Design a 3kW DC/DC smps

_nlc_

New member
Hi all,
I am new here. First, please scuze me if my english is not perfect, it's not my primary language :)
Second, MicrosiM I want to congrulate you about your work and your experiments. I laugh a lot when you often tell that you talk about thing you really tested comparing to other which claims a lot but without show anything :)
It makes me laugh because I am facing the same on some experiments I worked on in the past :D

If I post here it's because a friend asked me if I can develop for him a 3kW DC/DC converter. Input voltage 48V DC ( 42V min 55V max), and output adjustable from 220 to 240V DC.
It's to use in its hybrid car, to extend the range in pure electric mode with a supplementary battery pack.

He first used a kit from Enginer, and had a lot of problem with the converter and even with the battery pack. Now Enginer refuse to change another time it's broken converter under warranty !! It's why he asked to me to try to repair it, it's what I done (but for how many time !?), but also to look if I can build a new one, more reliable.

Because I like discover new area of electronic, I am looking for the maximum of information on SMPS before try to do anything. It's why I found this site and this forum !
I quickly understood the basic theory of this kind of DC-DC converter, but I know it's really complex in practice when we reach these output power !

My first idea to build this converter was to parallelize multiple small converter based on smaller transformer instead to use only one big transformer. Also, my idea was to parallelize the transformers's primary, but put in serie the transformer's output.
After my friend gave my the broken converter from Enginer and open it, I see they use exactly the same technic. Thus I think the basic idea is good, but the enginer realization is not reliable.

So what I want to do is to use a push pull technology with 6 transformers and 12 associated mosfets. If I put the 6 output transformer in serie just before the ultrafast rectifier bridge, the transformer can be 1:1, because with 42V at the input I will have 42x6 = 252V at the output (with max duty cycle)
Do you think this configuration can be a good idea ?

Next, I need to learn about transformers, because I think about winding them myself on standard core ETDxx.
If I use 6 transformer (1:1), one transformer needs to manage 3000/6 = 500W. From your experience, what is the best transformer ETD size for this power ? ETD44 ? From the current side, if I take the most unfavorable configuration (input at 42Vdc), each transformer need to manage 12A.

With the 6 transformers solution, the advantage is that I can first work on a "small" 48Vdc/48Vdc 500W converter based on a single transformer, to optimize the transformer winding, the electronic and the regulation, and after this step I can realize the big converter with 6 transformers with the ouputs in serie.

Thank again for this site and this forum !
Regards,
Cyril
 

MicrosiM

Administrator
Staff member
Smps 3kw

Hello Cyril.

Thanks for joining diysmps, what you see here is effort from all of us, after hard work.

Regarding my statement *Showing True Results* I stick to that because I believe into it.

I dont claim! I only show what happened! others around showing 4000W SMPS while it delivers 1800W RMS, and then when they get stuck they tell its for Audio, so I am fed up from claims, specially in SMPS.

Kindly note that I never came across (Low Power SMPS) units, like the one you are looking for.

You may check these threads as a start, they might contain some helpfull informations.

*http://www.diysmps.com/forums/showthread.php?108-Le-Monstre-Hiraga

* http://www.diysmps.com/forums/showthread.php?33-I-m-looking-to-make-a-smps-for-a-car-amp/page2

I will help you as much as I can, I will read your message again and will put more informations soon!

thank you
 

_nlc_

New member
Hello MicrosiM, thank you very much for your answer.

I think I will be able to develop and test the electronic myself, but the important point I will need help is for the transformers. I have no experience at all in transformer calculation and winding. Actually I read a lot about them (high frequency transformers) to understand all the subtlety, for example skin effect, leakage inductance and so on...
I carefully read your pdf about transformer construction you posted on another topic, and I already have an interrogation. You tell to wind a 1/2 primary on the first layer, then 1/2 secondary, then 1/2 secondary, and then 1/2 primary. But if winding are done in layer, the winding resistance will be different between the two 1/2 primary (and even for the 2 secondary), because wire length will be differents, it's not a problem ?

