"over current" use of a Pulse transformer - ( or an alternative )

[mrpackethead]

https://product.tdk.com/info/en/catalog/datasheets/090007/trans_alt4532p_en.pdf

TDK have these small pulse transformers which look really useful, as they have a very small footprint which makes them useful for tight packs.    they are rated for POE+ or 600mA.    I want to run run at 900mA, which is what the 802.3bt standard for POE can go to. 

Other than thermal considerations, ( the device would potentially be disipitating approximately twice the power from 600 to 900mA ) would there be any other issues. 

In this use case, i am confident that i can provide considerable heat releif and cooling, so it might just be possible to get past this issue.

[T3sl4co1l]

If you're okay with qualifying for the application then yeah, there's no one else stopping you from doing so, and it is at your own risk.

Those could also be paralleled, without much impact on frequency response.

Tim

[2N3055]

Tim is right on the money.
Put two in parallel.

[mrpackethead]

In my applicaiton, i'm just running 100M ethernet.  It's worth considering using them in parallel, however part of the attraction is the size.    For this application i can design cooling in from the start.

[T3sl4co1l]

Is the cure worse than the poison? -- how much are you spending on high-k thermal goop / potting / resin / thermal pads keeping those wires cool, and how much space is taken up by that plus the heatsinking, versus just using two, or using a higher rated part?

Tim

[mrpackethead]

Is the cure worse than the poison? -- how much are you spending on high-k thermal goop / potting / resin / thermal pads keeping those wires cool, and how much space is taken up by that plus the heatsinking, versus just using two, or using a higher rated part?

Tim

Yes, this is the balacing act of engineering is'nt it. Constantly having to make compromises and balacing acts.

A higher rated part would be great.  Right now, that means a much much bigger torrodial transformer pair. 

[T3sl4co1l]

Can you get individual toroids (single Ethernet lane coupler) and use both sides, same as this part?

I would think a pack (2 or 4 transformers plus CMCs) is much larger, but this should be similar.  In any case, it can't be much smaller due to signal and thermal limitations, and that hard-limits your device.

Tim

[floobydust]

Why is there no isolation (voltage) rating for the transformer?

[eb4fbz]

https://product.tdk.com/info/en/catalog/datasheets/090007/trans_alt4532p_en.pdf

TDK have these small pulse transformers which look really useful, as they have a very small footprint which makes them useful for tight packs.    they are rated for POE+ or 600mA.    I want to run run at 900mA, which is what the 802.3bt standard for POE can go to. 

Other than thermal considerations, ( the device would potentially be disipitating approximately twice the power from 600 to 900mA ) would there be any other issues. 

In this use case, i am confident that i can provide considerable heat releif and cooling, so it might just be possible to get past this issue.

Yes, there is another issue: core saturation due to excesive current imbalance between the primary halves. So inductance could drop below Ethernet requirements (350uH OCL).

[mrpackethead]

Yes, there is another issue: core saturation due to excesive current imbalance between the primary halves. So inductance could drop below Ethernet requirements (350uH OCL).

Thats an issue irrispective of it being one of these devices or just a 'normal' transformer surely?  In this case it shoudtn be too much of an issue, as the current will be largely balanced.

[2N3055]

Why is there no isolation (voltage) rating for the transformer?

Withstanding Voltage: Compliant with IEEE 802.3, Impulse 2400V 1.2/50μs

If you look into datasheet, you will see that current curve is flat 600mA from 0°C to 70°C at which point temp derating starts.
600mA is probably not thermal bound but defined by core saturation. It will have a security factor, but how much, you will have to characterize yourself.
But I will speculate it won't be 60% more...

[mrpackethead]

Why is there no isolation (voltage) rating for the transformer?

Withstanding Voltage: Compliant with IEEE 802.3, Impulse 2400V 1.2/50μs

I did see that somehwere but then it went missing on me.

If you look into datasheet, you will see that current curve is flat 600mA from 0°C to 70°C at which point temp derating starts.
600mA is probably not thermal bound but defined by core saturation. It will have a security factor, but how much, you will have to characterize yourself.
But I will speculate it won't be 60% more...

 i've only got little knowledge of transformers.     Since the current flow in each side of the primary is towards the center tap ( opposite to each other ), won't the magnetic field cancel the other side out?

[Zero999]

Why is there no isolation (voltage) rating for the transformer?

Withstanding Voltage: Compliant with IEEE 802.3, Impulse 2400V 1.2/50μs

I did see that somehwere but then it went missing on me.

If you look into datasheet, you will see that current curve is flat 600mA from 0°C to 70°C at which point temp derating starts.
600mA is probably not thermal bound but defined by core saturation. It will have a security factor, but how much, you will have to characterize yourself.
But I will speculate it won't be 60% more...

 i've only got little knowledge of transformers.     Since the current flow in each side of the primary is towards the center tap ( opposite to each other ), won't the magnetic field cancel the other side out?

No, because the current doesn't flow from both sides of the tap simultaneously. It alternately flows from each side to the tap.

[T3sl4co1l]

No, because the current doesn't flow from both sides of the tap simultaneously. It alternately flows from each side to the tap.

