# differentail to single conversion using isolation transforomer

Thread Starter

#### LavanyaThumati

Joined Sep 22, 2020
25
I need to convert differential data signal to single ended signal using isolation transformer
Expected configuration is

I cannot able to measure differential signal so i have used the below configuration

The Input is fed from function generator 5V sine signal, the output is connected to oscilloscope. The same input should be displayed at output. But the output was not in the expected way. The amplitudes are changing for different frequencies, the output amplitude gets distorted for lower frequencies (i.e, 50KHz) and gets amplified at high frequencies (10MHz) . Can anyone recommend any solution for conversion of differential to single signal using the transformer.

#### andrewmm

Joined Feb 25, 2011
1,456
You are doing well.

But falling into the trap of "perfect" parts.

You need to think of the transformers as not being perfect.

They are coils ( inductors ) that have wires wrapped next to each other ( a capacitor ) and they have resistance.

If you think about it , an RLC circuit has a frequency dependent component, its a filter.

At some frequencies the transformer is going to be better than at others,

Try dropping the amplitude of the sine wave, does that make a difference ?
that's because the transformers are none linear magnetics,

Take step back,

what is the specification of the input you are expecting ?
Is it a digital or an analog signal ,
is it a sine wave or is it random ?

Thread Starter

#### LavanyaThumati

Joined Sep 22, 2020
25
Thanks for response

In first images, first transformer part number : MABA-007159-000000 (link :https://cdn.macom.com/datasheets/MABA-007159-000000.pdf) and second transformer part number : MABAES0060 (link : https://www.mouser.in/datasheet/2/249/MABAES0060-961544.pdf).
Now, I am giving input(-2.5V to +2.5V sine wave , 32KHz) from function generator and the output is observed at osciiloscope.
Observations :
The signal amplitude was attenuated at 32KHz, 320KHz, 200KHz
at 5MHz the output was same as input but at >15MHz, output was amplified.

Now I swapped the input and output. same procedure was repeated , same observations were made.

can you please suggest any part numbers for achieving this application
The circuit should work in bi-directional way i.e input should act as output also and output should act as input also i.e when data is fed at data+ and data-, this circuit should convert the signal to single ended , when input is O/P in first image then it should convert the single ended signal to differential signal.
The frequency of signal is 32KHz and signal is -12 to +12V square wave.

#### Ian0

Joined Aug 7, 2020
2,162
Is there a DC offset on your data? It won't work unless your data has a net DC component of zero, using some technique like Manchester coding.

#### andrewmm

Joined Feb 25, 2011
1,456
And are you using the 10:1 probe on the scope, and low capacitance / matched impedance cables from sig gen ?
what is the output termination ?

Transformers are 4 port networks, the reality, they only approaches theory when the network is seeing the correct termination at input and output,.

Show us a picture of your set up please.

#### AnalogKid

Joined Aug 1, 2013
9,244
I don't see anything in post #1 about the frequency range of the signal you are trying to couple, or its waveshape. A transformer cannot couple DC, or the DC component of a complex signal. For a digital data stream, this means that the average value of the bit stream must be 0 V. Something like the output from a UART fails to meet this requirement, which is why NRZ (non-return zero) and Manchester codings were invented.

So, what is the actual signal?

Wait - I found it at the end of post #3. So, for a 32 kHz signal, why do you care about the transformer performance at 10 MHz? Also, 32 kHz is a relatively low freq for data. Magnetics designed for MHz frequencies do not have the right core material for this, and probably not the mass either. Electrically, your signal is more like high-end audio than data. Consider a transformer designed for ground isolation in video applications, designed for a 30 Hz to 5 MHz bandwidth.

ak

Last edited:

#### crutschow

Joined Mar 14, 2008
27,170
The frequency of signal is 32KHz and signal is -12 to +12V square wave.
If that is your signal, why are you testing at such high sinewave frequencies and not with that square-wave signal?

#### rick.curl

Joined Jul 14, 2020
5
Go back and take another look at the transformer spec sheets. Those transformers have a frequency range of 4.5Mhz to 3 Ghz. They're no good at 32 Khz. Not enough material in the core.

#### andrewmm

Joined Feb 25, 2011
1,456
Concur with @rick.curl

those are small , RF ballums.

#### RPLaJeunesse

Joined Jul 29, 2018
164
If you are running at 32KHz with a 24V Pk-pk signal you might be able to get by with a beefy 600:600 Ohm audio coupling transformer. Take a look at the Mouser 42TU016-RC. Yes, it is only rated for voice frequencies but you may find it works adequately at 32KHz.

#### MisterBill2

Joined Jan 23, 2018
8,711
Certainly transformers have frequency ranges and also amplitude ranges, because the magnetic material will saturate and become non-linear if the signal is excessive. And if there is any DC in the circuit there may be non-symetrical saturation. Also, transformers are inductive and so their impedance varies with frequency. So that is why the results are not perfect.