0-50V 0-10A Variable SMPS

[fethiyeli]

I have designed adjustable output voltage and curent SMPS having 500W output. I would like to share circuit schematic since there are many experienced people in SMPS and power electronics in this forum. Do you have any suggestion about this circuit ? or Do you see any critical fault ? This project may help to people who want to build a project something like this.

I have calculated the components value according to some power electronics and transformer textbooks.

 

[schmitt trigger]

First of all, thanks for sharing a very complete design.
You must be very proud of your work.

My only comments are, why would you use a 1N5822 (D3,with only 40V rating) as the bulk capacitor pre-charge? Doesn't it go into breakdown?

Also D5, D6 should be schottkys, like the ones you show on D29, D30.

 

[fethiyeli]

First of all, thanks for sharing a very complete design.
You must be very proud of your work.

My only comments are, why would you use a 1N5822 (D3,with only 40V rating) as the bulk capacitor pre-charge? Doesn't it go into breakdown?

Also D5, D6 should be schottkys, like the ones you show on D29, D30.

Thnx for your comment. There is nothing to be proud of my work i think, because it is not a professional project. But thank u very much for this good comment.

I used 1N5822 because i have seen similar diodes in ON semiconductor app notes as i remember but i am not certain. Actually, i dont know what the criteria is for selecting pre charge diode?

Yea, D5 and D6 should be schottky as well. I change them soon.

It is very good thing to take a comment from experienced person like u.

Thnx.

 

[treez]

hi , can you get it on pdf?, it is hard to look at in that application, we can only see bits at a time. I reduced it in paint, but cannot see part numbers now. I don't see snubbers across the two output diodes. Your PFC output diode is like one of the proper pfc type ones from eg Infineon.?
Cant see all in one , but at moment looks like 2 transistor forward.
Rem not to use schottky if it will get overvoltage transient. schottky definitely need snubber.
Cap on output of curr sense tran is very big (10uf)

 

[Fragrance]

Hi

complete design with pcb and all info for 0-60 0-10 adjustable smps

Regards
Fragrance

 

[fethiyeli]

hi , can you get it on pdf?, it is hard to look at in that application, we can only see bits at a time. I reduced it in paint, but cannot see part numbers now. I don't see snubbers across the two output diodes. Your PFC output diode is like one of the proper pfc type ones from eg Infineon.?
Cant see all in one , but at moment looks like 2 transistor forward.
Rem not to use schottky if it will get overvoltage transient. schottky definitely need snubber.
Cap on output of curr sense tran is very big (10uf)

Hi i can upload pdf version but u can download the schematic via clicking the image. Even i was on mobile, i was able to download the schematic and see all the components.

My Pfc is classic boost type pfc of on semi. I put snubbers across the windings of the transformer rather than putting across the diodes. The values of the snubber components were selected arbitrary because there is no specific calculation for these snubbers. Best one is to measure the oscillation between transformer windings and calculate the components according to oscillation. Yea main converter is 2T forward converter. There is no need a snubber for schottkys for now i guess. Is there any disadvantage of using very big cap for output of current transformer ? but i know its value is pretty much.:-D

 

[Dan Mills]

The transorbs to ground (And MOV between neutral and ground) are not somewhere I would go, they offer little protection and increase leakage current.

Your DC aux supply has the low side of the bridge grounded, which is surely wrong.

1mH for a boost inductor? Just how slowly are you running the PFC stage?

Some of those small transformers for axillary supplies might be hard to wind respecting mains isolation requirements, particularly FB2 where the 8 turn winding must be isolated from the other low voltage winding.

Regards, Dan.

 

[fethiyeli]

The transorbs to ground (And MOV between neutral and ground) are not somewhere I would go, they offer little protection and increase leakage current.

Your DC aux supply has the low side of the bridge grounded, which is surely wrong.

1mH for a boost inductor? Just how slowly are you running the PFC stage?

Some of those small transformers for axillary supplies might be hard to wind respecting mains isolation requirements, particularly FB2 where the 8 turn winding must be isolated from the other low voltage winding.

Regards, Dan.

MOVs were put according to Switchmode Power Supply Handbook 3rd Edition. In electronics, sometimes there is no certain information about specific things. For example, u say MOVs offer little protection and others or some books say that it must be put between lines for extra protection. So, it makes me uncertain about what to do But i agree that they increase leakage current.

Yea, low side of bridge is grounded but if it is realy wrong, i can remove the ground. I agree also isolation requirements but especially between 15turns and others. Since 15 turns are not grounded. It shares the same reference point with UC3844.

PFC inductor was calculated as 895uH and it was rounded to 1mH.

 

[Dan Mills]

Nothing inherently wrong with the MOV between the lines, it is the stuff to earth that worries me as apart from anything else it makes the supply fail an insulation test.
But, yea there are differences of opinion in this area.

What frequency are you operating the PFC at?

1mH sounds very large to me, and the huge gap will make for significant copper losses if you are not careful with the winding layout.

Regards, Dan.

