Signal isolation using a transformer: help interpreting waveform sought

[jamesb]

Hi all,

I am building a decoupling / isolation circuit for use with amateur radio and I've run into an issue.
Using a conventional 1:1 600ohm transformer, I have had success in passing audio tones from the computer and the radio (and vice-versa).

I am trying to use some SMT transformers (DA102MC) and have come across a very strange phenomenon (see attached pic).
When injecting a 1kHz tone at 5Vpp I see something that looks nothing like a sine wave!



I realized that I may be driving the little transformer into saturation so I backed the amplitude down to 100mVpp and managed a decent sine wave output:


I notice that there is a voltage drop of ~60mVpp at the 100mVpp input level. The effect is similar at the varying input levels.

The problem is that I need to have the transformer handle 1.9Vpp from a sound card.
Backing the input signal down to 1.9Vpp still yields the same output waveform which prompted me to add some resistance to the input.
I tried a range of values and even at 22? the resulting waveform is far from ideal.

22? input resistance:


270? input resistance:


What I find interesting is the output waveform of the 270? input resistance test. It appears to be more similar to a triangle wave than the original sine wave.

When I probe the primary (input) side of the transformers, they all exhibit the same waveform as the secondary.
Performing the same tests with a different transformer (TYP145), I see effectively the same results as when using a capacitor to isolate the signal.

Questions:
1. What effect is causing the waveform distortion seen using the DA102MC transformer? Is it saturation?
2. I am looking to isolate a radio (which may carry stray RF over the audio lines) from a computer, what is the best (preferred) method in such a case?

[w2aew]

My first thought would be core saturation.  What is the load on the secondary?  The datasheet indicates these devices are suitable for 75 and 110ohm circuits.  If your load is heavier than this, that might be the problem. 

[jamesb]

My first thought would be core saturation.  What is the load on the secondary?  The datasheet indicates these devices are suitable for 75 and 110ohm circuits.  If your load is heavier than this, that might be the problem.

Hi Alan,

Thanks for the reply! Love your YouTube videos btw.
My initial testing was done with no load on the secondary, then I tried using resistors of varying values which did not seem to have much of an effect.
With the USB sound card attached, I see a much cleaner output which more closely matches the input signal - with the addition of some 60Hz bias it would seem.



When I pull back the input frequency, to say 300Hz the resulting waveform looks a lot more like the first few.



Supposing that the impedance of the sound card is something like 600ohm then another transformer would be more appropriate I would imagine. I looked but could not find the impedances of the SB1140. Another interesting finding is that if the USB sound card is not plugged in to the computer, it is as though there is no load on the transformer at all (perhaps a CMOS switch or something on the sound card?).

I'll try sourcing another SMT transformer - it seems that the 600ohm ones I used in my PTH design worked better, though they are considerably larger.

Alan, am I over-engineering the interface by using transformers? Should I go with capacitors instead?

Have you thought of just getting an audio isolator designed for this kind of signal? They make them for use in cars where there can be a ground loop between the head unit and the amp.

I appreciate your response, however, purchasing a ground-loop filter would defeat the purpose of what I am doing entirely. It is also not really the best solution anyways. Besides, there is no learning in just purchasing a COTS solution :p and ham radio is all about the learning (for me) which is why I seek the support of much more knowledgeable people such as Alan.

Edit: Kind of frustrating that omega (Alt-234) shows up as ? on an EE board :p

[Jay_Diddy_B]

Hi,
 I think that you missed a little Subtlety  on the datasheet:




These Murata transformers are for Digital Audio. These transformers are really pulse transformers. The DA102 can only support 20Vus before saturation. The volt - microseconds comes from  Faraday's Law of Induction:

V = NA d phi / dt

where phi (the total flux) = B (flux density) x Area (of the core)

Since the Area is fixed.

V= NA dB/dt

This equation links:

 the Number of turns N

the cross-section of the core A

The frequency  1/time

And the peak flux density in the transformer.

You need to look for a transformer designed for analog audio.


Jay_Diddy_B

[jamesb]

Hi,
 I think that you missed a little Subtlety  on the datasheet:

These Murata transformers are for Digital Audio. These transformers are really pulse transformers. The DA102 can only support 20Vus before saturation. The volt - microseconds comes from  Faraday's Law of Induction:

V = NA d phi / dt

where phi (the total flux) = B (flux density) x Area (of the core)

Since the Area is fixed.

V= NA dB/dt

This equation links:

 the Number of turns N

the cross-section of the core A

The frequency  1/time

And the peak flux density in the transformer.

You need to look for a transformer designed for analog audio.


Jay_Diddy_B

You are very correct - I did not even consider Et.
I appreciate your detailed response as it nicely explains why the volt-time specification is important!
In retrospect, I should have looked a bit more closely at the units of measure and question their derivation - something which is now perfectly evident.
At least my gut feeling about the resultant waveform was on the right track