low voltage high current power supply

[mr_byte31]

Hi All,

I am planning to build a 5v power supply which can provide at least 50A. I found many project on internet that use MOT (Microwave oven transformer)
it can produce a lot of current. I am focusing now on the losses and low ripple DC voltage.

I am planning to use Pi-Filter to remove the AC ripple voltage (Forget about values in figure, I just add it as reference):

**broken link removed**



I will configure the secondary windings to have 6.4 v AC to overcome the losses on the rectifier

I have two points here

losses:
-------
The bridge rectifies would reduce waste 2x0.7v = 1.4 v
The losses would be 1.4 v x 50 A = 70W
the remaining voltage would be (6.4-1.4=5 v), the power here would be = 5 v x 50 A = 250 W
we get a efficiency of approx = 1- 70/250 = 72% !! I think this is low
any idea how to improve this ?

AC ripples:
-----------
is it possible to have a very low AC ripples at this high current using Pi-filter? (forget about the C and L values in figure)

 

[schmitt trigger]

There are several incorrect assumptions, which I will allow other forum members to point out.

I'm going to explain only what is wrong with choke weight and cost.

See the attached image. It is a 5 amp choke I have. From the web page, this weights 14 pounds.

Now, on chokes, the energy capability (which is directly related to volume) goes with the square of current, e = 1/2*L*I^2
A 50 amp choke of the same inductance, would be 100 times more volume.
Since weight is directly proportional to volume, and by maintaining the same copper/steel ratios, then your proposed choke would weight a whooping 1400 pounds.

You could halve the inductance, or you could go further and reduce it by 10......it would still be 140 pounds.

Now for the price. This one retails for $93 US dlls. Imagine how much your proposed one would cost.

 

[FlapJack]

schmitt trigger, you have a bad attitude. 1400 lbs is less than the curb weight of a 1966 Volkswagen beetle, Curb Weight: 1672 Lbs. Get with the times!

mr byte, what are your specs for regulation and voltage tolerance. What are you using this for. Maybe we can suggest alternatives.

 

[Easy peasy]

unfortunately there would be a large voltage change from no load to full load, which may well be too great for your application... without a knowledge of power electronics it might be better to buy a supply off ebay...

 

[chuckey]

If you go for a centre tapped secondary, there is only one diode volt drop in series with the supply, but the current through the diode will be twice as much. A 1H choke is far too large, .01 H is more like it. I have one rated at 12A, it weighs about 10 kG. Likewise you caps are too small, more like 10,000 MF would be better.
Frank

 

[mr_byte31]

I just added the image in first thread for the layout of my desired circuit. I didn't mean to use the same parameters.

here is a modified version of the picture without values not to miss lead others.

I will use large capacitors , not sure about the value yet but I think values of around 20,000 uF.

I think these values will let me have smaller L

 

[dick_freebird]

If you just want the result and not the adventure, an
ATX PC power supply will get you right about where
you want to be. Maybe not as quiet as you'd like, but
I've used one such for testing my POL DC-DC chips
and the noise on the 5V output was unremarkable.

 

[Audioguru]

I think a microwave oven transformer is too cheap. Have you heard the fan speed suddenly increase a lot when the oven times out? Because its voltage suddenly rises a lot because a microwave oven does not need good voltage regulation and if the cheap transformer gets hot then who cares? It is turned on only for a short time and the more heat the better.

 

[Easy peasy]

How are you going to get 5V, 50A out of a uWave transfomer? you need a suitably made transformer to do this, even then if you have 5V AC rms out of your transformer this is <4.5V DC average under load, and 7.1V DC no load.

A Tx with 9V ac out (rms) with 9% leakage, two scr's to create the +ve and two diodes to create the negative, and say 22,000uF of output C (or larger), then control the SCR's to get your 5VDC out, this is one of the easiest ways, the 9% leakage gives you nearly sinusoid currents at full power in the Tx and lowest I^2R losses in the Tx, also limits any fault currents in the SCR's (80A scr's) should you short the output.

 

[mr_byte31]

I used ATX power supply before. It could deliver 20A @5v but I would still need to provide 50A.
since I have questions about MOT and how to use it , I took the idea from this video : https://www.youtube.com/watch?v=d5pGN6pqkyY

 

[D.A.(Tony)Stewart]

Hi All,

I am planning to build a 5v power supply which can provide at least 50A. I found many project on internet that use MOT (Microwave oven transformer)
it can produce a lot of current. I am focusing now on the losses and low ripple DC voltage.

I am planning to use Pi-Filter to remove the AC ripple voltage (Forget about values in figure, I just add it as reference):

**broken link removed**



I will configure the secondary windings to have 6.4 v AC to overcome the losses on the rectifier

I have two points here

losses:
-------
The bridge rectifies would reduce waste 2x0.7v = 1.4 v
The losses would be 1.4 v x 50 A = 70W
the remaining voltage would be (6.4-1.4=5 v), the power here would be = 5 v x 50 A = 250 W
we get a efficiency of approx = 1- 70/250 = 72% !! I think this is low
any idea how to improve this ?

AC ripples:
-----------
is it possible to have a very low AC ripples at this high current using Pi-filter? (forget about the C and L values in figure)

Possible but not practical

As %ripple is reduced with larger C, Iavg/Ipp increases such that it is = %rippleVpp

For line frequency LC+RC filter time constant > 5x (1/f*2) for 5% ripple and 20x peak pulse current.
Thus Cap value and ESR value are expensive.

5V/10A= 0.1Ω and for 50Hz ,1/2f= 10ms
thus C=500ms/0.1Ω= 50mF with ESR of <5% of load or < 5mΩ

Thus the Cap needed RC time constant= 50mF*5mΩ= 250us which is 0.05ms/10ms = 0.5% or 5% of the pulse surge time.

This would cost ~$30 (1of) if you could find one.
https://www.digikey.com/product-search/en?v=478&mpart=BZ015A503ZLB
THese are only $3.49cdn ea. (2500pcs) not stocked and not sold for 1pc
But only 15mF and 40mΩ and has an RC = 600us. USing two Caps gives and RC value of the same 600us so the ripple would be 0.6ms/10=+/-6% roughly.

SMPS are cheaper and more efficient since it uses smaller caps and lower voltage drop switches from RdsOn.

You would be lucky to get 30% efficiency with this high peak current and low ripple, duty cycle.

It would be far better to use a PC PSU.