 

[fethiyeli]

Nothing inherently wrong with the MOV between the lines, it is the stuff to earth that worries me as apart from anything else it makes the supply fail an insulation test.
But, yea there are differences of opinion in this area.

What frequency are you operating the PFC at?

1mH sounds very large to me, and the huge gap will make for significant copper losses if you are not careful with the winding layout.

Regards, Dan.

Actually, i have little bit knowledge about insulation test and EMC.

PFC is operating at 67kHz. It is 67kHz version of NCP1653. This low frequency makes the coil large in value. All of the transformers and choke coil have been calculated according to Transformer and Inductor Design Handbook - 4th Edition. Fringing flux also has been taken into account. Huge gap might be a problem for fringing flux and can reduce the efficiency. To reduce this problem, number of turns was decreased to 51 from 93. The book calculates the gap for 93 turns at first, then reduces the number of turns after gap calculation.

- - - Updated - - -

First of all, thanks for sharing a very complete design.
You must be very proud of your work.

My only comments are, why would you use a 1N5822 (D3,with only 40V rating) as the bulk capacitor pre-charge? Doesn't it go into breakdown?

Also D5, D6 should be schottkys, like the ones you show on D29, D30.

Yea u re right. Now i changed the diodes. I used MUR460 for pre charge.

 

[treez]

I believe you need a resistor in series with primary of gate drive transformer to cut down ringing between the series cap and the primary mag inductance of the gate drive transformer

 

[fethiyeli]

I believe you need a resistor in series with primary of gate drive transformer to cut down ringing between the series cap and the primary mag inductance of the gate drive transformer

I see. I thought due to resistors at secondary outputs, there is no need to put a series resistor. But ur opinion makes sense to me.

 

[treez]

yes you still need in primary for C(block).L(mag) resonance

 

[fethiyeli]

yes you still need in primary for C(block).L(mag) resonance

4.7R is enough i think or i can drive a mosfet which drives the gate drive transformer. I see the mosfet implementation in high power applications. Do u think i need that implementation ?

 

[treez]

well you can check in simulation if you wish..
The attached is ltspice simulations of your 2 tran forward.
One runs very slow because it has non untiy coupling in the transformers, the other is quicker , but less realistic, as the coupling in the transformers is unity.
The gate drive tran should ideally have r series = sqrt (L/C)
Where L = mag inductance
C = series cap

- - - Updated - - -

**broken link removed**

I think the above type of PNP's that have high hfe at high IC, are best for the gate turn off use, ie, preferable to the bc327's that you are using.
The FMMT549 datasheet also specifies the switching times, whereas the bc327 does not.
You need a switching transistor for gate turn off, bc327 is stated as being an "amplifier" transistor....not the right type for your "switching" purpose.

 

[fethiyeli]

well you can check in simulation if you wish..
The attached is ltspice simulations of your 2 tran forward.
One runs very slow because it has non untiy coupling in the transformers, the other is quicker , but less realistic, as the coupling in the transformers is unity.
The gate drive tran should ideally have r series = sqrt (L/C)
Where L = mag inductance
C = series cap

- - - Updated - - -

**broken link removed**

I think the above type of PNP's that have high hfe at high IC, are best for the gate turn off use, ie, preferable to the bc327's that you are using.
The FMMT549 datasheet also specifies the switching times, whereas the bc327 does not.
You need a switching transistor for gate turn off, bc327 is stated as being an "amplifier" transistor....not the right type for your "switching" purpose.

I know using ltspice, orcad, multisim, proteus, psim etc. but i have always had annoying errors i couldnt fix. Therefore, i am not simulating this type of projects anymore but i may check them again later.

I have seen many times that bc327 or similar transistors had been used for gate turn off. Thats why i used bc327 for gate turn off. I even havent done a research for better one. Maybe u re right there might be a better, suitable one for switching. I will take consider ur opinion. Moreover, i couldnt understand anything from ur simulation text attachments. Sorry.

 

[treez]

they are .txt files, because this forum does not accept .asc files, you just have to convert them to .asc files then run them on ltspice.

 

[fethiyeli]

they are .txt files, because this forum does not accept .asc files, you just have to convert them to .asc files then run them on ltspice.

Ok bro thnx. I am gonna look at them tomorrow. Thnx a lot.

 

[fethiyeli]

they are .txt files, because this forum does not accept .asc files, you just have to convert them to .asc files then run them on ltspice.

I did what u have said and simulated both of them. Simulation results are quite good but may i use LT IC for UC3844 ? Are they similar ? If they are, i can simulate the most of the circuit, especially, "control, power, output" and "voltage and current control" stages. It may give me pretty good results.

- - - Updated - - -

In another forum, they suggest me to put dummy load to SMPS output. How much in value do i need to put? If yes, then i need to put another dummy loads to flyback one.

 

[treez]

Yes, also, i think the datasheet of LT124X does say that LT124X are pin for pin compatible with uc384x series.
I think the UCC38c4x series is also the same but draws less bias current, so you may prefer it.
Remember also you only need the 50% duty one as its 2TFC.
There is also the NCP12XX series which may do the trick.