For a 3000W DC/DC converter, I think I will stay on the multiple transformers configuration, with primary in parallel and secondary in serie. I took photos on the 5kW converter of my friend, I will post them tomorrow. They use 8 transformers used in this way :

- 4 transformers with primary in parallel on 48V DC and secondary in serie + ultra fast diode bridge + 220uH inductor + capacity bank
- 4 transformers with primary in parallel on 48V DC and secondary in serie + ultra fast diode bridge + 220uH inductor + capacity bank (same as the previous one)

I don't know what the transformer are, it's only marked EC53-15. I think this is this one : http://www.gztydz.com/ShowProductEn.aspx?type=1&guid=f0536d25-1774-46d0-8dc6-9c51ac4295be
I think the transformers are not optimised on this converter because with no load at nominal output at 240VDC and input at 48VDC, if I look the mosfet gates signal there is a very very small duty cycle, below 5% if I remember !!

Regards,
Cyril
 

ludo3232

Member
Salut nlc

Vu le drapeau français je suppose que tu parle français? moi aussi :D

Si tu as besoin d'aide pour ta SMPS je suis la..

;)
 

_nlc_

New member
Bonjour Ludo, oui en effet je parle Français, mais il vaudrait mieux parler anglais pour que tout le monde comprenne bien ?
En fait pour tout ce qui est autour du transfo ça ne me pose aucun problème, c'est dans le transfo lui même que je n'ai aucune expérience et que je pense avoir quelques difficultés. Déjà je voudrais décider de quel core ETD j'ai besoin, sachant que je vais avoir 6 transfos pour passer 3000W, soit 500W par transfo. Je ne veux pas utiliser le core le plus petit possible en lui faisant passer le maximum de puissance, mais par exemple utiliser le core au dessus pour avoir moins d'échauffement et donc un meilleur rendement. Est-ce qu'un ETD44 pourrait faire l'affaire pour passer 500W en étant à l'aise ? Ou plutôt un ETD49 ?

Ensuite pour le bobinage, j'ai bien compris que pour avoir le moins de pertes par échauffement, il faut le plus de cuivre possible, mais qu'il fallait utiliser plusieurs fils de petite section en parallèle pour éviter l'effet de peau. Par contre au niveau du bobinage, je ne sais pas encore si je dois plutôt faire le primaire sur une première couche et le secondaire sur une 2ème, ou si je dois faire 1/2 primaire, 1/2 secondaire, 1/2 secondaire et 1/2 primaire. J'ai peur qu'en faisant comme ça les primaires aient une résistance différente et que ça pose des problèmes de saturation du core (je vais être un mode push pull avec le point central du primaire rélié au +) ?

Ludo Hello, yes indeed I speak French, but it would be better to speak English so that everyone understands?
In fact for all that is around the transformer that I have no problem is the transformer itself I have no experience and I think I have some difficulties. Already I would decide what I need ETD core, knowing that I'll have to spend six transformers 3000W or 500W per transformer. I do not want to use the core as small as possible by making him spend as much power, but by using the example above core for cooler running and thus better performance. Is a ETD44 might do the trick to pass 500W being comfortable? Or rather a ETD49?

Then to the winding, I understand that to have the lowest heating losses, it must be possible over copper, but he had to use several small section of son in parallel to avoid skin effect. By cons in the coil, I do not know if I should rather be on a first primary and secondary layer on a second, or if I should do 1 / 2 primary, 1 / 2 second, 1 / 2 and secondary 1 / 2 primary. I fear that by this way the primaries have different resistance and that it poses problems of saturation of the core (I will be a push-pull mode with the center point connected to the primary +)?
 

ludo3232

Member
Bonjour Ludo, oui en effet je parle Français, mais il vaudrait mieux parler anglais pour que tout le monde comprenne bien ?
En fait pour tout ce qui est autour du transfo ça ne me pose aucun problème, c'est dans le transfo lui même que je n'ai aucune expérience et que je pense avoir quelques difficultés. Déjà je voudrais décider de quel core ETD j'ai besoin, sachant que je vais avoir 6 transfos pour passer 3000W, soit 500W par transfo. Je ne veux pas utiliser le core le plus petit possible en lui faisant passer le maximum de puissance, mais par exemple utiliser le core au dessus pour avoir moins d'échauffement et donc un meilleur rendement. Est-ce qu'un ETD44 pourrait faire l'affaire pour passer 500W en étant à l'aise ? Ou plutôt un ETD49 ?

