2 deleted 138 characters in body; edited title
source | link

Isolated vs non-isolated dc/dc in multi-rail design.

I want to build a efficient, small and lightweight multi rail dc/dc power supply. I've been thinking about several topologies and ideas. However, the main question of "isolated vs non-isolated" still stands. I am working with a 3cell lipo (11.1V nom. 9.2V min, 12.6V max). So the input voltage will vary over time. I need atleast 12V @ 5A output, 5V @ 5A, 3V3 and perhaps some other specific voltage levels I yet have to receive info about. So my initial thought was quite biased for my love of isolated power supplies: With the following block scheme as my idea: The

The idea behind this was to have a single core with multiple output windings. So I thought, since N1:N2:N3:Nn will always remain constant, any offset in the voltage on a single rail will also be present on the other rail. Correction would be done via a microcontroller. This simplifies the circuit to a single switching stage that corrects itself and a sense circuit. I have no idea how well this works, if it does. A problem with this would be that accuracy of the voltage outputs will be crude, since the amount of turns rarely turn out to give perfect 5000 mVDC for example. The lack of regulation on each output can be problematic with some sensors which this converter will power.

My other idea was to have a non-isolated dc/dc converter for every single output. Like this:

non-isolated

For the buck-boost I would use the SEPIC topology and for the buck I would consider a synchronous buck converter. The advantage of this is that it won't use a heavy transformer and all the outputs are properly regulated. A problem would be that it requires multiple dc/dc converters.

I would love to know all your knowledge, tips and experience regarding this design stage. If you could elaborate your preference I would highly appreciate it and certainly take it into my final decision!Which concept is more suited for building a DC/DC power supply?

Isolated vs non-isolated dc/dc in multi-rail design.

I want to build a efficient, small and lightweight multi rail dc/dc power supply. I've been thinking about several topologies and ideas. However, the main question of "isolated vs non-isolated" still stands. I am working with a 3cell lipo (11.1V nom. 9.2V min, 12.6V max). So the input voltage will vary over time. I need atleast 12V @ 5A output, 5V @ 5A, 3V3 and perhaps some other specific voltage levels I yet have to receive info about. So my initial thought was quite biased for my love of isolated power supplies: With the following block scheme as my idea: The idea behind this was to have a single core with multiple output windings. So I thought, since N1:N2:N3:Nn will always remain constant, any offset in the voltage on a single rail will also be present on the other rail. Correction would be done via a microcontroller. This simplifies the circuit to a single switching stage that corrects itself and a sense circuit. I have no idea how well this works, if it does. A problem with this would be that accuracy of the voltage outputs will be crude, since the amount of turns rarely turn out to give perfect 5000 mVDC for example. The lack of regulation on each output can be problematic with some sensors which this converter will power.

My other idea was to have a non-isolated dc/dc converter for every single output. Like this:

non-isolated

For the buck-boost I would use the SEPIC topology and for the buck I would consider a synchronous buck converter. The advantage of this is that it won't use a heavy transformer and all the outputs are properly regulated. A problem would be that it requires multiple dc/dc converters.

I would love to know all your knowledge, tips and experience regarding this design stage. If you could elaborate your preference I would highly appreciate it and certainly take it into my final decision!

Isolated vs non-isolated dc/dc in multi-rail design

I want to build a efficient, small and lightweight multi rail dc/dc power supply. I've been thinking about several topologies and ideas. However, the main question of "isolated vs non-isolated" still stands. I am working with a 3cell lipo (11.1V nom. 9.2V min, 12.6V max). So the input voltage will vary over time. I need atleast 12V @ 5A output, 5V @ 5A, 3V3 and perhaps some other specific voltage levels I yet have to receive info about. So my initial thought was quite biased for my love of isolated power supplies: With the following block scheme as my idea:

The idea behind this was to have a single core with multiple output windings. So I thought, since N1:N2:N3:Nn will always remain constant, any offset in the voltage on a single rail will also be present on the other rail. Correction would be done via a microcontroller. This simplifies the circuit to a single switching stage that corrects itself and a sense circuit. I have no idea how well this works, if it does. A problem with this would be that accuracy of the voltage outputs will be crude, since the amount of turns rarely turn out to give perfect 5000 mVDC for example. The lack of regulation on each output can be problematic with some sensors which this converter will power.

My other idea was to have a non-isolated dc/dc converter for every single output. Like this:

non-isolated

For the buck-boost I would use the SEPIC topology and for the buck I would consider a synchronous buck converter. The advantage of this is that it won't use a heavy transformer and all the outputs are properly regulated. A problem would be that it requires multiple dc/dc converters.

Which concept is more suited for building a DC/DC power supply?

1
source | link

Isolated vs non-isolated dc/dc in multi-rail design.

I want to build a efficient, small and lightweight multi rail dc/dc power supply. I've been thinking about several topologies and ideas. However, the main question of "isolated vs non-isolated" still stands. I am working with a 3cell lipo (11.1V nom. 9.2V min, 12.6V max). So the input voltage will vary over time. I need atleast 12V @ 5A output, 5V @ 5A, 3V3 and perhaps some other specific voltage levels I yet have to receive info about. So my initial thought was quite biased for my love of isolated power supplies: With the following block scheme as my idea: The idea behind this was to have a single core with multiple output windings. So I thought, since N1:N2:N3:Nn will always remain constant, any offset in the voltage on a single rail will also be present on the other rail. Correction would be done via a microcontroller. This simplifies the circuit to a single switching stage that corrects itself and a sense circuit. I have no idea how well this works, if it does. A problem with this would be that accuracy of the voltage outputs will be crude, since the amount of turns rarely turn out to give perfect 5000 mVDC for example. The lack of regulation on each output can be problematic with some sensors which this converter will power.

My other idea was to have a non-isolated dc/dc converter for every single output. Like this:

non-isolated

For the buck-boost I would use the SEPIC topology and for the buck I would consider a synchronous buck converter. The advantage of this is that it won't use a heavy transformer and all the outputs are properly regulated. A problem would be that it requires multiple dc/dc converters.

I would love to know all your knowledge, tips and experience regarding this design stage. If you could elaborate your preference I would highly appreciate it and certainly take it into my final decision!