Bipolar Power Supply Load Current Imbalance And Transformer Core Saturation

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
mediatechnology
Posts: 5437
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Bipolar Power Supply Load Current Imbalance And Transformer Core Saturation

Post by mediatechnology »

A Lesson In Split Power Supply Load Balancing

The story resides here: viewtopic.php?f=12&t=763&p=15586#p15586

Moral of the story: Keep DC out of the magnetic circuit...
User avatar
mediatechnology
Posts: 5437
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Split Power Supply Load Balancing

Post by mediatechnology »

Picking up this where I left off...

The larger current draw from the positive supply caused the Texar Audio Prism's power transformer magnetics to saturate.
The imbalance was due to a large amount LED current from the positive supply returning to ground. (An LM3914 was used so there was no choice.)


The positive DC supply current ranges about +300-450 mA; the negative is -90 mA.
The imbalance results in a large amount of DC current in the windings biasing the transformer core.

This is the hum signature of the unit with the current imbalance: Most of the hum is caused by the magnetic flux from the transformer and isn't in the actual DC rails.

Image

This is the hum signature when the negative supply has an added load making it about equal to the positive.

Image

Increasing the total load on the transformer, to balance the load currents and reduce the DC component, greatly reduced the harmonic content of the hum field.
The DC component caused even-order distortion.


Rebalancing the load currents kept the magnetics from "clipping" on the positive half-cycle.

This got me to thinking about split power supplies' rectifiers...

Here is the Texar's schematic which is pretty generic.

Image

Most power transformers have separate independent secondaries that are connected in a center tap configuration.
The outer winding ends are usually connected to a single "full wave" bridge rectifier.
The center tap is at 0V; the DC outputs of the bridge go to the filter caps.

Examining the rectifier bridge and its connection to the transformer it becomes apparent that the actual rectifier topology is "full-wave center-tapped."

The image below provides a visual of the half-wave center-tapped rectifierl. Each polarity of the supply only has two diodes feeding it.

Image

The single "full wave bridge" is actually two "half-bridges" wired full-wave center tap for its respective polarity.

The secondary current requirements have significant differences when full-wave center-tapped is compared to full-wave bridge rectifiers.

When unbalanced load currents are unavoidable it seems to me that the solution to avoid DC in the core is to use two full wave bridge rectifiers like this example from Douglas Self:

Image
Dual Full Wave bridge From Self, Small Signal Design.

(Ignore the fact that the above example uses two positive regulators.)

It looks to me that current imbalance isn't going to setup DC in the windings and saturate the transformer core when two bridges are used.
One secondary will just have a higher load current.

For medium current supplies the extra bridge diodes are cheap and seem to avoid problems when loads are not symmetric.
(Small load imbalances with single bridge supplies are no big deal - they don't shift the transformer core's operating point very far.)

Gross load current imbalance effects on the transformer was something I had never really looked at before.

IMHO the Texar transformer was running beyond it's capacity and near saturation anyway because the manufacturer had switched from dot to bar display. The bodacious amount of hum it generated made me think about symmetry.


For the Texar I solved the problem by installing switching power supplies. :lol:
User avatar
terkio
Posts: 304
Joined: Thu Feb 07, 2019 3:17 am
Contact:

Re: Split Power Supply Load Balancing

Post by terkio »

mediatechnology wrote: Tue Sep 22, 2020 9:59 pm Picking up this where I left off...

The larger current draw from the positive supply caused the Texar Audio Prism's power transformer magnetics to saturate.
The imbalance was due to a large amount LED current from the positive supply returning to ground. (An LM3914 was used so there was no choice.)


The positive DC supply current ranges about +300-450 mA; the negative is -90 mA.
The imbalance results in a large amount of DC current in the windings biasing the transformer core.

This is the hum signature of the unit with the current imbalance: Most of the hum is caused by the magnetic flux from the transformer and isn't in the actual DC rails.

Image

This is the hum signature when the negative supply has an added load making it about equal to the positive.

Image

Increasing the total load on the transformer, to balance the load currents and reduce the DC component, greatly reduced the harmonic content of the hum field.
The DC component caused even-order distortion.


Rebalancing the load currents kept the magnetics from "clipping" on the positive half-cycle.

This got me to thinking about split power supplies' rectifiers...

Here is the Texar's schematic which is pretty generic.

Image

Most power transformers have separate independent secondaries that are connected in a center tap configuration.
The outer winding ends are usually connected to a single "full wave" bridge rectifier.
The center tap is at 0V; the DC outputs of the bridge go to the filter caps.