Ensuite pour le bobinage, j'ai bien compris que pour avoir le moins de pertes par échauffement, il faut le plus de cuivre possible, mais qu'il fallait utiliser plusieurs fils de petite section en parallèle pour éviter l'effet de peau. Par contre au niveau du bobinage, je ne sais pas encore si je dois plutôt faire le primaire sur une première couche et le secondaire sur une 2ème, ou si je dois faire 1/2 primaire, 1/2 secondaire, 1/2 secondaire et 1/2 primaire. J'ai peur qu'en faisant comme ça les primaires aient une résistance différente et que ça pose des problèmes de saturation du core (je vais être un mode push pull avec le point central du primaire rélié au +) ?

Ludo Hello, yes indeed I speak French, but it would be better to speak English so that everyone understands?
In fact for all that is around the transformer that I have no problem is the transformer itself I have no experience and I think I have some difficulties. Already I would decide what I need ETD core, knowing that I'll have to spend six transformers 3000W or 500W per transformer. I do not want to use the core as small as possible by making him spend as much power, but by using the example above core for cooler running and thus better performance. Is a ETD44 might do the trick to pass 500W being comfortable? Or rather a ETD49?

Then to the winding, I understand that to have the lowest heating losses, it must be possible over copper, but he had to use several small section of son in parallel to avoid skin effect. By cons in the coil, I do not know if I should rather be on a first primary and secondary layer on a second, or if I should do 1 / 2 primary, 1 / 2 second, 1 / 2 and secondary 1 / 2 primary. I fear that by this way the primaries have different resistance and that it poses problems of saturation of the core (I will be a push-pull mode with the center point connected to the primary +)?

Yes is better in english for forum but is not my first language, sorry for mistakes..

My opignon is a etd44 is sufficent for 500W, i use now a etd49 with 1000W load.. core heat, but no too warm, i can measure temp if you want..

My new 500W smps is based on ETD39.

1 / 2 primary, 1 / 2 second, 1 / 2 and secondary 1 / 2 primary is the better solution for transformer, but in push-pull mode i have never tested.. the two winding are differents length and resistance.. this have probably effects (bad?) with push pull topology. i recommand you experiment first with one transformer to find the best winding solution.

why u prefer push-pull topology?

for winding a recommand you "litzes" wire for best results. For my first transformer i use 5 x 0.4mm (diameter) Wires. i have good results.. later i have make a new transformer with 20 x 0.2mm and these heat a bit less and lower voltage drop at secondary. ( no very significant.. 2v)

Now i use litzes wires 50x 0.2 and 100 x 0.1.. it's costly but best results and easy to wind at transformer and welding..

for good transformer isolation u need kapton tape.

this smps is regulated?
 
Last edited:

_nlc_

New member
Hello Ludo !

My opignon is a etd44 is sufficent for 500W, i use now a etd49 with 1000W load.. core heat, but no too warm, i can measure temp if you want..

My new 500W smps is based on ETD39.

ETD49 @ 1000W :eek: It's impressive !! :UP:
Thus maybe I can use 4 ETD49, 750W each to reach 3000W. In my case the board will be in a closed case but with a forced air flow, with a fan at each end of the box. An elevation of 40°c from ambient is largely acceptable.

1 / 2 primary, 1 / 2 second, 1 / 2 and secondary 1 / 2 primary is the better solution for transformer, but in push-pull mode i have never tested.. the two winding are differents length and resistance.. this have probably effects (bad?) with push pull topology. i recommand you experiment first with one transformer to find the best winding solution.
You are right I think I must do experimentation, but I want undestand the maximum before try.

why u prefer push-pull topology?

Push pull is less complex than full bridge, and don't use the 2 capacitors needed for half bridge. Without these capacitor the bom is smaller and the pcb routing will be easiest I think.


for winding a recommand you "litzes" wire for best results. For my first transformer i use 5 x 0.4mm (diameter) Wires. i have good results.. later i have make a new transformer with 20 x 0.2mm and these heat a bit less and lower voltage drop at secondary. ( no very significant.. 2v)

Now i use litzes wires 50x 0.2 and 100 x 0.1.. it's costly but best results and easy to wind at transformer and welding..

Yes it's the first thing I understood on high frequency transformer, the skin effect and AC copper loss !! Litzes wires is better I will use this technic.

for good transformer isolation u need kapton tape.

Thank for information !

this smps is regulated?
Yes it will be regulated in voltage and current, to have at the ouptut a constant voltage but with current limit (CV / CC regulation, such as laboratory power supply).

Regards
 

_nlc_

New member
The photo of the Enginer 5kW converted (48VDC ---> 240VDC) :



The principle is here :



2 groups of 4 "EC53" transformers with primary in "parallel" and secondary in serie.

