# Isolation transformer for multiple output flyback SMPS

#### jgrv

Joined Mar 2, 2018
36
Hi,

I'm designing a flyback SMPS power supply with the UCC28740 controller.
The design will have a universal input (approx. 95-264VAC), and two DC voltage outputs (5V and 24V).

Max 24W on the output (24V will draw max around 0.8A, and around 50mA on the 5V circuit).

To get away with only one SMPS circuit I was thinking to use an isolation transformer with multiple outputs, so that I could get both 5V and 24V straight from there.

Does anyone know of an isolation transformer I could use for this purpose? I found this, but the turns ratio between 24V and 5V will not give me exactly 24V and 5V. If I design the circuit for 24V output, I will get around 5.3V on the 5V circuit which is a bit too high. The 5V is going to power a microcontroller (atmega 324) and a 16x2 LCD. The 24V doesn't have to be exactly 24V but can be 24V +/- 3%.

Thanks!

#### ebp

Joined Feb 8, 2018
2,332
It is very simple to get multiple outputs from a flyback converter just by adding additional secondary windings. Tracking of the voltages is never perfect, but generally quite good. The usual approach is to close the feedback loop using the output that is most critical. In your situation, because the power requirement is higher for the 24 volt output it would make sense to regulate to that. I would design the transformer for 6 volts for the other output and use a low dropout linear regulator - it would only dissipate 50 mW with a 1 volt drop. Depending on the actual turns required, I would consider one extra turn for the low current output, just to have a little extra margin.

#### crutschow

Joined Mar 14, 2008
29,489
You haven't included the voltage drop of the output rectifier diodes in your voltage calculations for that transformer.
If you use a Schottky diode drop of about 0.5V for each output, then the 24V transformer output would be about 24.5V and the 5V rectified output would be (4/18 * 24.5V) -0.5V = 4.94V

Alternately, you could use the Aux output winding for the 5V, which would give you 8/18*24.5V -0.5V = 10.17V out and drop that to 5V using a linear regulator.
This would reduce the overall efficiency about 2.6%.

#### jgrv

Joined Mar 2, 2018
36
You haven't included the voltage drop of the output rectifier diodes in your voltage calculations for that transformer.
If you use a Schottky diode drop of about 0.5V for each output, then the 24V transformer output would be about 24.5V and the 5V rectified output would be (4/18 * 24.5V) -0.5V = 4.94V

Alternately, you could use the Aux output winding for the 5V, which would give you 8/18*24.5V -0.5V = 10.17V out and drop that to 5V using a linear regulator.
This would reduce the overall efficiency about 2.6%.
Did not think of that! Thanks for clearing that up. That means I can use that transformer after all

I'll design the circuit with the excel tool TI has for this controller. This does not include multiple outputs, but I should be fine designing it for one output of 24V, 0.9A, and then just change the transformer to the one I linked to in the first post, right? And then just add a capacitor to reduce ripple on the 5V output.

#### crutschow

Joined Mar 14, 2008
29,489
I'll design the circuit with the excel tool TI has for this controller. This does not include multiple outputs, but I should be fine designing it for one output of 24V, 0.9A, and then just change the transformer to the one I linked to in the first post, right? And then just add a capacitor to reduce ripple on the 5V output.
From the Schottky rectifier for the 5V, of course.
But note that if you regulate the controller from the 24V output, then the 5V output will not be regulated to changes in its load, and will change some for a change in the 24V load.

Will the 24V load change much in use?

If the 5V changes too much, you might be able to regulate it with an LDO (low drop out) series regulator or (since the current is small) a shunt regulator.

#### jgrv

Joined Mar 2, 2018
36
From the Schottky rectifier for the 5V, of course.
But note that if you regulate the controller from the 24V output, then the 5V output will not be regulated to changes in its load, and will change some for a change in the 24V load.

Will the 24V load change much in use?

If the 5V changes too much, you might be able to regulate it with an LDO (low drop out) series regulator or (since the current is small) a shunt regulator.
Hi,

Yes, the load on the 24V will vary. It will control two solenoid valves, 10W and 8W. Time constant of the valves for 24V step, I don't know.
Here is the sequence:
- First valve 1 will be turned on,
- After around 5 sec valve 2 will be turned on
- After around 1-2 sec valve 1 will be turned off
- After around 5 sec valve 2 will be turned off.

This sequence might be running once every hour.
So mostly one valve will be on, only to overlap with the other for maybe 2 sec at max.

What do you reckon to be the best solution to prevent a voltage drop that will shut down the microcontroller? Maybe have 24V-5V = 19V on the feedback line, and regulate for 19V? Or just place a large enough capacitor?

