Hz upto 5 MHz, and should be usable upto 10 MHz. Not bad for £10!

Make your own insertion transformer for measuring loop gain-phase of power converters
http://www.simprojects.nl/images/DIY_signal_injection_transformer.pdf

https://www.amazon.com/slp/ground-loop-isolator-for-audio-video/yjyw4vfsybns4h2