Power supply

[abdullahseba]

I am designing a 50V 20A power supply. I got as far as the rectifier and smoothing capacitor but I could not find a zenr diode that
has the right specs. is there any way I could make the output liner, smooth as much as possible?

[Dave]

You are designing a 1kW power supply. Are you sure you want to stabilize your output voltage with a zener diode? Do you want to have a space heater in your lab?
If you really want to make a linear power supply, I suggest you look up voltage regulation techniques and then build a suitable regulation circuit (you might also want to add current limiting, OVP, OCP, thermal protection, etc).

I may be wrong now, but circuits that require several 100 watts usually aren't that picky about the noise of their input voltage. A switch mode power supply would be a much better solution in this case.

[Pillager]

(you might also want to add current limiting, OVP, OCP, thermal protection, etc.)

Not to mention at least eye protection for the testing...

[homebrew]

Give us some specs, like:

What are the transformer ratings ?
What rectifier are you using ?
What capacitors have you chosen ?
What are the requirements on stability for your load?
What are the constraints on the environment(like max operating temperature etc etc)?

[abdullahseba]

the transformer is 240v in 50v 20A out
Rectfier: GBPC2501W
Capacitor: 100000uf
Load: underground train

[Andy Watson]

Load: underground train

!
A whole train, or just part of a train?

[Jeroen3]

Why would you want to design a 1 kW power supply yourself. Save the effort (and weeks of designing) and buy one, you should have one for $1k.
If you compare the off-the-shelf price with the estimated hours of effort you'd need to put in, I'm sure it's cheaper. And you'll have someone to blame if the device explodes.

[rob77]

ROFLMAO 
this thread really made my day

[SatoruSatake]

Could you provid some more specific parameters of your filters,capacitors and maybe other electronic components?

[Richard Crowley]

Is this a small scale-model of an "underground train"?  Or is this the real thing?
1KW seems like too much for a scale model, and not nearly enough for a real train.

Why does a train need regulation?  Or even filtering?
Your questions just raise more questions.

[Richard Head]

Look for a telecoms rectifier. They output 53V @ 30-60A generally, with active PFC. A bit better than a resistor and a Zener diode!

[hamdi.tn]

1kw linear power supply  switch mode power supply is the right solution

[rolycat]

Is this a small scale-model of an "underground train"?  Or is this the real thing?
1KW seems like too much for a scale model, and not nearly enough for a real train.

Why does a train need regulation?  Or even filtering?
Your questions just raise more questions.

I suspect the OP is referring to this London Underground carriage previously discussed on the forum here.

It's not going anywhere, so the power will be for lighting etcetera rather than traction.

He is 15 years old or thereabouts, so designing a 1KW+ switch mode power supply might be a bit ambitious.

[abdullahseba]

i just want to know if there is a smoothing circuit that keeps the output at 50v 20A 

[wraper]

There is but if you want to do it as a linear PSU it will dissipate a freaking lot of heat. And certainly you cannot do it with a zener.

[abdullahseba]

There is but if you want to do it as a linear PSU it will dissipate a freaking lot of heat. And certainly you cannot do it with a zener.

the heat is no problem there is lots of space for fan cooling on the train.
so where is this plan? do you have a link somewhere?

[mariush]

It's not cost efficient to make your own power supply to output 50v at 20A.  The power transformer would be heavy and large and you'd also need additional components to stabilize the output voltage.

You can buy a 48v power supply from stores and most of them allow some small control over the output voltage, which would allow you to tweak the power supply to output 50v.  See for example Meanwell power supplies on eBay, for example here's a 600w unit... two of these will be capable of providing that 1000 watts of power : http://www.ebay.com/itm/Meanwell-SE-600-48-AC-DC-Switching-Power-Supply-48V-12-5A-600W-Single-Output-/161350891549?pt=LH_DefaultDomain_0&hash=item259143381d

If you really want to make a linear power supply, the best approach would be to use a custom transformer, let's say something that outputs about 42v AC and maybe a couple of bridge rectifiers in parallel to reduce the heat dissipation...
After rectification, you'll have 42v x 1.414 = 60v , minus about 2-3v dropped on the rectifiers, so about 58v peak DC voltage.  At 20A, that 2-3v drop on rectifier means lots of heat, that's why it would be wiser to use 2 bridge rectifiers, each doing only 10A of current.

