Variable Voltage Regulator and Pass Transistor

[fatman57]

I have recently purchased a device that requires 1000w of power at 48 Volts. My next problem is how to power it.

I have found that manufactured power supplies are expensive, so I am wondering what the possibility is of building my own. Of course this device is going to consume quite high average, so could be quite dangerous. Ultimately I would like to build a variable voltage power supply so it can be used in multiple projects, at the expense of maybe taking on too much.

I have found a great power supply which could deliver 100 amps @ 12v - dangerous! It does however provide an isolated output from mains and is manufactured from a good source so I trust its quality. Next of course is how I would boost the voltage and supply amperage required - I am thinking of using either a manufactured step up buck converter or linear regulator (LM317AHV seems an interesting one: http://www.farnell.com/datasheets/1949043.pdf) and then a pass transistor to deliver the current.

Is this all possible or am I taking on too much? This will be my first attempt at building a power supply and with my luck has to be a bigger project than I would like (such high amperage and variable voltage functionality), so I am worried as to how to approach this.

Can I use a manufactured step up buck converter with pass transistor (haven't seen much info on this)? Is it difficult to implement a variable voltage linear regulator with pass transistor? Does the variable voltage requirement make this all too complex?

Much obliged for any help.

 

[AnalogKid]

Without being snotty, your preliminary approach indicates that you're probably already out of your depth. A few things...

Double-converting power never is efficient, and always is expensive. At 80% efficiency, your boost converter would need to handle 100 A input current to deliver 20 A of output current at 48 V. And your converter never will hit 80%.

The LM317 is a linear regulator, which is a form of a down or buck (in the switching world) converter circuit. It's output voltage always is less than its input. There are weird switching circuits that use the part, but never at these power levels.

You absolutely need a fully isolated power system.

For a single piece, nothing you build will cost less than what you can buy.

Nothing you build will have world-wide safety certifications based on actual lab testing.

Does size matter? You can build a relatively simple giant linear power supply with a 10 or 20 pound transformer, big diode bridge, bigger filter capacitors, and a regulator with heatsinks and fans. Other than scale it won't be very different in concept from little $10 kits. But time is money, weight is money, and size is money. What is your time worth? Also, what is your target cost for the power system?

ak

 

[fatman57]

You are dead right with me being out of my depth. During this journey of investigation I have learned a lot, but I think this might be too much to take on for my current skill level.

I did read that you need a higher input than output but wan't sure. From what I can tell a step up converter requires an inductor to work but I suspected with the variable voltage requirement this might mean I require some kind of variable inductance (something else I am unsure of).

I might bit the biscuit and get a power supply, I can always sell it afterwards and get some money back if required. I have found this item, do you think it will do the job (will not hold your opinion accountable don't worry!):
http://www.aliexpress.com/item/s-10...2034985262.html?spm=2114.40010508.4.18.H2KGYf

I think the price is ok from what I have seen so far, can I expect to get one for any less?

As a side note: maybe I should just string together 4 x 12v car batteries and get a cheap charger....

 

[davenn]

I might bit the biscuit and get a power supply, I can always sell it afterwards and get some money back if required. I have found this item, do you think it will do the job (will not hold your opinion accountable don't worry!):
http://www.aliexpress.com/item/s-10...2034985262.html?spm=2114.40010508.4.18.H2KGYf

looks pretty respectable

I think the price is ok from what I have seen so far, can I expect to get one for any less?

units priced any cheaper and it's likely to be junk

the original linear supply you were originally contemplating would have cost upwards of 10 x that price

As a side note: maybe I should just string together 4 x 12v car batteries and get a cheap charger....

that's still going to be more expensive than this supply, assuming you use reasonable car batteries that could handle the load


Dave

 

[dorke]

I agree with the above writers on the issue of buying a power supply.

There are other issues you need to think about.

1st,
what is the equipment you will power with the PS.
Is it an emergency device, a medical device, what are the consequences of it's failure ? etc.
This will lead you to the required regulations that the PS should adhere to.

2nd,
How expansive is the device to you?
This will lead you to the protection mechanisms required from the PS.
e.g. over-voltage protection,input surge protection etc.


P.S
Unless you need a LAB Power Supply (which you probably don't),you should go for a fixed Voltage one for high currents-they are a lot cheaper and with far less damage potential (accidentally cranking up the the voltage knob etc.)
There are the normal steps of operating voltages 12V,24V,48V
get what you need,not a variable one.

 

[rickselectricalprojects]

If you were going to use an lm317 and a heap of pass transistors you would need a very, very, very big heat sink...
but seriously the cheapest way to do it is get a bunch of car batteries and put them in series.
Ebay might have something as well.

 

[AnalogKid]

The supply linked to in post #3 is less than 7 cents per watt for a single piece. That is stupidly cheap, and I do mean stupid. I wouldn't touch it.

ak

 

[fatman57]

Thanks for the advice! Some really good tips there.

Building my own is out the window - I might do this some day but for much lower voltages/amperage and using manufacturer specified circuits as found in component datasheets.

I will look into this further and either get a dedicated supply or use batteries. Its mostly for experimental purposes but could merit a dedicated power supply.

Once again thanks!