I want to keep the same idea, which is a good idea I think.

I dismounted a transformer, the primary is foil based and the secondary is Litz based. I counted 15 turns for the secondary, and 8 turns for each 1/2 primary. But I am not sure, it's maybe 7 turns because the first turn is not really complete.

Photos in next post
 

_nlc_

New member
Photos of one transformer

Second primary pin :



Center primary pin :



First primary pin :



Secondary :



As you can see, the two 1/2 primary cannot have the same length, and yet it's a push pull topology. Curious, normally the core will saturate if current are not exactly the same in the 2 primary winding ? Maybe the current difference is too low to be a problem ?
 

_nlc_

New member
I looking about full bridge which seems to have advantages :

- The transformer is easy to build, only 2 windings, the primary and the secondary.
- I think more power can be achievable because the 2 winding are used all the time
- The flux unbalance will be inexistent in the transformer, but may be present because of the switching mosfet if rdson is not perfectly the same between mosfets
- Mosfets only see one VCC and not 2 VCC as the push pull topology

The disadvantage is that there is 4 mosfet gates to control, with 2 floating mosfet gate. Thus this add some gate transformer in the bom.
But maybe to build a 3000W converter with the full bridge topopogy I can use 2 transformers only, 1500W each ? I look on ETD54 or 59 to see how many power I can achieve with them.

One question, where can we buy copper foil for transformer construction ?

Regards
 

CobraNet

Member
Copper foil is good choice at high currents, and for skin effect.

Full bridge is better

I wonder if you are rebuilding this SMPS for personal use? because a ready made one is cheaper for you

Good pictures
 

Murilo

DIY ADICTED
It's nice to see copper foil winding, i woul'd like to have something like this on my transformer.

I had to make a coil holder (hehehe it's crap!), but with PM87 core i'm able to get more then 10kW in a single core (was placed over a heatsink).


www.epcos.com/inf/80/db/fer_07/pm_87_70.pdf

For switching transistors, you can use IRFP4004, in transformer, maybe with the PM87 and like 10 copper foils in paralalel in two turns for primary, and 12 turns with litz wire on secondary would save a lot of space!

But like you said, two transformers can be a good choice too!

All what i said if you want to rebuild it, but as CobraNet said, isn't cheaper you try to use this which you have?

Good Luck!

Murilo
 

Yosemite

New member
It's nice to see copper foil winding, i woul'd like to have something like this on my transformer.

I had to make a coil holder (hehehe it's crap!), but with PM87 core i'm able to get more then 10kW in a single core (was placed over a heatsink).


www.epcos.com/inf/80/db/fer_07/pm_87_70.pdf

For switching transistors, you can use IRFP4004, in transformer, maybe with the PM87 and like 10 copper foils in paralalel in two turns for primary, and 12 turns with litz wire on secondary would save a lot of space!

But like you said, two transformers can be a good choice too!

All what i said if you want to rebuild it, but as CobraNet said, isn't cheaper you try to use this which you have?

Good Luck!

Murilo


10Kw? can you show some pictures?

Topology? schematics?

that looks cool
 

Murilo

DIY ADICTED
10Kw? can you show some pictures?

Topology? schematics?

that looks cool

I'll open a topic to discuss about it when i get a better internet to upload the photos: :D

I didn't get 10kW on it, like i said, this core is able to handle more than 10kW (calculations), my smps was designed to about 4,5kW (15v @300A).

http://imageshack.us/photo/my-images/31/nucleopm.jpg/

http://img228.imageshack.us/img228/8648/conversor1.jpg
http://img695.imageshack.us/img695/6708/conversor2.jpg
 

Murilo

DIY ADICTED
Hosted on diysmps :UP:
 

Attachments

  • conversor1.jpg
    conversor1.jpg
    37.7 KB · Views: 269
  • conversor2.jpg
    conversor2.jpg
    40.3 KB · Views: 334
  • nucleopm.jpg
    nucleopm.jpg
    27.7 KB · Views: 251

_nlc_

New member
Hi all,

I just received a notification but I see that there is a lot of new messages !! I advance slowly in this project, because I have a lot of work at office and a lot of work at home (I redo the bathroom and 2 bedrooms). I began the smps shematic, I received my ETD59 core and 3C90 material, but I am unable to find the copper foil :(
I sent some message at some manufacturer and I never received any answer :(
 
Top