To get at least 50v minimum DC voltage, you can use capacitors to smooth out the output and you can use this formula to approximate capacitance  :  C  =  Current / (2 x AC Frequency x V ripple)  where Vripple is how much you're willing to let the voltage go down.
For peaks of 20A and assuming you want a minimum dc voltage of 52v then you have  C  =  20A / [ 2 x 50 Hz x  (58v peak dc  - 52v minimum dc) ]  = 20 / 600 = 0.0333 Farads  or 33.333 uF

So you'd need 33.333 uF to keep the minimum voltage to about 52v when your train uses 20A, but when the power consumption is lower (fewer amps) the capacitors will charge more often and the voltage will go closer to that peak DC voltage of 58v.

 So you still need some regulation circuit to keep that output voltage to 50v, and for that you can use several power transistors in parallel and a feedback circuit (operation amplifier to monitor the output voltage and restrict the current flow through the transistors so that you always get only 50v)

As an example, you can see attached a power supply capable of 5A and as you can see they use 4 C2750  (2SC2750)  power transistors just to make sure the power supply is capable of 0-30v at 5A.
They have to use a lot of power transistors even for only 5A because you see, it's not only the current it's also the voltage difference that matters... with that adjustable power supply if user wants 1v out at 5A, the transistors receive about 12v from the transformer so the power transistors need to dissipate about  (12v - 1v) x 5A  =  55w as heat.... it's easier for each of those 4 transistors to dissipate only 55/4 = ~ 14 watts each. 

In your case, you'll have a minimum voltage of about 52v and a maximum of about 58v, so the voltage difference will be only about 2-6v. Still, at 20A you'll have about 160 watts of power dissipated in the transistors. Assuming a maximum of about 20w per power transistor you'd want at least 8-10 such power transistors in parallel, installed on a big heatsink and also a fan cooling that heatsink.

When you draw the line, you can see the power transformer, the large heatsink, the fans to keep the heatsink cool, the components needed, they all add up to the price and it's more cost effective to buy a ready made power supply.

[abdullahseba]

It's not cost efficient to make your own power supply to output 50v at 20A.  The power transformer would be heavy and large and you'd also need additional components to stabilize the output voltage.

You can buy a 48v power supply from stores and most of them allow some small control over the output voltage, which would allow you to tweak the power supply to output 50v.  See for example Meanwell power supplies on eBay, for example here's a 600w unit... two of these will be capable of providing that 1000 watts of power : http://www.ebay.com/itm/Meanwell-SE-600-48-AC-DC-Switching-Power-Supply-48V-12-5A-600W-Single-Output-/161350891549?pt=LH_DefaultDomain_0&hash=item259143381d

If you really want to make a linear power supply, the best approach would be to use a custom transformer, let's say something that outputs about 42v AC and maybe a couple of bridge rectifiers in parallel to reduce the heat dissipation...
After rectification, you'll have 42v x 1.414 = 60v , minus about 2-3v dropped on the rectifiers, so about 58v peak DC voltage.  At 20A, that 2-3v drop on rectifier means lots of heat, that's why it would be wiser to use 2 bridge rectifiers, each doing only 10A of current.

To get at least 50v minimum DC voltage, you can use capacitors to smooth out the output and you can use this formula to approximate capacitance  :  C  =  Current / (2 x AC Frequency x V ripple)  where Vripple is how much you're willing to let the voltage go down.
For peaks of 20A and assuming you want a minimum dc voltage of 52v then you have  C  =  20A / [ 2 x 50 Hz x  (58v peak dc  - 52v minimum dc) ]  = 20 / 600 = 0.0333 Farads  or 33.333 uF

So you'd need 33.333 uF to keep the minimum voltage to about 52v when your train uses 20A, but when the power consumption is lower (fewer amps) the capacitors will charge more often and the voltage will go closer to that peak DC voltage of 58v.