Examining the rectifier bridge and its connection to the transformer it becomes apparent that the actual rectifier topology is "full-wave center-tapped."

The image below provides a visual of the half-wave center-tapped rectifierl. Each polarity of the supply only has two diodes feeding it.

Image

The single "full wave bridge" is actually two "half-bridges" wired full-wave center tap for its respective polarity.

The secondary current requirements have significant differences when full-wave center-tapped is compared to full-wave bridge rectifiers.

When unbalanced load currents are unavoidable it seems to me that the solution to avoid DC in the core is to use two full wave bridge rectifiers like this example from Douglas Self:

Image
Dual Full Wave bridge From Self, Small Signal Design.

(Ignore the fact that the above example uses two positive regulators.)

It looks to me that current imbalance isn't going to setup DC in the windings and saturate the transformer core when two bridges are used.
One secondary will just have a higher load current.

For medium current supplies the extra bridge diodes are cheap and seem to avoid problems when loads are not symmetric.
(Small load imbalances with single bridge supplies are no big deal - they don't shift the transformer core's operating point very far.)

Gross load current imbalance effects on the transformer was something I had never really looked at before.

IMHO the Texar transformer was running beyond it's capacity and near saturation anyway because the manufacturer had switched from dot to bar display. The bodacious amount of hum it generated made me think about symmetry.


For the Texar I solved the problem by installing switching power supplies. :lol:
Texars schematic.
I do not see how an imbalance of the currents drawn at the +15 V versus the -15 V can induce an imbalance at the currents from the transformer.
Let's consider there is only the +15V loaded, no load on the -15V.
Then the current at the transformer is AC with no DC.....Unless the diodes have different Vf and/or the two secondary center tap windings have different voltages. This AC current is far from a sine, it looks more like positive and negative pulses at the time the diodes are conducting. Nevertheless it should have no DC component, the two said pulses being the same with opposite polarities.
Whatever imbalance of the +15 V versus the -15 V, DC at the transformer cannot come from this, it can only come from an imbalance in the two half waves rectification, diodes mismatch and windings mismatch.
The Dual Full Wave bridge From Self is the way to avoid any of this issue from the windings. It is my favorite way to implement a +V 0 -V regulated PSU; Why waste time with opposite polarity design when duplicating same polarity works fine ?
User avatar
mediatechnology
Posts: 5437
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Split Power Supply Load Balancing

Post by mediatechnology »

Referring only to the top example (the Texar) which is a full-wave center-tap topology.
The positive rail is fed only by a single diode pair.
It looks like a 4 diode bridge but, for only one output polarity, it's not.

Looks to me that when the +15 output is fully-loaded and the -15V output is unloaded that the load return current in the center tap winding flows only on positive half-cycles.

Its analogous to 240V (120-0-120) US house wiring where all the load is on one phase.
Under that condition a neutral current equal to the load current flows.
When both phases are equally loaded (resistively) there is no neutral current: The current flow is from phase-to-phase.

When the loads are balanced the current in the center tap is at a minimum.

How do you explain the change in the harmonic component?
Looks like an operating point change in the core - DC - would shift it to even-order and it is.
User avatar
terkio
Posts: 304
Joined: Thu Feb 07, 2019 3:17 am
Contact:

Re: Split Power Supply Load Balancing

Post by terkio »

The single phase Full wave Center tap ( The middle diagram in the three ) driving one load.
It makes no DC at the transformer. If it were, this widely used configuration would heat, the core saturate, eventually blow up.
DC and spurious harmonics can occur only from a mismatch in the two rectifying legs. From diodes, windings, wiring, solder joints.
Adding two diodes to drive a second load for an opposite polarity doesn't make any difference.

The configuration as of the Self schematic is a sure way to eliminate such issues from windings, wiring, solder joints. Then, the only thing left to make odd 60 Hz harmonics is a not so perfect diode bridge, like Vf1 + Vf3 not exactly equal to Vf2 + Vf4.
User avatar
mediatechnology
Posts: 5437
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Split Power Supply Load Balancing

Post by mediatechnology »

The single phase Full wave Center tap ( The middle diagram in the three ) driving one load.
The middle diagram of the three (FWCT) IS driving one load and I agree it will not produce DC in the core.
(Edit 11/15/20: It does actually produce DC in the core.)

But when the negative supply is added there are two loads - the second is not shown in that drawing.

When I added load to the negative supply to balance out the currents the harmonic content lowered.
The diode's Vf differences were still in the circuit.
Why did the harmonics radiated by the core lower?
By adding current something took the core out of saturation.

