Ethernet Magnetics as Analog Signal Isolator (could it work?)

[eleguy]

Something that just that came to my mind. Could ethernet magnetics be used as a signal isolator for high frequency path. Not that I am now even trying to say this would be a proper way to make robust isolation but just as an idea. Apparently IEEE 802.3 standard requires 1500VAC isolation for 1min. Did not now go deeper and check what is the requirement for "standing" = long term isolation. Anyhow, it quickly looked that those cheap SMD devices could be used in some cases. Before thinking further I quickly scratched something up into LTspice (attached as zip). Values are mostly just thrown in without deeper thoughts or analysis. Overall it does not look so bad, right? Certainly some sort of isolated low frequency path and input stage (etc) should be added in order to create DC coupled "probe".

LTspice seems to refer to GND symbol by default which means one has to click on the signal then hold and finally release mouse on the desired reference "zero voltage". Otherwise you can get fancy figures.

EDIT (added after original post, story was not finished)

There are also devices which isolation level goes up to 6kV and some of them can be officially used in the field of safety isolation. Check link below.

[fchk]

Ethernet signal transformers only have FUNCTIONAL isolation, i.e. DC removal etc. You must not use them for SAFETY isolation of ANY kind.

[eleguy]

Ethernet signal transformers only have FUNCTIONAL isolation, i.e. DC removal etc. You must not use them for SAFETY isolation of ANY kind.

Yes, it is apparently clear when you take such a small magnetics SMD component into your hand. There is no clearance at all. But no no... I was not thinking it as a safety isolator of any kind. I just happened to find spice model that particular part easiest.

BUT! It seems that there is something better nowadays available and from this page the idea originally came to my mind. https://www.mouser.co.uk/new/pulse-electronics/pulse-ethernet-transformers/

"provide safety isolation of up to 6kV and are ideal for both industrial and medical applications"

[Karel]

Ethernet signal transformers only have FUNCTIONAL isolation, i.e. DC removal etc. You must not use them for SAFETY isolation of ANY kind.

Depends on the model. Some can be used as a safety transformer, even in medical devices:

https://productfinder.pulseelectronics.com/api/open/part-attachments/datasheet/HXU6200NL

[TimFox]

Spice itself reserves node "0" as a voltage reference for all internal node voltages, usually thought of as "ground".
No node is allowed to "float":  there must be a DC path from each node to node 0.
When the circuit does not have these paths, it is common to add an absurdly-high resistor (e.g., 1 G\$\Omega\$) to the model from the floating node to ground so that the Spice initialization works.

[tszaboo]

Overall it does not look so bad, right? Certainly some sort of isolated low frequency path and input stage (etc) should be added in order to create DC coupled "probe".

If you are thinking about making an isolated probe, then CMRR and noise coupled into your device might be important.

[moffy]

Any idea as to the V*t constant of the transformers? That will provide the signal level by frequency limit, also I imagine distortion levels would be of interest.

[eleguy]

If you are thinking about making an isolated probe, then CMRR and noise coupled into your device might be important.

True. In this quick hack idea the first fully differential amplifier should give CMRR 86/119dB (min/typical, +-1V dc, based on the datasheet). But this setup does not go down to dc (at least not yet). So, to be simulated...

[jonpaul]

See Ethernet PHY stds.

The transformers are typically cheap Chinese toroids , in plstic encapulation, very high capacitacne P-S,

The toroid size ( 5mm ) and limited turns limit the BW approx 10 kHz-100 MHz..1 GHz.

j

[eleguy]

The toroid size ( 5mm ) and limited turns limit the BW approx 10 kHz-100 MHz..1 GHz.

Isolated Analog Probe having 100MHz BW would be more than enough for me. Officially available ones are just dammed expensive. Single chip "solutions" (called isolated amplifiers) offer BW in the range of 200-300kHz. DIY isolated versions (those I have found) are going to some MHz but none up to 100M.

[moffy]

High frequency ADCs can often use a wideband transformer to couple the signal: https://www.analog.com/en/resources/analog-dialogue/articles/transformer-coupled-front-end-a-d-converters.html
might be of interest.

p.s. https://www.coilcraft.com/en-us/products/transformers/wideband-rf-transformers/th/swb/

[eleguy]

High frequency ADCs can often use a wideband transformer to couple the signal: https://www.analog.com/en/resources/analog-dialogue/articles/transformer-coupled-front-end-a-d-converters.html
might be of interest.

I have used (mostly) Mini-Circuits' RF transformers as impedance converters but not with ADC. They typically have pretty low isolation rating. Like 500V or so (if even mentioned at all). They (hiddenly) mention this in the article you linked. As a solution (to this and something else) they offer cascading many transformers. Nice idea, but we would need pretty many of these before getting to any markable isolation levels. Good link though!