Deriving a 220v source from two 120V receptacles

[GoldDigger]

There have been devices sold that combined two 120V 15A cords to drive a 240V receptacle which require both 120V circuits to be live and of opposite polarity before closing a relay that energizes the receptacle.
Trying to do this with fixed (even if temporary) wiring instead would run afoul of the need to use a two pole breaker at the source panel.
To avoid GFCI problems you would certainly have to drive the two 120V circuits from a two-pole GFCI breaker.

 

[winnie]

Nice.

If I'm not mistaken, this could be one physical transformer of the style called 120x240 - 120/240, where the primary has two coils, H1-H2 and H3-H4.
....
But with (2) two wire primary circuits, L1, N1, L2, N2, you could wire the primary as L1-H1, N1-H2, N2-H3, L2-H4.

You'd still need a GEC connection to earth the secondary,

Correct if the phases of the selected circuits are suitable and the voltages are well matched.

I'd want to have some built in procedure that ensures correct phasing.

Jon

 

[wwhitney]

With that connection, unless the magnitudes of the L1-N1 and L-2-N2 voltages are equal, relatively high currents may flow between the circuits though the two primary windings. It's kind of like putting two low impedance sources of voltage in parallel.

So are you saying that the voltage on, say, the H1-H2 coil is going to try to induce a voltage on the H3-H4 coil, and vice versa? How are the 4 coils on a transformer like that wound?

If it were wound as a common core with one half of the core length having the H1-H2 and X1-X2 coils interwound on it, and the other half of the core length having the H3-H4 and X3-X4 coils interwound on it, would it still be an issue? But perhaps winding it that way doesn't work properly for one of the more usual use cases, so it's wound differently.

Thanks, Wayne

 

[LarryFine]

This reminded me of a discussion about an auto-transformer (even used the same disproportionate numbers):

How is power used in an autotransformer?

In an application where the “AT” is being used as a step down with load balancing for a 220v genset to a 120v inverter, How is the genset power used? What I mean by this, I understand it induces load on the L2 leg yo balance the genset, so is this load wasted as a “dummy heat load”, or actually...

forums.mikeholt.com

 

[wwhitney]

So are you saying that the voltage on, say, the H1-H2 coil is going to try to induce a voltage on the H3-H4 coil, and vice versa? How are the 4 coils on a transformer like that wound?

OK, thinking about this some more, I remembered that I need to stop thinking of a single phase transformer core as a bar. It's actually a loop (open square). So let me know if the following is correct:

A single phase core carries a single "magnetic current" around the loop. When one coil around it is energized, it creates the "magnetic current" and that establishes a voltage per turn; all other coils get an according voltage induced.

I'm thinking that this 120x240 - 120/240 transformer consists of 4 different coils wound on the single phase core: H1-H2, H3-H4, X1-X2, and X3-X4. I think for a first understanding the physical location of the coils on the core doesn't matter, and that the coils can be considered all the same.

So if you are going to energize two coils on the core, you need to ensure that the applied voltages are identical. When you put H1-H2 and H3-H4 in parallel and energize them, that's automatic. When you put H1-H2 and H3-H4 in series and just energize H1-H4, the voltage point of H2/H3 is floating, so it will naturally end up as the midpoint.

But if you energize H1-H2 and H3-H4 separately as I proposed, there will be an issue if the two voltages aren't the same. If the two 120V circuits are on the same leg, then they should have the same voltage other than effects from the length of the upstream conductors. If they are on opposite legs, then it will depend on how well balanced the 120/240V supply is. And of course if you get the relative polarity wrong on the two coils, you end with a 240V dead short. [In the sense that if you were hooking it up live, everything would be fine for the first 3 connections, but as you are about to make the fourth connection, you'd find there is a 240V potential there.]

Cheers, Wayne

 

[wwhitney]

A couple further questions/comments:

So if a 120x240 - 120/240 transformer is really a single phase core with 4 identical coils, then you could use it as a 1:3 isolation transformer, yes? E.g. energize H1-H2, and then put H3-H4, X1-X2, and X3-X4 in series, to get a H3-X4 secondary of 3 times the voltage. Or is there something I'm missing?