#### crutschow

Joined Mar 14, 2008
29,489
What do you reckon to be the best solution to prevent a voltage drop that will shut down the microcontroller?
Not sure what's best, but here are some options:

Since solenoid values should be fairly insensitive to voltage variation, you could regulate the 5V output, and let the 24V output change some with load.

If the 5V is high enough, you could regulate it as I suggested in post #5.

Another approach is regulate the 24V and use the 10V winding, dropping it to 5V with a linear regulator as stated in post #3.
That will only drop the overall efficiency by about 1.3%, not the 2.6% I originally stated, for a 50mA, 5V load.
This option is the most likely to give you both outputs well regulated.

#### ian field

Joined Oct 27, 2012
6,536
Hi,

I'm designing a flyback SMPS power supply with the UCC28740 controller.
The design will have a universal input (approx. 95-264VAC), and two DC voltage outputs (5V and 24V).

Max 24W on the output (24V will draw max around 0.8A, and around 50mA on the 5V circuit).

To get away with only one SMPS circuit I was thinking to use an isolation transformer with multiple outputs, so that I could get both 5V and 24V straight from there.

Does anyone know of an isolation transformer I could use for this purpose? I found this, but the turns ratio between 24V and 5V will not give me exactly 24V and 5V. If I design the circuit for 24V output, I will get around 5.3V on the 5V circuit which is a bit too high. The 5V is going to power a microcontroller (atmega 324) and a 16x2 LCD. The 24V doesn't have to be exactly 24V but can be 24V +/- 3%.

Thanks!
Most SMPSU manufacturers use custom transformers, or some even manufacture their own. They're usually vacuum impregnated with laquer and baked hard - its fairly rare that you can add windings.

AFAICR: Power Integrations publish various application motes & bulletins - some include winding details for the transformer as well.

Their SMPSU chips are simple low component count variety too.

#### jgrv

Joined Mar 2, 2018
36
Not sure what's best, but here are some options:

Since solenoid values should be fairly insensitive to voltage variation, you could regulate the 5V output, and let the 24V output change some with load.

If the 5V is high enough, you could regulate it as I suggested in post #5.

Another approach is regulate the 24V and use the 10V winding, dropping it to 5V with a linear regulator as stated in post #3.
That will only drop the overall efficiency by about 1.3%, not the 2.6% I originally stated, for a 50mA, 5V load.
This option is the most likely to give you both outputs well regulated.
The feedback in my circuit is isolated using an opto-coupler. Using the AUX to power the microcontroller would break that isolation.

I could regulate for 25V (which is max for UCC28740?), to get get somewhere around 5.2V / 5.3V, and use a linear regulator. Do you think a sudden load on the 24V would drop the voltage on the 5V output more than 0.2V?

Last edited:

#### crutschow

Joined Mar 14, 2008
29,489
The feedback in my circuit is isolated using an opto-coupler. Using the AUX to power the microcontroller would break that isolation.
I don't understand.
The Aux winding is isolated from the primary winding with a 1500V rating.
Are you using the Aux winding for something else?
I could regulate for 25V (which is max for UCC28740?), to get get somewhere around 5.2V / 5.3V, and use a linear regulator. Do you think a sudden load on the 24V would drop the voltage on the 5V output more than 0.2V?
Without knowing the winding resistance (which is not given in the data sheet, that can't be readily determined.

#### ebp

Joined Feb 8, 2018
2,332
I recommend you regulate to the 5 V output and let the supply for the valves fall where it may. It is very unlikely that the valves would fail to operate with a voltage 10% or more below nominal, and for such short on times and low duty cycle there is no risk of overheating from voltage ten or twenty percent high. If the valves are well specified, a minimum must-operate voltage will be included in the spec's, as will a maximum allowable voltage.

The TI spreadsheet probably covers this but just in case:
Flyback converters work the output filter capacitors very hard. It is not unusual for the peak current to be at least 5 times the average current, so ESR is much more important than it is for other topologies. Fortunately, the long-term average current for your application is very low, which is a great help in terms of longevity of the caps.

#### ebp

Joined Feb 8, 2018
2,332
Presumably he'll be using the Aux winding to power the controller. The other issue with it is that it is pinned out immediately adjacent to the primary winding so it becomes impossible to meet safety clearance and creepage would require a slot routed in the board.

