Lost in supplies. Traditionel vs SMPS ?

[Malmir]

Hi there. I have always been told that traditionel power supplies with huge transformers and large electrolytic bridges is much more Space consuming than a smps, also it is much more difficult to regulate when having large voltage or current outputs compared to smps.
So when i was going to replace a power supply we already got in my Company, I started to do a smps. and the clever people killed my idea pretty fast, then after some reconsiddering I did attempt number 2 and it two was killed before seeing the light.

Now I ended up with 2 question (actually a lot, but some follows the answer to those two).
1 : how to figure out what core is best and most usefull for a specific smps ?
2 : This is a stupid non good smps (according to others), but is it realy that bad ? and if, then why.

View attachment smps.pdf

Some people says it would be much smarter to do this with a primary regulated smps but I have not seen any one yet with a large output voltage span, and I need from 24V to 130V @ 3A. Now back to the comment on sizes. those i have spoken with claims it is impossible to do a 420W smps on 100mm*100mm PCB space, why should that be so impossible. Our traditional only takes 100mm*150mm PCB space.

If the one in the attachment is not good. Could any one guide me to a schematic, that I could use as a start. I need a supply that has :

Universal input : 85Vac - 265Vac
1 or multiple outputs for coverring : 24Vdc - 130Vdc
Current output unaffected by output Voltage : 3Amp

Thanks in advance

Sincerely
Klavs Rommedahl.

 

[KX36]

Your PDF file appears to be broken.

It should be possible to make a 400W SMPS in that footprint. I've seen higher power SMPS in a smaller footprint (with forced air cooling). 117x61x13mm is roughly "full brick" size. The question is more how small can you personally make it without much/any previous experience. I think you might struggle to get it in a brick size, but it shouldn't be too hard to get it smaller than your linear supply.

High output voltage is always a little tricky because you may not be able to use standard schottky rectifiers, and wide input AND output makes for a very wide range of duty cycle which is also tricky and if you want a single output to swing the full range, you might need to add a boost PFC just to get a regulated bus voltage to work with.

 

[Malmir]

Dont get it, i just open the PDF ???
What the heck. In that design i do have 4 equal outputs from the ac/dc part and these outputs are then managed by each own dc2dc device, to lower the power waste from one device. The "ac" output was ment to be around 15 - 35V in span and as they are equal, they would hopefully follow each other or rather the one regulated.

I have done several pcb designs for SMPSs and switched amplifiers for ICE Power and others. That part is not what i fear . I am used to have a complete pc on a credit card sized board, so minimization is not the problem, and i am aware of the trace 2 trace distances for safety and so on.

 

[KX36]

I can open the file on my phone, but my home (adobe 9)and work PC (very old adobe) say it's corrupt, so it must be a very version specific file.

The schematic is not particularly clear, so it's hard to see what you're trying to do. It looks more complicated than I expected of your original specs of

Universal input : 85Vac - 265Vac
1 or multiple outputs for coverring : 24Vdc - 130Vdc
Current output unaffected by output Voltage : 3Amp

Now you're talking about AC outputs, multiple outputs, multiple stages of regulation etc.

What exactly did the "smart people" say you'd done wrong?

 

[Malmir]

What people told me was : It would be much easier to do the regulation at the primary site instead of at the secondary site as i do. And I could control the duty cycle to reach the span i needed.
My idea here was if 130V equals 100% or close to duty cycle, how far could i go down before the controller says it is to low and stops. And worse if max duty cycle is 80% or lower, then it is a very narrowed duty cycle span to control a 106V span. Some controllers stop at 25 % duty cycle. How easy would it be to have a span from 24Vdc to 130Vdc, controlled by a duty cycle span from 25% to lets say 78% ?

 

[KX36]

If you have a PFC to take care of the input voltage variation, it's not that difficult to have such a range of output voltages. You'd have to use a push-pull, half bridge or full bridge so that you can have duty cycles above 0.5. PP is less suitable for high input voltages, so HB or FB. I've drawn a rough draft as a half bridge. There's still plenty of work for you to do if you want to use this circuit, but I don't see a reason why it can't be done. There may be the odd mistake in there too as I've rushed it but hopefully nothing major.

 

[renga92]

Hi Everyone. I am almost trying to build the same type of power supply which varies over a wide output voltage. I am using Full bridge to achieve the same and UC3525A instead of UC2525A. I am struck with the output filter design. Can you guys suggest me how to go about the filter design for a wide output voltage ?

 

[KX36]

Firstly what's the ratio of Vout_max to Vout_min? If i's up to 4:1 you should be OK with a duty cycle ranging from 20-80%. 5:1 is proabably OK, but if you end up with duty cycles below 15% or above 85% (>5.67:1), that's not so easy to implement. Calculate your transformer turns ratio for minimum Vin and maximum Vout at Dmax. Calculate your inductor based on dI/dt=V/L. assuming it's CCM, I'd probably aim for 40% ripple current at 50% duty cycle, when ripple is highest, so dI=0.4*Iload, V=Vout(@D=0.5), dt=0.5/Fs (where Fs is the inductor switching frequency which is double the FET or transformer frequency). Capacitor value is not critical, but you can do a rule of thumb calculation for ripple voltage. Don't make it larger than necessary. Then together with the ESR of the capacitor, the input voltage and the turns ratio, you should have enough data to plot the frequency response of the power stage (plant) and design a compensator (type III for voltage mode control) by whatever method such as by following the ON semiconductor TL431 guide listed in my signature.

 

[renga92]

Hi KX36. I just made the ratio calculation and wanted it to vary from 20v to 200v. So the ratio is 10:1 which out of the normal range which you have said. Can you suggest why having larger ratio is bad or difficult to make ? I guess I can it get down to 5:1 at max.

Thanks
renga92