A 3-phase transformer has a figure 8 core with coils wound on the 3 parallel segments and relies on the constraint that the sum of the voltages in a 3 phase system is zero (vs 3 individual single phase transformers with 3 separate cores). If you had a 12-wire 3 phase transformer (is that a thing?) where both ends of all six coils are brought out (just like you'd have with 3 single phase transformers), then you could use it as 2 independent single phase transformers by just using 2 primary coils and 2 secondary coils. The unused primary coil would have the induced voltage of the unbalance between the two applied primary voltages. Correct?

Cheers, Wayne

 

[LarryFine]

A single phase core carries a single "magnetic current" around the loop. When one coil around it is energized, it creates the "magnetic current" and that establishes a voltage per turn; all other coils get an according voltage induced.

Yes. Just like in the discussion to which I linked.

I'm thinking that this 120x240 - 120/240 transformer consists of 4 different coils wound on the single phase core: H1-H2, H3-H4, X1-X2, and X3-X4. I think for a first understanding the physical location of the coils on the core doesn't matter, and that the coils can be considered all the same.

Yes. The two primaries will be wound together, as will the two secondaries.

So if you are going to energize two coils on the core, you need to ensure that the applied voltages are identical. When you put H1-H2 and H3-H4 in parallel and energize them, that's automatic. When you put H1-H2 and H3-H4 in series and just energize H1-H4, the voltage point of H2/H3 is floating, so it will naturally end up as the midpoint.

Yes. That's why we normally let the H-2/H-3 point float.

But if you energize H1-H2 and H3-H4 separately as I proposed, there will be an issue if the two voltages aren't the same. If the two 120V circuits are on the same leg, then they should have the same voltage other than effects from the length of the upstream conductors. If they are on opposite legs, then it will depend on how well balanced the 120/240V supply is.

Yes. Similar to, for example, parallel vs series battery connections.

And of course if you get the relative polarity wrong on the two coils, you end with a 240V dead short. [In the sense that if you were hooking it up live, everything would be fine for the first 3 connections, but as you are about to make the fourth connection, you'd find there is a 240V potential there.]

Yes. Just like connecting jumper cables between two vehicles; 0v vs 24v.

 

[LarryFine]

So if a 120x240 - 120/240 transformer is really a single phase core with 4 identical coils, then you could use it as a 1:3 isolation transformer, yes? E.g. energize H1-H2, and then put H3-H4, X1-X2, and X3-X4 in series, to get a H3-X4 secondary of 3 times the voltage. Or is there something I'm missing?

Theoretically, perhaps, except for the closer coupling between pairs (in my second comment above.)

A 3-phase transformer has a figure 8 core with coils wound on the 3 parallel segments and relies on the constraint that the sum of the voltages in a 3 phase system is zero (vs 3 individual single phase transformers with 3 separate cores). If you had a 12-wire 3 phase transformer (is that a thing?) where both ends of all six coils are brought out (just like you'd have with 3 single phase transformers), then you could use it as 2 independent single phase transformers by just using 2 primary coils and 2 secondary coils. The unused primary coil would have the induced voltage of the unbalance between the two applied primary voltages. Correct?

I have no idea, to be honest, but you can read more about it. One example:

https://www.electronics-tutorials.ws/transformer/three-phase-transformer.html#:~:text=With%20three%20single%2Dphase%20transformers,time%2Dphase%20by%20120%20degrees.

 

[synchro]

A 3-phase transformer has a figure 8 core with coils wound on the 3 parallel segments and relies on the constraint that the sum of the voltages in a 3 phase system is zero (vs 3 individual single phase transformers with 3 separate cores). If you had a 12-wire 3 phase transformer (is that a thing?) where both ends of all six coils are brought out (just like you'd have with 3 single phase transformers), then you could use it as 2 independent single phase transformers by just using 2 primary coils and 2 secondary coils. The unused primary coil would have the induced voltage of the unbalance between the two applied primary voltages. Correct?

You couldn't use it as two independent single phase transformers because the magnetic circuit within the core would have a shared leg where such transformers would couple to each other.

I am aware of figure-8 cores being used for single phase reactors by having opposing windings on just the outside legs, with the middle leg of the core unwound. I think that might be done just to reduce the inventory of cores that the manufacturer needs to carry.

 

[wwhitney]

You couldn't use it as two independent single phase transformers because the magnetic circuit within the core would have a shared leg where such transformers would couple to each other.

I'm not following that.

