50W DC/DC converter

[Dromroe]

I want to design a 50W DC/DC converter .

Spec

Input voltage : 5-25V
Isolation : Input isolated from Output
Type : Isolated Boost
Output Voltage : 28V - fixed by load connected to DC/DC converter output
Input Current : 4.0A 4.5A or 5A - selected by the user .DC/DC converter sets input current
Power: 50W

Questions :
1. What is the best topology . Is push/pull the best for for this
2. Has anyone got a design that could used or simply modified for this application
3. How can it be done to minimise the cost ?
4. Is there an off the shelf transformer i could use or modify ?

 

[IRstuff]

Is this for school? You can buy one of these fron Vicor.

TTFN
faq731-376

 

[Dromroe]

It needs to regulate the input current 4.0A or 4.5A or 5.0 A

The Vicor products regulate the output voltage

 

[OperaHouse]

If you have an input voltage of 5-25V and constant current for an input, how do you have a system where you have 28V out. Your spec seems very confused.

 

[Dromroe]

Hom Opera house . My spec is not confused . Think about it some more

 

[OperaHouse]

4A x 5V = 20W and 4A X 25V = 100W and you want a 50W converter. Why don't you just spell it out for us.

 

[IRstuff]

It's confused because the OP wrote input current instead of output current

TTFN
faq731-376

 

[Dromroe]

Agreed, the input current spec is confusing
The power is limited to 50W .
Therefore with 25V input the current will be limited to 2A .
The input current should be specified as 2A to 5A : 25V = 2A, 20V = 2.5A , 15V= 3.3A, 10V= 5A , 5V= 5A
The DC/DC converter regulates the input current .

 

[OperaHouse]

"Think about it some more",,,,,,, Indeed we have and more than once, If others have not noticed. this appears to br a continuation of nother thread. As pointed out by IR, nput current can be "regulated" by limiting output voltage and current. Last time output was tracking the input. This is a paper napkin design . but the devil is in the details like what happens when input goes under 5V. Choice of components also gets limited when you go down to 5V.

Stick a fork in me.....I'm done. Your spec is like nailing jelly to a tree, It is clear you haven't thought about it.

 

[Comcokid]

Back to the OP questions.
Questions :
1. What is the best topology . Is push/pull the best for this
Since you want to isolate, which requires using a transformer, an buck-derived, boost-derived, or flyback is possible. So, at 50 watts you could use any topology - flyback, forward, push/pull, full bridge. Choice becomes one of overall efficiency, complexity, cost, EMI, and such.

2. Has anyone got a design that could used or simply modified for this application
First – it’s best to check the design examples and application notes from the manufacturers out there. Texas Instruments (who also has the formerly National semiconductor and Unitrode parts and design examples/tools), Linear Technology, Micrel, etc.

3. How can it be done to minimise the cost ?
Minimization of cost is a very iterative process of paper design-analyze cost, paper redesgn-reanalyze cost, analyze-tradeoffs, finally working up to a working design. For minimum cost you will probably find a flyback topology is best, but a flyback will have the worst efficiency and EMI.

For the topologies I listed in question 1, they go from (generally) lowest efficiency to highest efficiency, lowest complexity to highest complexity (which translates into cost), and highest EMI to lowest EMI.

4. Is there an off the shelf transformer i could use or modify ?
There are winding companies out there that make transformers for the design examples that the IC companies publish. There are transformers made by companies like Pulse that IC companies will publish an example built-around. Otherwise, almost all DC/DC SMPS when you begin to build more than few bench items get into designing a custom transformer, which means conquering the skills of magnetics design. Only when doing AC/DC designs at 50Hz or 60Hz do you find a large supply of ready-made transformers. For switch-mode, there are only a few pre-designed transformers available, and some of the ones that vendors advertise are really non-available 'examples' that they publish as a starting-point for discussions on doing a custom. However, 50 watts is a common power level, and you Vin/Vout is fairly standard so there might be some off-the-shelf available.

Don't overlook the online tools and downloadable software that is out there that is available at Linear or TI (National). You may be able to plug-in your numbers and have it pop-out a design complete with BOM.

Of course, I will point out page one of Linear Technology application note AN73. Do your sanity check and proceed if you must. Doing SMPS design is not just another project to get out-of-the-way, it is, in itself, a complete and all-encompassing career-path if you want to do it well.




I just love pointing people to AN73 page 1 every time this type of question comes up on Eng-Tips.

 

[IRstuff]

I will rephrase your spec so that it's understandable. You have an unknown load that you want to program the output voltage so that you draw 50W from the input source which varies from 5V-25V, assuming that you weren't refering to the output power.

So, this means that your orginal current spec was not plausible, and your clarification didn't improve it much. Your input current range should be continuously variable from 2A-10A to achieve 50W. The end result is that your obsession with the input current is misplaced; this is not a current regulator, but a power regulator. Your regulator must actually control the output voltage so that 50W is consumed from the input. And this will essentially limit the input current, but not necessarily control it. Your regulation algorithm must therefore start with 0V output and crank it up until 50W is consumed on the input side and maintain the power level regardless of how the load might change.

TTFN
faq731-376

 

[OperaHouse]

Well, I think the maximum input current of 4-5A should be added to that along with a maximum output voltage of 28V. Generally a regulator should have a low voltage shutdown to guarantee a fet will have a guaranteed turn on. Most chip regulators will have a shutdown of 8-12V. Who knows how hard a spec that 5V is, It wasn't in his last thread. It all just sounds like Dr Phill's last book, "Ten Ideas I Just Pulled Out Of My ....."

 

[DHambley]

I interpret your spec to be this:

1. A constant current from the 5-25V input, programmable by the user.
2. A limit of 50W, controlled by your control circuitry.
3. The output load has some controller to keep its V at 28V (sounds like a battery)
4. ....and the whole thing is isolated.

I don't know what the purpose of your circuit is, but you could use a current-fed push-pull with the Weinberg topology and regulate your PWM based on the input inductor current.

 

[Dromroe]

Hi Hgldr

Sorry if my spec was not clearer but your understanding is pretty good

The converter topology you proposed seems complicated . Why have you proposed this topology ? What are the advantages of this topology over say a simple push/pull converter

Couls i use a Push/Pull converter that controls the input current based on a current demand setting the user supply's

 

[DHambley]

The only reason I suggested a current-fed converter is that I assumed you may want a constant current. (again, I don't know what the input current should look like). A simple push-pull will work also. If this current input is critical, it's only a matter of some analog circuitry to measure the average input current and create a feedback scheme which utilizes that signal in order to control your PWM.

 

[IRstuff]

There should not be any "programming" by the user. If you're limited to 50W draw, then your regulator should be able to handle this on its own, by adjusting the output power until only 50W is drawn from the input.

TTFN
faq731-376