Sorry but there's not really much I can add... it's a linear circuit, currents don't simply stop and start, it's a superposition of continuous, variable currents where that continuity includes values of zero and negative.   Also, the only nonlinearity we expect in the circuit, results in greater current flow, not lesser, nothing to make it discontinuous.

Tim

[mrpackethead]

Can you get individual toroids (single Ethernet lane coupler) and use both sides, same as this part?

I have not found any if they in fact exisit.

[mrpackethead]

No, because the current doesn't flow from both sides of the tap simultaneously. It alternately flows from each side to the tap.

The 'power' component is nominally 52V DC ( or GND ) , and its applied at the same time to to both sides, and it flows out the CT.   Its DC??        The signal component is of course is effectively AC, but the magntitude of that is very small in comparision to the power.

[mrpackethead]

No matter how nice looking is mentioned TDK small footprint pulse transformer, whole idea of more powerful transformer - it's bigger. Two parallel transformers will not pass IEEE 802.3, thus such cannot be used in other than DIY/hobby short cable runs. I do not see any other solution than using more powerful, 1000mA transformer such as ETH1-460L_

Why would  it not pass?   

The real question is what the limiting factor of this part is. Its it thermal, or is it something else. ( core saturation ).   I will have to find out.

[Zero999]

No, because the current doesn't flow from both sides of the tap simultaneously. It alternately flows from each side to the tap.

The 'power' component is nominally 52V DC ( or GND ) , and its applied at the same time to to both sides, and it flows out the CT.   Its DC??        The signal component is of course is effectively AC, but the magntitude of that is very small in comparision to the power.

If you're talking about the power flowing through both sides of the CT simultaneously, then of course you're right: the flux will cancel, not no net magnetic field induced in the core, both it sounds like a strange, non-standard setup, rather than how the transformer is supposed to be used.

Post a schematic.

[T3sl4co1l]

Mrpackethead has been deeply involved in PoE applications (I don't know how obvious this is from the OP or his posting history), and as far as I know this is a standard application, for which there are such transformers as linked above, as standard off-the-shelf components.   Yeah, differential data passes by, common mode power is drawn through the CT.  Everything is transformer-isolated so there should be no balance issues outside of insulation/conductor failures.

Nice name change..?

Tim

[mrpackethead]

If you're talking about the power flowing through both sides of the CT simultaneously, then of course you're right: the flux will cancel, not no net magnetic field induced in the core, both it sounds like a strange, non-standard setup, rather than how the transformer is supposed to be used.

Post a schematic.

This is a very standard Power over ethernet set up.  I'd suggest 10's if not 100's of millions of ports of POE has been done.  Generic schematic is set up.

[mrpackethead]

Mrpackethead has been deeply involved in PoE applications (I don't know how obvious this is from the OP or his posting history), and as far as I know this is a standard application, for which there are such transformers as linked above, as standard off-the-shelf components.   Yeah, differential data passes by, common mode power is drawn through the CT.  Everything is transformer-isolated so there should be no balance issues outside of insulation/conductor failures.

Nice name change..?

Tim

The difference between current flowing in each side of the transformer should only be minimal. The cable connecting each side is the same guage, and it will be the same length, ( with a few mm ).   This is no differnet from any transformer that is being used for this application.

My gut feeling says that the limitations here, are thermal, and that the device was only ever tested to 600mA because thats what was requried for POE+.

[cdev]

You're probably right about them never needing to carry more so likely being able to handle it.

But suppose it did get too hot, would there be any technical reason you could not use two with their windings simply *in parallel* with one another? - I suppose then you would lose the size advantage, though, so not worth it.

Are all of these POE-ready pulse transformers all rated at the same 600 ma because of the rating being that?

https://product.tdk.com/info/en/catalog/datasheets/090007/trans_alt4532p_en.pdf

TDK have these small pulse transformers which look really useful, as they have a very small footprint which makes them useful for tight packs.    they are rated for POE+ or 600mA.    I want to run run at 900mA, which is what the 802.3bt standard for POE can go to. 

[mrpackethead]

You're probably right about them never needing to carry more so likely being able to handle it.

But suppose it did get too hot, would there be any technical reason you could not use two with their windings simply *in parallel* with one another?

I'm not sure if there is not..

- I suppose then you would lose the size advantage, though, so not worth it.

yes.

Are all of these POE-ready pulse transformers all rated at the same 600 ma because of the rating being that?

There are transformers rated at 1000mA  such as coilcrafts ETH1-460L, theres other parts from Wurth, and no doubt from other manufacturers such as HanRun..  However they are all large.

[cdev]

If you find it runs hot and its potted I guess you could slap a heat sink on it although it might look funny-that would likely help.

[mrpackethead]

Worst case,  at 100m,  you 12.87ohms vs 12.13.     

total current = 900mA..   this would result in 463mA in one conductor and 437 in the other  26mA.       

Its unclear if this difference will be enough to cause saturation of the core.   It will need to cope with 18mA unbalance if its going to run at 600mA.

The articale you reference while interesting is to do with the inblance between the pairs,  this particular problem is related to a single pair.