Making of Lightforce 1 Class D Powerbriefcase

[Pierre]

Well, as you have guessed, in fact I am building a half-bridge SMPS too...
Have a look at some threads I have started in the Top->Power Supply Design forums.
You will find very useful info there.

Of course I will be a pleasure to contribute with any ideas.
It would be particullarly interesting to see how you solved the gate drive issue. It is quite a nightmare with integrated chips, and a bit expensive with transformers.

[Workhorse]

Hi Lars Clausen,

After seeing your circuit , i have some points for your SMPS..

1 The Bus Capacitor is of very low value, it should be at least 1500MFD

2 Why did you implement single capacitor accross the bus, you might use 2 series connected 1500MFD/250V caps in order to ensure some headroom for overvoltage and also it facilitates in providing a centre tap for connecting the primary of transformer.It also provides the provision to configure the circuit for 110V//220V mains supplies by the virtue of voltage doubler principle.

3. The relay + bypass resistor used for softstart should be placed before the Bridge rectifier in order to charge the Bus Caps slowly, in your circuit it is placed in series with primary winding which offers little task...

regards,
K a n w a r

[Pierre]

I agree with Kanwar except in one point: it is not necessary to have double input capacitors: the way Lars have done it is also correct and widespread in the industry. It only has one disadvantage, the doubling possibility for 120V countries is voided, as Kanwar has also pointed out.

About the capacitor value, there is a common rule of using around 1uF per watt, so I guess that for 400W output, 470uF would be enough. Of course, that's the total equivalent capacity, if you put 2 470uF/200V caps in series, you have half an equivalent 235uF/400V capacitance!
As supply is unregulated, capacitance shouldn't be lower than that for decent stiffness of the output.

Apart from gate driving, (we hope Lars will show how he has solved it, and I guess it will be a very simple and effective way, as always), other things to watch out are input line filtering.

BTW: almost all SMPS's for audio over there use full-bridge rectification. I think it can work well with only 2 diodes, but what do you think?

[Lars Clausen]

Thanks for all the useful input. About the bus cap i have given this a great deal of thought. Because of course i want a very stiff supply on the secondary.

Many use 1 uF caps on the secondary side which is enough to remove most of the HF ripple. However i am using 10000 uF because this way i can maintain some local 'stiffness' after the transformer. Not least to keep control of the rail pumping effects from the Class D modules.

So i am to some degree accepting 50Hz charging of my secondary caps, and not depending on a clean stabilized output voltage for my (otherwise) 1 uF caps. This means when i reduce the bus cap i am only trading in a small portion of the transfer effeciency and transformer impedance. But then this is very low to start with, so i am guessing it will not be a big problem.
In other words due to the big caps on the secondary, the charging load drops to almost zero at just a few volts under the peak voltage, so a bigger bus cap wouldn't make a difference anyway. Unless it was VERY big...

About using the 2 secondary diodes, instead of 4, this is of course a very tricky area, especially combined with the pumping effects and the varying load. My conviction at this point is, that the DC level of the 1 uF cap will be held steady by the 50% duty cycle of the driver system. And so can only charge evenly on the secondary side. But we will see, when the circuit is completed....??

[Pierre]

OK, keep on and tell us your results, Lars.
I hope to be able to contribute with my experiences in this area.

[Pierre]

Some more comments before I leave...

Lars, your core is a RM14, with effective area Ae=2.01cm^2.
If you use 15 primary turns and 320V rectified mains voltage, your peak flux density is:

B=(Vbus/2)*1E8/(4*f*N*Ae)=160*1E8/(4*250000*15*2.01)=500G=50mT

At that operating point (50mT, 250KHz), you must be using a high-frequency material, at least 3F3, in order to reduce core losses, aren't you? (3C90 material won't work well at that freq, for example).

And, according to some references, max. available output power, with that core and operating point is around 750W, not bad for its small size!

[Workhorse]

Originally posted by Lars Clausen

Many use 1 uF caps on the secondary side which is enough to remove most of the HF ripple. However i am using 10000 uF because this way i can maintain some local 'stiffness' after the transformer. Not least to keep control of the rail pumping effects from the Class D modules.

Lars, High value caps at secondary side of High frequency switching supply....is this a good practise...what do you think??

[Lars Clausen]

If i was making analog amplifiers, i would of course use a smaller cap. But for the reasons already posted, i think it's a good solution.

What do you think?

[phase_accurate]

I was once thinking of making an unregulated PSU for an amp too. This has the advantage of simplicity (and intrinsic stability)over a regulated one and of course the advantage of size/weight over a conventional one.

I am not sure but maybe a PFC followed by such an unregulated PSU might also be an attractive solution for audio purposes due to 4 features:

- Could be made working on any mains voltage
- has increased regulation
- can be made CE compliant
- soft start can be made intrinsic to the PFC

Regards

Charles

[Lars Clausen]

Hello Charles, and welcome to this thread

- soft start can be made intrinsic to the PFC

Yes i have seen this claim by some vendors of PFC chips, but to me it seems the only thing that is limiting the DC current is the PFC choke, and that will not last more than a few milliseconds. So how does that work?