Say I have a symmetrical figure 8 core (3 legs) with 3 identical coils, A, B, C, one on each leg, with the B coil on the center leg. I apply a sinusoidal voltage V_B to the B coil, and I measure the voltage (if any) on the A and C coils. What do I get, and is it stable?

I thought you would get two voltages V_A0 and V_C0 for which the relation V_A0 + V_B + V_C0 = 0, for the proper sign convention. If the system is truly symmetric, the magnetic flux (?) would divide evenly between the two outside legs, so further V_A0 = V_C0, but if in practice there are small asymmetries, their magnitudes may vary slightly, and the voltage division may not be stable (?). I.e. the system has a degree of freedom.

Then if you further apply a voltage V_A to the A coil (for simplicity sinusoidal of the same frequency), I thought that would make use of the degree of freedom and then shift the system to a fixed point. The new voltage V_C on the C coil would be determined by the constraint V_A + V_B + V_C = 0 imposed by the geometry of the core.

If it's not that simple, what am I missing?

Thanks,
Wayne

 

[kwired]

Two 120V-120V 1.5kVA isolation transformers could be used with each driven by a 120V outlet. Then the 120V secondary windings of the two transformers could be placed in series to produce 240V. GFCI's on 120V circuits should not trip because there would be negligible common-mode current drawn from their outputs because each would be feeding its own isolated primary winding. A 2-pole breaker or handle tie between breakers would be needed as Larry mentioned above.



If the two 120V circuits were on opposite phases of a multiwire branch circuit supplied by a 2-pole GFCI breaker, then it should not trip with a L-L load across the two branch circuits. However, it would not meet the requirements of 300.3(B) which Larry mentioned in post #6 if the conductors of the L-L circuit are not contained within the same cable, raceway, etc.:

Unless permitted in 300.3(B) (1) through (4).

(3) non ferrous wiring methods would be a somewhat common possible situation allowed.

 

[synchro]

You couldn't use it as two independent single phase transformers because the magnetic circuit within the core would have a shared leg where such transformers would couple to each other.

I'm not following that.

Say I have a symmetrical figure 8 core (3 legs) with 3 identical coils, A, B, C, one on each leg, with the B coil on the center leg. I apply a sinusoidal voltage V_B to the B coil, and I measure the voltage (if any) on the A and C coils. What do I get, and is it stable?

I thought you would get two voltages V_A0 and V_C0 for which the relation V_A0 + V_B + V_C0 = 0, for the proper sign convention. If the system is truly symmetric, the magnetic flux (?) would divide evenly between the two outside legs, so further V_A0 = V_C0, but if in practice there are small asymmetries, their magnitudes may vary slightly, and the voltage division may not be stable (?). I.e. the system has a degree of freedom.

Then if you further apply a voltage V_A to the A coil (for simplicity sinusoidal of the same frequency), I thought that would make use of the degree of freedom and then shift the system to a fixed point. The new voltage V_C on the C coil would be determined by the constraint V_A + V_B + V_C = 0 imposed by the geometry of the core.

If it's not that simple, what am I missing?

I agree, as long the core remains in its linear range. Since the magnetic flux through leg C would be equal in magnitude to the sum of the fluxes through legs A and B, the magnetic saturation of leg C (which is the magnetic return path for A and B) will depend on the voltage waveforms of both the A and B coils. And so the windings on leg A and leg B would not act as totally independent transformers in all respects. I also suspect that there may be interactions of harmonics (i.e., intermodulation) between separate transformers on legs A and B because of the nonlinear nature of core materials, but that would be complicated to analyze.

 

[winnie]

Given a design goal of: run a 240V 3000W load from _two_ 120V 15A circuits, without tripping GFCIs, without violating codes, and without blowing up given incorrect phasing of the supply circuits.

1) IMHO any solution which avoids tripping the GFCIs will also prevent violating 300.3(B). To avoid tripping the GFCIs you must never draw current from one receptacle and return it to the other receptacle. This is a basic flaw of the simple 'adapters' that simply take 5-15 plugs and combine them to create a 6-15 (or 14-15) receptacle.

2) I don't believe it is necessary to use a transformer that takes _both_ inputs and combines them. One input can directly supply the load.

3) If the second input goes via a 120V:120V transformer to provide galvanic isolation, and the second input connected in series with the first, then neither GFCI will trip. Any current leaving the second supply simply goes through the primary of the transformer and returns to the second supply. Any current leaving the first supply passes through the load and the transformer secondary, but by KCL has to return to the first supply.

