smps design component placement

[jackbob]

Hello eevbloggers I am planning on building a smps for a class d amplifier kit i bought. I actually bought the kit just for a good load to have for the smps I am building. This is the first professional smps I am building not just one i slap together in a couple days. I bought most of the supplies already so here is what it is. The heart is a TL494 it runs off 12vdc nominal in and puts out -60v 0v +60v. I am using an EL55 transformer which I will wind myself, it will be driven by a pair of irfp2907 mosfets. I can draw a schematic in more detail if anyone would like to see. Of course it has the general safety and filtering components such as diodes on the fets as well as input and output filter caps.  I want to design the supply to its maximum efficiency and have done research. I find that there are ways to put capacitors and inductors on the high frequency of the primary and secondary of the transformer to achieve a resonance. I also noticed some designs put a capacitor in series with the transformer primary as a high pass filter. Can capacitors really pass this much power through them happily? My main question is how can I add capacitors and inductors to increase efficiency of the power supply?  I will post some pictures I saw of basic configurations of these components. The second one is a qsc amplifier psu and the last one is a psu from a pkn controls amplifier, it operates on silgle or three phase, thus three identical power supplies put together.

[MaxPower]

Non of these circuits seems to be correct, I have never seen any capacitor in series with the transformer, they are usually in parallel followed by a resistor and a diode which forms the snubber circuit. Snubber is there to protect against high voltage spikes generated through the primary side of the transformer due to the high frequency switching. 
The series inductor is usually part of the PFC circuit which is a boost converter that tries to increase the power factor. That is where you increase your efficiency. Read about power factor correction. your input filter caps are way too large for an SMPS which will lead to a huge inrush current. The whole point of an SMPS is to increase efficiency while reducing component size.
That is not how you make a three phase SMPS. of course you can parallel three separate SMPS and connect the outputs, but that is not how its done. you need a three phase rectifier bridge and etc.

What topology are you using??? Flyback, Forward, Push Pull, or Full Bridge??? you need to be more specific.   

[jackbob]

These are all 100% working production schematics from actual designs. I agree some seem a bit wild which is why I am here, to find out why. I understand most of how most of it works already such as pfc, however I am not familiar on how to calculate the inductor sizes to get the best power factor. large input caps are needed because it is 120vac rectified and it has an inrush limiter. I actually own a qsc plx3602 (middle schematic) and it sure does have a cap in series with the transformer. It sort of makes sense because caps let ac through and this cap is sized perfect to let the high frequency (100khz) and filter out lower frequency noise. However it seems like a bit of power to be pushing through a capacitor (passive audio speakers use them this technique all the time though). These supplies are different than the one I am designing they are line vac input and about +- 200vdc out. I am running 12vdc in and +- 60 out.  I am using these as examples because of the way they used the caps and inductors to increase efficiency. I am not sure yet whether I should have a center tap primary with + 12v at the center and the other two taps in push pull from the mosfets or just a single winding with one side +12v and the other pwm from the mosfets. Because of low voltage at the primary current will be very high. I will post a quick schematic of the two possible configs. 

[jackbob]

I made a quick schematic of possible configurations for the transformer primary. obviously leaving out other components as i am just showing transformer configurations. Also the transformers are not to scale, but just showing how they are connected. The FET's are irfp150's in the schematic because irfp2907's were unavailable in multisim 12.0. Irfp2907's will be used in the actual project. If i put the two fets in parallel like the first configuration the max duty cycle will be 48% of I use the second configuration the max duty cycle will be 96% (48% per transistor) alternating in both directions. The FET's im using are very high current and will be properly cooled. They are rated at 209 amps continuous (although we all know those max ratings are under unrealistic conditions).

[MaxPower]

There is an easy way to generate the +-60 volts that you need from a single +12 Vdc supply. You can use a full bridge to generate +- 12 volts alternating square wave and a transformer to step it up to +-60 volts. then use two diodes and two caps to get +- 60 Vdc. have a look at my attachment. in this configuration, each MOSFET only operates at 50% duty cycle.

[jackbob]

Your schematic will work but im not looking to go that simple (not saying its bad) BUT that is a perfect example of what i said in my first post. It is something I would build in two days without much design thought.  I want to build this one professionally and take consideration into all the detail of pfc and other small aspects. It should be easily a 600W power supply when its done and I dont want it to draw 800W total for inefficiencies in the design.  I would also like the output power to be as clean as possible (little ripple).  If I add more components such as snubber circuits, pfc, and resonance, then it will be much more efficient and reliable. Too many times I build these things and they work but get steaming hot and eventually a component gives due to stress because i didn't implement any extra safety components. One big thing I'm not sure on is the frequency to operate at. I have seen them operate as low as 20khz and high as 200khz. Most of the time 30-50khz. I did some bread boarding and find 40khz works good, however i'm not sure if it is the optimal frequency. I know as the frequency goes up problems get greater ( transistor configuration, heat on the transistors, less efficient, diodes unhappy) but the reward is more power output in less space. Its a balance.

[jahonen]

That topology with series capacitor seems to be a half-bridge, and it usually has a series coupling capacitor in series of the primary, so nothing wrong there. There is no intentional resonant circuit.

I think that there is no trivial way to improve the efficiency substantially. Maybe using a synchronous rectification at the secondary but it is not a trivial thing.

Regards,
Janne

[jackbob]

That PDF has lots of good info, I will look into it with more detail when I get home.  So the point of the coupling capacitor is just filtering?  I guess there isn't a huge difference even if I put all the fancy components but it still may make it a bit more reliable and stable. One thing I noticed that I haven't seen before... In the second schematic qsc power supply it has diodes at the gates of the FET's where a pull down resistor normally is anybody seen this?

[jackbob]

Yes they are igbt's wasn't paying attention. I will post a better picture of the gate drive. But if you were to see the whole schematic, the gate signal looks as if it would be grounded out by the diode which is the part that doesn't make sense. I will look at flux walking.

[jackbob]

Ah taking a closer look the diodes don't go to ground but rather back to the ic. The driver for the igbt's probably drives negative and positive to switch off the fets and it uses the diodes as a low resistance path to turn them off quicker. I have a better picture attached. The pdf about flux walking had lots of other useful info. Thank you guys for finding good detailed documents with formulas.  Those are what I have been in search of.

[donpetru]

jackbob, you accomplices pointless with three SMPS stages from the first post.  I understand that you want to create a SMPS which is fed from three-phase? If yes, then you have to create an electronic device that has a diagram like the one posted below:

[jackbob]

I am not making a three phase supply but just used it as an example. I am actually making a dc to dc supply.  Why does everyone say the three phase supply is bad?  That supply is a real power supply from an amplifier that is in production.  Here is the link to the amplifier http://www.pknc.com/3phase_eng.html  these amps aren't cheap Chinese things either they have very good specs made in Hungary.  It's really a good idea for live audio to help keep the three phase distro balanced.