DIY Variable Multi-Output Power Supply Project

[Kippre7790]

I have been learning about switching power supplies, namely buck and boost transformers but LLC transformers as well and I am really excited at making my own power supply now that I see how powerful of a tool they can be. I am learning electronics as I go along so there may be some simple concepts and rules of thumb that I may be missing but I'm not shy to learn. I'm open to criticism and better opinions on the designs but it would be awesome if you can provide explanation on new concepts and confusing ones as I am learning as I go along. Here is a block diagram of the way I'd like to see the power supply function.



The following are the targets I would like to achieve in the design. However it could hit these points is fine with me.

*Can supply +100W of power and a decent amount of current.
*Does not require much heat dissipation. (efficiency?)
*Outputs a wide range of voltage adjustment. (60v?)
*Multiple regulated voltage supply outputs.
*Regulated output for a basic oscilloscope (From China)
*120v AC plug capable of TRIAC/AC PWM adjustment.
*Can be carried in a backpack or something.
*Can be rescaled to larger power outputs without completely redesigning. (optional)
*Limit the use of complex IC's and code based Chips.
*Prefer using only through hole packages where possible.
*Keeping frequencies under 150khz for scope visibility.

I know it's not a small project and all but I want to have a power supply that is good and doesn't have many power limitations.
I will post progress and notable mentions below.
------------------------------------------------------

[Kleinstein]

This sounds like a single complicated large project. For a start there are too many points of failure. So usually it is better to start with separate parts and only put the things together one the single parts work. Other wise it tends to be a project that never gets finished.

The usual way to build a multi channel supply is to have completely separate channels that essentially only share a case and mains transformer. So it would be just more boards of a single channel supply. Separete Supples can be used in series, with a common ground or with a little extra circuit also in parallel.

A SMPS directly from mains can be tricky an is definitely not a beginners project. There are some dangers involved and it needs some experience and suitable gear to work with mains safely. It may be easier here to start with a ready made SMPS to some 24 .. 48 V.

[Kippre7790]

oh yes I am breaking it down but I wanted to provide an overview of my plan.
would a transformer that brings it down to those voltages work then use SMPS to adjust as needed?

[DannyTheGhost]

As stated above, it is indeed very serious project. You still didn't answer several more questions for yourself:
1. More exact total power output
2. What kind of output noise you want to get out of it
3. Do you need isolated output channels (in case if you want to make bipolar supply)
and so on.
I suggest you to look more closely on professional lab PSUs and their topologies to have more clues about what you actually need.

[Kleinstein]

It is a slightly unusual combination to use a transformer and than a switched mode regulation. However there still is some use to it: the 60/50 Hz transformer can give isolation against common mode noise. So it likely would not be as clean as a pure linear supply, but can avoid some of the interference of a full SMPS solution.

If the power level is not that high, one could still go pure linear, and leave the pre-regulation for a 2nd higher power version.

[Kippre7790]

It is a slightly unusual combination to use a transformer and than a switched mode regulation. However there still is some use to it: the 60/50 Hz transformer can give isolation against common mode noise. So it likely would not be as clean as a pure linear supply, but can avoid some of the interference of a full SMPS solution.

I know it doesn't really make sense to run a transformer and then throw the output through an SMPS but I cant think of another way of isolating it. again I'm learning and so I thought it would be better if I started from the mains in and go to the outputs... but it doesn't matter to me where I start so lets say I'll start at the linear supply side for starters and go back. I already made a few current and voltage limiting circuits so I know what I'm doing there so that's fine and dandy. next is the adjustable SMPS for coarse adjustment which from my original drawing takes 60V in and lowers it to X voltage with not much loss. Ok 60V is too high of a voltage. Say I supply the lowest I want to have which is 36v then, doesn't matter to me because my idea is that the general layout doesn't really change just the value of the components changes.

So 36V in and I want it to switch it at about 50khz and be able to adjust the output rate based on the variation from the output voltage, and also be able to adjust the target output voltage via a potentiometer. I made a circuit with a set of 555 timers that did the switching portion but they are not fast enough for what I need. so I guess my first question would be, is there a timer out there that I can use at 50khz? I know there are buck and boost controllers that I can use but I'm not sure how exactly they function. (whether the frequency is adjusted or the pulse width or whatnot)

So lets start there, anyone got any recommendations on a through hole switching IC? or maybe some sort of square wave circuit that would work?

[LeoTech]

Just out of curiosity, why do you have a 9 V regulator followed by a linear regulator for your scope?