Some tool would still be needed to ensure proper phasing. If the second input went to a rectifier and then a transformer isolated inverter, then this could be made to work with any phase angle between the two sources.

-Jon

 

[wwhitney]

I agree, as long the core remains in its linear range.

OK, to follow up on this, I understand from some reading that a single phase transformer may use a 2 leg core (with primary and secondary each split between the two legs, even if that split is not brought out as in a dual voltage transformer) or a 3 leg core (shell type transformer, windings are only on the central core). Likewise a 3 phase transformer may use a 3 leg core or a 5 leg core (shell type transformer, the outer legs have no windings).

So if you have a 3 phase transformer with 5 leg core, and if non-linear effects like saturation or leakage flux arent an issue, can such a transformer be used as 3 independent single phase transformers? Because the outer legs can carry whatever magnetic flux is required to balance the 3 legs with coils?

I imagine that on a commercially available product there'd be no guarantee that the outer legs were designed such that you wouldn't end up running into non-linear effects. My question is about the basic theory.

Cheers, Wayne

 

[wwhitney]

Some tool would still be needed to ensure proper phasing.

If you knew the two supplies were from a split phase system (not 120/208V "single" phase), then a little cleverness with a latching relay should work. It should be possible to arrange the circuit so that if the relative polarity is wrong, yielding ~0V at the output rather than 240V, the relay gets toggled to the other state, switching the polarity on one input.

And like any product that requires two supply cords, some interlock will be required to ensure that connecting one supply cord and the load doesn't present voltage on the plug of the other supply cord. So the relay solution to do that could probably be incorporated with the "reverse polarity" correction relaying.

Cheers, Wayne

 

[synchro]

So if you have a 3 phase transformer with 5 leg core, and if non-linear effects like saturation or leakage flux aren't an issue, can such a transformer be used as 3 independent single phase transformers? Because the outer legs can carry whatever magnetic flux is required to balance the 3 legs with coils?

It sounds plausible. You need enough core cross section in the magnetic paths to keep the flux density within material limits, but having to magnetize excess material increases the magnetization current. The worst case would probably be the zero sequence case when all of the coil currents are in phase with each other.

 

[synchro]

IMHO any solution which avoids tripping the GFCIs will also prevent violating 300.3(B). To avoid tripping the GFCIs you must never draw current from one receptacle and return it to the other receptacle. This is a basic flaw of the simple 'adapters' that simply take 5-15 plugs and combine them to create a 6-15 (or 14-15) receptacle.

In my house there are some 120V receptacles on MWBC's that have 2-pole GFCI breakers. All circuits are in EMT as required in the Chicago area. And so I could power a L-L 240V load directly across the "hots" of two 120V receptacles that are on different phases of the same MWBC, and the 2-pole GFCI that's powering them should not trip. But I believe driving a 240V load that way would not meet the intent of 300.3(B) or 300.20 because the neutral of the 2-wire circuit to the receptacle would not be used, even though the premises wiring itself meets NEC requirements.

 

[winnie]

I agree. I was assuming 120V GFCIs. The solution that I proposed would use the appropriate neutral for whichever hot gets loaded to neutral, and thus would not trip 120V GFCIs

Jon

 

[LarryFine]

. . . I could power a L-L 240V load directly across the "hots" of two 120V receptacles that are on different phases of the same MWBC, and the 2-pole GFCI that's powering them should not trip. But I believe driving a 240V load that way would not meet the intent of 300.3(B) or 300.20 because the neutral of the 2-wire circuit to the receptacle would not be used, even though the premises wiring itself meets NEC requirements.

My "objection" is in reference to obtaining the line-to-line power from two separate circuits and outlets.

If this was a 3-wire MWBC to one box, then the new wiring would be merely an extension of the circuit.

Once a MWBC becomes separate 2-wire extensions, I don't see feeding a single 3-wire load from them.

 

[winnie]

Once a MWBC becomes separate 2-wire extensions, I don't see feeding a single 3-wire load from them.

IMHO this objection is absolutely correct for circuits and currents.

I don't believe the objection applies to power, and that with suitable transformer isolation it wouldn't apply. If the current on each hot is perfectly balanced by the current of its respective neutral then 300.3(b) is met.

There are tons of other issues which would apply.

Jon