strange transformer frequency response

[andia]

Hi all, I am experimenting again. This time I found some nice nickel UU laminations and wound a test transformer for experimenting. the transformer has a 1:7 turns ratio (1000:7000 windings), split between 2 bobbins, interleaved windings (sec-pri-sec-pri-sec) and the two bobbins are connected in reverse to each other. when I measured the frequency response has a strange deep dip at around 20kHz. where does it come from? see picture...

 

[emrr]

hum bucking reverse winds create this sort of cancellation notch, you see it on many old things. Most UTC A / HA / LS for example.

 

[micolas]

@andia
In fact this attenuation of high frequencies could help circuits without de-emphasis but with bright capsules.
This is the case with my Sterling ST-77 microphone which sounds very velvety, smooth, being my favorite on female vocals. Finally, the frequency response of the microphone is fairly flat.

 

[andia]

I have wound another experimental transformer with similar properties. this time I tried to reduce the DCR of the primary by using thicker wire (0.15mm instead of 0.07mm), less turns (750 instead of 1000) and another bobbin with smaller stackheight (1.5 instead of 2 times the width). turns ratio is the same 1:7. also bobbin 1 and 2 are connected in reverse again.
with the new transformer the dip has shifted to above 30kHz. I also found that the dip gets less and lesser the more I load the secondary until it gets reversed to a peak. the transformers frequency response also seems to get flatter the more I load it.
the graph shows loading from 150k down to 100R. source impedance was 150ohms in all measurements.
I don*t really know what to do with the results. I am trying to make a mic input transformer to use with api 312 clones. Is this getting me anywhere?

other measurements taken:
primary inductance 4.2H
primary dcr 28.5ohms
secondary inductance 150H
secondary dcr 885ohms
leakage inductance 0.3mH

 

[CJ]

20k hz is a bit low for any type of rolloff

How much insulation do you have between PRI and sec?

Try putting some more insulation between them, maybe 20 mil.

Neve mic and line inputs have copper foil between the PRI and sec and another wrap around the whole mess.

 

[andia]

there is indeed very little insulation between layers. just one wrap of tape between every primary/secondary split. the bobbins are very small, I fear there is not much room left for more insulation, but I will try, if I can fit copper foil between the layers.these copper layers are then connected together and grounded?

 

[CJ]

Grounded to the core with a small dab of solder to one of the lams, then the lams get grounded via the clamps to the case, then the case gets grounded somehow.

You do not have to use copper, you can just try a bit more tape, a few more mils will work wonders,