Wouldn't it be far more efficient and practical to lower the 60 Volts to whatever the scope needs in one module?

On the account of modules, I would highly recommend to break this project down into smaller modules - i.e. smaller projects - that you could design and build, and then slowly merge them together.

Lars

[Kippre7790]

Just out of curiosity, why do you have a 9 V regulator followed by a linear regulator for your scope?

Wouldn't it be far more efficient and practical to lower the 60 Volts to whatever the scope needs in one module?

well I'm thinking I would like the cleanest power I can get to go to the scope which runs at 9V and pulls about 110mA so if I just made a linear regulator from whatever voltage I decide to use on the rest of the circuit, be it 60v, 50v, or 36v. that's no small amount of waste heat. (since basic regulators dump all the excess power as heat)

So I would switch it down to a more manageable voltage like 12v then use a linear regulator to keep the power going to it clean. it's still wasting heat but nowhere nearly as much as directly regulating.

****EDIT****
Did the math and yea it would probably work because it's only 6 watts of heat to dissipate.

[LeoTech]

Just out of curiosity, why do you have a 9 V regulator followed by a linear regulator for your scope?

Wouldn't it be far more efficient and practical to lower the 60 Volts to whatever the scope needs in one module?

well I'm thinking I would like the cleanest power I can get to go to the scope which runs at 9V and pulls about 110mA so if I just made a linear regulator from whatever voltage I decide to use on the rest of the circuit, be it 60v, 50v, or 36v. that's no small amount of waste heat. (since basic regulators dump all the excess power as heat)

So I would switch it down to a more manageable voltage like 12v then use a linear regulator to keep the power going to it clean. it's still wasting heat but nowhere nearly as much as directly regulating.

****EDIT****
Did the math and yea it would probably work because it's only 6 watts of heat to dissipate.

That's a good idea, considering the heat, but 6 Watts should be doable.

Just keep in mind, that if your scope needs a 9 V input, then the pre-regulator from your main supply has to be higher than 9 Volts, because the linear regulator will cause a voltage drop. Although you probably know that, and just did a typo in the block diagram.

[Kippre7790]

That's a good idea, considering the heat, but 6 Watts should be doable.

Just keep in mind, that if your scope needs a 9 V input, then the pre-regulator from your main supply has to be higher than 9 Volts, because the linear regulator will cause a voltage drop. Although you probably know that, and just did a typo in the block diagram.

Deffinetly, wasn't thinking when I put the 9v there. should have been on the linear but I might just remove that for simplicities sake.

Got any input on the switching supply part of things?

[prasimix]

I'd like to recommend you with a whole heart to use SMD from the start as much as possible. I've started my adventure with THT believing it is more DIY-friendly, but it is not! Prototyping where you need to add, remove and change components often is much easier with SMD. Also if you're going to do something with power switching you can expect a lot of trouble because THT cannot be packed and arranged as SMD parts.
Ok, there is parts like bulk elcos, connectors, etc. where THT is still better choice. Also you don't need to go in another extreme and use QFN and BGA packages that requires stencil, microscope, pick&place, etc.

[Kippre7790]

thanks for the concern, I've never tried using SMD mainly because I can't stick them in a bread board and test them as I like to test most of my circuits before putting them together. That's one of the main reasons which is keeping me from switching.

What do you guys think about using the LM2596 to drive a power transistor to get a higher current output compared to the 3A of the LM2596, or better yet use the LM2596 to drive the transistor and switch a higher voltage? Doubt I'd go over 3A of current but if I can do either of those, then I could use it to supply some low voltage high power applications later on.

Here is the datasheet: http://www.ti.com/lit/ds/symlink/lm2596.pdf

[Kleinstein]

Combining the LM2596 or similar SMPS chip with integrated power stage with an extra power transistor may not work that well. Usually one would than better start with SMPS chips that is made to work with an external power transistor.  So more like TL494/TL497 or uC284x / uC38../LT124x .
A transformer followed by a SMPS regulator is not that bad - it is just not common.

For a beginner I would start low power. If this works it is often a relatively small step to scale up to a little higher power.

[LeoTech]

That's a good idea, considering the heat, but 6 Watts should be doable.

Just keep in mind, that if your scope needs a 9 V input, then the pre-regulator from your main supply has to be higher than 9 Volts, because the linear regulator will cause a voltage drop. Although you probably know that, and just did a typo in the block diagram.

Got any input on the switching supply part of things?

Unfortunately no, I have mostly been playing around with building semi-discrete linear regulator power supplies.

Leo