Sometimes in flyback converters with multiple windings, lightly loaded winding have a tendency to "peak charge" the capacitor, resulting in higher than expected voltage. It is a bit hard to predict because it depends on the winding design. In practical transformers with windings on different "levels" there is always some leakage inductance between them. This is another reason I recommend regulating to the 5 V winding. The 24 V outputs may peak charge a little, but again the valves aren't going to care. No practical amount of capacitance on the 24 V output would hold enough charge to keep the voltage too high for more than a few milliseconds.

#### jgrv

Joined Mar 2, 2018
36
I don't understand.
The Aux winding is isolated from the primary winding with a 1500V rating.
Are you using the Aux winding for something else?
Here is the schematic generated with WEBENCH.

I recommend you regulate to the 5 V output and let the supply for the valves fall where it may. It is very unlikely that the valves would fail to operate with a voltage 10% or more below nominal, and for such short on times and low duty cycle there is no risk of overheating from voltage ten or twenty percent high. If the valves are well specified, a minimum must-operate voltage will be included in the spec's, as will a maximum allowable voltage.

The TI spreadsheet probably covers this but just in case:
Flyback converters work the output filter capacitors very hard. It is not unusual for the peak current to be at least 5 times the average current, so ESR is much more important than it is for other topologies. Fortunately, the long-term average current for your application is very low, which is a great help in terms of longevity of the caps.
The valves can operate at a 10% voltage drop. But there will also be voltage drops up to 4% i cables.

When regulating for 5V. What is it exactly that will cause the 24V output voltage to drop on sudden load? Is it because when the valves are turned on, there is some kind of delay before the 5V circuit with the feedback will sense it?

#### crutschow

Joined Mar 14, 2008
29,489
The valves can operate at a 10% voltage drop. But there will also be voltage drops up to 4% i cables.
Can you use larger wire in the cables to reduce that drop?
When regulating for 5V. What is it exactly that will cause the 24V output voltage to drop on sudden load?
It's from the transformer 24V output secondary winding resistance.
The 5V doesn't see that voltage drop so can't compensate for it.
It can compensate for any drop due to the primary winding resistance.

If necessary, you can use two rectifier diodes in series for the 5V output.
That will increase the 24V output by about 10%.

#### jgrv

Joined Mar 2, 2018
36
Can you use larger wire in the cables to reduce that drop?
No, for long distances this will be the case. There's a cost/benefit thing here
It's from the transformer 24V output secondary winding resistance.
The 5V doesn't see that voltage drop so can't compensate for it.
It can compensate for any drop due to the primary winding resistance.
That makes sense.

If necessary, you can use two rectifier diodes in series for the 5V output.
That will increase the 24V output by about 10%.
Great tip, thanks!

Last edited:

#### AnalogKid

Joined Aug 1, 2013
9,944
Since the 5 V output is only 1/4 W, I suggest this: A linear regulator that takes 24 V down to 5 V will dissipate less then 1 W of excess heat, and make the transformer much easier to design.

Separate from that, I don't think you can design, prototype, and build a 20 W switching power supply for less than you can buy one, especially in small quantities. But if that is the task, then I think a switching/linear approach will save a ton of development time. Also, thanks to National and Power Integrations, several transformer companies make single output transformers for small switchers as stock components for the more popular controller chips. Premier Magnetics comes to mind, but it's been a while.

ak

#### jgrv

Joined Mar 2, 2018
36
Since the 5 V output is only 1/4 W, I suggest this: A linear regulator that takes 24 V down to 5 V will dissipate less then 1 W of excess heat, and make the transformer much easier to design.

Separate from that, I don't think you can design, prototype, and build a 20 W switching power supply for less than you can buy one, especially in small quantities. But if that is the task, then I think a switching/linear approach will save a ton of development time. Also, thanks to National and Power Integrations, several transformer companies make single output transformers for small switchers as stock components for the more popular controller chips. Premier Magnetics comes to mind, but it's been a while.

ak
That is worth looking into. Do those companies also sell just the PCB with the components? I need to integrate the power supply PCB into my control module (small plastic container) next to my control PCB

#### jgrv

Joined Mar 2, 2018
36
I've decided to make things simple and regulate for 24V, and use a linear regulator from 24V to 5V. Mostly it's only going to be the microcontroller in sleep mode, which will only draw a few milliamps, so the loss over the regulator will be very small.

I'm looking to find a suitable transformer for this (universal mains input, 24V output, 0.8-0.9A max output current, around 700uH +/- 200 @ around 50kHz).
I am having a hard time finding a transformer that matches my requirements. Custom transformers is not possible at this stage, so if anyone know of one I could use, please help me out

Thanks

#### crutschow

Joined Mar 14, 2008
29,489
Why not use the one you found?
Just don't use the unneeded windings.