 So you still need some regulation circuit to keep that output voltage to 50v, and for that you can use several power transistors in parallel and a feedback circuit (operation amplifier to monitor the output voltage and restrict the current flow through the transistors so that you always get only 50v)

As an example, you can see attached a power supply capable of 5A and as you can see they use 4 C2750  (2SC2750)  power transistors just to make sure the power supply is capable of 0-30v at 5A.
They have to use a lot of power transistors even for only 5A because you see, it's not only the current it's also the voltage difference that matters... with that adjustable power supply if user wants 1v out at 5A, the transistors receive about 12v from the transformer so the power transistors need to dissipate about  (12v - 1v) x 5A  =  55w as heat.... it's easier for each of those 4 transistors to dissipate only 55/4 = ~ 14 watts each. 
https://www.eevblog.com/forum/Smileys/default/icon_smile_thumbsup.gif
In your case, you'll have a minimum voltage of about 52v and a maximum of about 58v, so the voltage difference will be only about 2-6v. Still, at 20A you'll have about 160 watts of power dissipated in the transistors. Assuming a maximum of about 20w per power transistor you'd want at least 8-10 such power transistors in parallel, installed on a big heatsink and also a fan cooling that heatsink.

When you draw the line, you can see the power transformer, the large heatsink, the fans to keep the heatsink cool, the components needed, they all add up to the price and it's more cost effective to buy a ready made power supply.

thank you very much
this has helped me a lot   

[Marco]

You can get Delta 1200W 48V power supplies pretty cheap on ebay.

[Andy Watson]

It might be useful if you could be a bit more specific about what you are trying to power. I'm struggling to think of anything on a tube train that would required smoothed DC, let alone stabilised DC.

[Jeroen3]

If indeed used to power the lights on a film set, a smooth DC voltage and current might benefit the camera's.

[Richard Crowley]

If indeed used to power the lights on a film set, a smooth DC voltage and current might benefit the camera's.

For over 100 years we have been shooting film/video under ordinary incandescent lamps operating on raw AC (not DC, not smoothed, not regulated).  The thermal inertia of the filament is sufficient to "smooth" the light output from most lamps.  But if I were refurbishing an old rail car, I would strongly consider replacing all the incandescent (and even fluorescent) lamps with LED.  They don't need filtering or regulation, either.

Deploying a home-made, high-power, mains-connected supply for use in a public place (like a museum) may even be illegal in some jurisdictions.  If the situation REALLY requires smooth DC, and especially if it requires regulated DC at such high power, I would strongly discourage the OP from attempting to DIY.  This wouldn't even be advisable for a home project by someone with limited experience.  Recommend deploying a commercial unit that has all the regulatory and safety inspections and ratings, etc.

[abdullahseba]

It might be useful if you could be a bit more specific about what you are trying to power. I'm struggling to think of anything on a tube train that would required smoothed DC, let alone stabilised DC.

everything on the train excluding motor,compressor(240V) & lights(110V).
the speakers,doors,fans,external lights & computers work on 50V
the fans work on 240v through an inverter.

[abdullahseba]

If indeed used to power the lights on a film set, a smooth DC voltage and current might benefit the camera's.

For over 100 years we have been shooting film/video under ordinary incandescent lamps operating on raw AC (not DC, not smoothed, not regulated).  The thermal inertia of the filament is sufficient to "smooth" the light output from most lamps.  But if I were refurbishing an old rail car, I would strongly consider replacing all the incandescent (and even fluorescent) lamps with LED.  They don't need filtering or regulation, either.

Deploying a home-made, high-power, mains-connected supply for use in a public place (like a museum) may even be illegal in some jurisdictions.  If the situation REALLY requires smooth DC, and especially if it requires regulated DC at such high power, I would strongly discourage the OP from attempting to DIY.  This wouldn't even be advisable for a home project by someone with limited experience.  Recommend deploying a commercial unit that has all the regulatory and safety inspections and ratings, etc.

the lights are fluorescent tubes that operate on 110v.
the transformer is to replace the battery's which are expensive to run & 78Kg each   

[Marco]

What's wrong with something like this?

http://www.ebay.co.uk/itm/Delta-DPS-1200AB-A-1200W-Power-Supply-TESTED-/200732965696?pt=US_Server_Power_Supplies&hash=item2ebc9dfb40

The copper in a 50/60 Hz transformer would probably cost you more.