High power SMPS

Thread Starter

Calleb

Joined Jan 21, 2016
5
Hi everyone,

I would appreciate some advice/criticism about my current idea of a high power SMPS. I need to design high power source about 600W (preferably more than that) with voltage range from 0 to 100V and resolution about 0,5V. The voltage should be adjustable in range mentioned (0-100V) and the output current varies according to the load. So it's a "simple" adjustable voltage source with some current maximum/limit.
Right now my idea is that I should stick to some known ATX designs and modify them to fit my needs. So bridge the input voltage to high voltage and then use a half bridge topology output. To be specific I consider using SG3525 PWM driver and IR2110 H/L side driver.
Do you think that this is realistic considering the topology and used drivers? Or should I use a full bridge topology? I am worried about instability with high output power and I am not even sure if I can get the output voltage range that I want.
I appreciate any advice... Thank you very much...
 

wayneh

Joined Sep 9, 2010
17,496
I need to design high power source about 600W...
Let's start with "why?" I mean, any DIY power supply is going to be a reinvention of the wheel. Power supplies come in infinite variety and commercial products are going to be cheaper and better than anything you design. (No offense, just a fact of nature.) Why do you want to design your own? Your answer will help folks here understand your needs and specifications.
 

Thread Starter

Calleb

Joined Jan 21, 2016
5
My answer is why not... I need a powerful power supply and this seems to me like a good idea for gaining some new knowledge, experience and practice...
And to be fair I didn't came across anything similar to what I am proposing... I found a lot of high power SMPS but only fixed ones... Maybe Vicor have some high tech modules but they aren't cheap...
I am open to any alternative but from my point of view it seems to me that I have to do it myself...
 

Dodgydave

Joined Jun 22, 2012
11,284
Either the SG chip or the TL494 are good to work with, most atx psus i have modified to variable will go upto 28V dc,by altering the feedback loop, there will be a minimum output voltage of around 5V, if your going upto 100V its bigger fets bigger pulse transformer bigger output capacitors.
 

ronv

Joined Nov 12, 2008
3,770
My answer is why not... I need a powerful power supply and this seems to me like a good idea for gaining some new knowledge, experience and practice...
And to be fair I didn't came across anything similar to what I am proposing... I found a lot of high power SMPS but only fixed ones... Maybe Vicor have some high tech modules but they aren't cheap...
I am open to any alternative but from my point of view it seems to me that I have to do it myself...
Isolation transformers can be pretty cheap. You don't say how much current you are after so it might be easier to just use an isolation transformer and a buck converter than to get special transformers for a high voltage switcher.
 

tcmtech

Joined Nov 4, 2013
2,867
Although I would not go after trying to build a SMPS based wide voltage range DIY power supply it believe it is possible.

Large inverter based welders and plasma cutters easily handle that range of output voltages while providing far higher currents so yes it is possible to build such a device and have it work with good stability.

You are on the right track with the H bridge design being that's what they use but they typically add a low ESR high voltage AC capacitor in series with the transformers primary windings.

By doing that voltage and current control can be done by both PWM and frequency shifting which when properly combined and controlled can give you very good output regulation.

By using switching frequency variation the equivalent impedance of the series capacitor can be varied and PWM under that can be used for fine voltage or current stabilization and control depending on what you are after.

With the right circuitry most any driver IC like the SG3525 can do both functions simultaneously with only the addition of a few op amp comparators and a master frequency source like a 555 timer.
Part of the SG3525 feedback loop controls the PWM and a second part of the feedback loop goes to the master clock to adjust the primary switching frequency if the output voltage or current goes beyond what the PWM feedback loop can control.

It's just something to think about.
 

Papabravo

Joined Feb 24, 2006
21,159
If "learn by doing" is your goal, then a bit of advice from the DIY grinding of telescope mirrors. If you want to make a 12' mirror, you'll get there faster by grinding a 2' mirror, a 4' mirror, and an 8' mirror first. In power supply terms you'll get to the high power supply faster and more efficiently by starting with a series of more modest targets. Don't believe me? Then knock yourself out with design, component selection, layout, and oh....probably smoke.
 

Thread Starter

Calleb

Joined Jan 21, 2016
5
Thanks for your responses...
I think I am not that off regarding the feasibility of the design. Like tcmtech said welders etc. should have use something similar. My goal is to make it happen and learn something doing it but I really don't want to get stuck in the middle. Like I said...I want to make it happen...
 

Papabravo

Joined Feb 24, 2006
21,159
I don't understand "stuck in the middle". Does that mean you don't really want to learn what you need to know to accomplish your design? If this kind of design was easy, then everybody could do it, but they can't. It is not quick or easy and its does take substantial time and effort. In any case, I hope you succeed and reach your goal.
 

Thread Starter

Calleb

Joined Jan 21, 2016
5
By "stuck in the middle" I mean that for example the result/parameters that I want is impossible to get with the selected topology...
The advantage of my design is that I can quite easily convert half bridge into full bridge by adding another IR2110 driver. But I am really concerned by minimal output voltage and the voltage range in general...
 

Papabravo

Joined Feb 24, 2006
21,159
I still think you'll get there faster by playing "money ball" rather than swinging for the fences. I would concentrate on getting the required current and high efficiency -- that is where the real challenge is. Next I would concentrate on the magnetics, third the search for capacitors with ultra low ESR. After those three are nailed down you can go for any other bells an whistles you like.
 

Thread Starter

Calleb

Joined Jan 21, 2016
5
ehmmm... can you elaborate it a little bit? :) I am not from US nor UK so this kind of expression is new to me :)
 

Papabravo

Joined Feb 24, 2006
21,159
"Swinging for the fences" means trying to hit a home run everytime you step up to the plate. In football(soccer) it would be like trying to score a goal everytime you touched the ball.

"Money ball" refers to a process of getting the most efficient use out of the resources you have available for the lowest possible cost in time and material. In actual practice it was putting together a championship baseball team from players nobody else wanted. You have a set of goals and objectives and I'm advancing the hypothesis that you will get there faster by concentrating on things that are important in SMPS design, leaving the rest until you have accomplished the critical stuff first.

I really don't care if you take my advice or you don't. You're in a much better position to evaluate the alternatives. Just consider it.
 

tcmtech

Joined Nov 4, 2013
2,867
To be honest the simplest way to make a wide voltage range power supply is to use a variac to make a basic AC to a basic DC power supply then use a standard linear voltage regulator behind that for the finer voltage control.

Years ago I had a 0- 3000 VDC 450 ma regulated HVDC power supply that worked that way. It just used a large variance going into a step up transformer followed by a large vacuum tube based linear regular stage.

To keep things fairly efficient the unregulated DC stage only ran a few tens of volts above the regulated DC outputs voltage at full load keeping the heat losses in the huge vacuum tube to a minimal.

A very easy concept to replicate into a rugged reliable lower voltage form with solid state components for the linear voltage regulation stage.
 
Top