When the positive and negative loads are imbalanced where does the difference in current flow?
In the center-tap right?

I agree that the Self circuit, using two full-wave bridges is superior.

FWIW this transformer ran hot, very hot.

I did an experiment here with a split supply using half-wave rectification for each polarity and an unbalanced load which setup DC in the core.

https://proaudiodesignforum.com/forum/p ... =672#p9299

What I observed on the Texar supply was similar despite it being FWCT for each polarity.
User avatar
terkio
Posts: 304
Joined: Thu Feb 07, 2019 3:17 am
Contact:

Re: Split Power Supply Load Balancing

Post by terkio »

mediatechnology wrote: Thu Nov 12, 2020 5:16 am
The single phase Full wave Center tap ( The middle diagram in the three ) driving one load.
The middle diagram of the three (FWCT) IS driving one load and I agree it will not produce DC in the core.
But when the negative supply is added there are two loads - the second is not shown in that drawing.
I think that when the negative supply is added, there is superposition of the two supply currents.
Each supply produces no DC in the core.
The superposition should be no DC, too.
But......
Now, I am not sure that superposition is applicable here, as usually done with linear systems.
User avatar
mediatechnology
Posts: 5437
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Split Power Supply Load Balancing

Post by mediatechnology »

I think there is superposition but when the load currents are imbalanced a difference current flows in the center tap which has an average DC component that causes core magnetization.

Since both loads reference to the center tap, the CT seems to be the only path where the difference in load currents can flow.

The half-wave example I showed using the Velleman supply in the earlier thread is easier to visuallize.

What we have in this situation are two full-wave rectifiers of opposite polarity that appear to behave the same way as a half-wave rectifier when faced with a gross load imbalance.

Consider the transformer without rectifiers feeding load resistors.
One winding is feeding a load resistor connected to the CT.
The second winding is also feeding a load resistor connected to the CT but it is of lesser value.
Where does the difference current flow?
In the CT right?
User avatar
mediatechnology
Posts: 5437
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Split Power Supply Load Balancing

Post by mediatechnology »

I decided to put theory to test.

Using the K8042 Velleman +/-15V supply and an external transformer I was able to confirm that load imbalance does setup a DC current in the center tap.

The test circuit was the Velleman supply feeding a fixed 470Ω load on the -15V supply and a switchable 220Ω load on the positive rail.
I- is -32 mA; I+ is +68 mA or 0 mA.
I inserted a 1Ω sense resistor in the center tap connection to the transformer in order to indirectly read current.

The average DC current in the center tap sense resistor is equal to the DC imbalance of the load currents.

It looks like pulsating DC:

This is with I+ = 68 mA and I- = 32 mA. The difference is +36 mA avg.
The 'scope references to DC 0V so the displayed polarity is negative.

Image

When the imbalance is switched from the positive to the negative rail, I+ = 0 mA, the polarity reverses. The difference is -32 mA avg.

Image

The above shows that load imbalance does setup a DC current in the center tap.
If the DC current is high, the core will become magnetically biased, saturate and produce high order harmonics which include a lot of even-order products.
User avatar
mediatechnology
Posts: 5437
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Split Power Supply Load Balancing

Post by mediatechnology »

I decided to look to see if DC voltage appears across the transformer secondary when there is a load current imbalance.
There is.

When the transformer CT is used as the 0V reference and only the positive output is loaded in 32 mA there is about -120 mV DC developed from the CT to either outer winding.
Measured across the entire secondary 0V is developed.

There is no DC in the overall secondary but DC is developed from the CT to each end.
The DC currents flow from the CT out to the ends.

When the negative output is loaded by the same amount and the positive output is unloaded, the polarity, as expected, reverses.
When both outputs are loaded equally there is no DC developed from the CT to the end windings.

The transformer secondary DCR is about 4Ω per winding so the DC developed across it produces a DC current essentially equal to the load current imbalance as I've shown earlier.
This test confirms those earlier current measurements.

The AC waveform feeding the input to the bridge has, as expected, greater flattening and sharper corners on positive peaks when that output is loaded.
The positive peak current saturates the core more.
When loaded unequally the waveform becomes asymmetric, the distortion adds even-order components and the area under the curve and peak voltage is greater for negative peaks.

I haven't wrapped my head around the polarity of the DC magnetic fields and their effect on the core.
Does the magnetic DC bias from each winding induced into the core add or cancel?

In either case the current in the center tap increases and has much greater harmonic content when the outputs are loaded unequally.

If the transformer is heavily-loaded overall and already near saturation the load imbalance causes a rapid increase in harmonics.
Post Reply