Power Electronics Power Supply

[SG-1]

I am building, hopefully, a power supply with the following primary specs. 

1. Powered from a 15 or 20 ampere receptacle.
2. 0 to 125 VDC output. 
3. 12 to 16 ampere output max.
4. 1 volt ripple @ max output.
5. Power supply is to be portable.  Meaning light weight as possible.
6. Must be able to withstand repeated short circuit current on the output. It will be used to energize control circuits for the first time and test them.  Many times short circuits are discovered due to either bad wiring or engineering, usually the wiring.

To ditch a major amount of mass I have chosen a Crydom controller to replace the traditional varaible auto-transformer.
http://www.alliedelec.com/Images/Products/Datasheets/BM/CRYDOM_CO/682-3038.PDF

I will feed the output into a rectifier then to some filtering.  This is where I need help. A simple RC filter will require approx. a 10F capacitor to control the ripple at the currents desired.  This size capacitor will probably need a "soft start" relay so not to trip the supply circuit breaker, 15 or 20 ampere.  I see regulators on-line, but the current & voltage rating are a fraction of my needs.  I have not investigated using a pie filter ahead of a RC filter. 

Any suggestions would be greatly appreciated.

I am ignoring the input & output protection for now, except using a 10 dollar fuse is out.  That is why the controller is over sized to get more I2t. 


 

[SG-1]

One more overlooked detail is that the input voltage is 120VAC.

[Zero999]

What about power factor correction? A large power supply such as this will need power factor correction.

That relay is not a voltage converter. It appears to be a phase controller and would probably not like a non-linear load such as a transformer with a rectifier and huge capacitors on the output.

120V rectified will give 168V so the capacitor will need to be rated to 200V. A 200V 10F capacitor will need to store 0.5*10*200² = 200kJ. To put this into perspective a 10,000µF 50V capacitor stores 12.5J and has a diameter of 25mm and is 50mm high, your capacitor would be 16000 times that volume.

You need a switched mode power supply and to know what you're doing to do this properly/safely.

[Time]

This is kind of a very brute force approach to this don't you think?  kW class power supplies are almost always switching.  I am interested to see this evolve for you as 1.5 kW is not typically portable and light weight.

[Zero999]

It depends on what you mean by portable? I'm thinking it could fit into something the size of a large briefcase, say 500x300x150mm.

Still you need a lot of experience with power electronics design to do this, its not something for a beginner.

[DrGeoff]

I am building, hopefully, a power supply with the following primary specs. 

1. Powered from a 15 or 20 ampere receptacle.
2. 0 to 125 VDC output. 
3. 12 to 16 ampere output max.
4. 1 volt ripple @ max output.
5. Power supply is to be portable.  Meaning light weight as possible.
6. Must be able to withstand repeated short circuit current on the output. It will be used to energize control circuits for the first time and test them.  Many times short circuits are discovered due to either bad wiring or engineering, usually the wiring.

I don't see any line or load regulation specs here.

[SG-1]

What about power factor correction? A large power supply such as this will need power factor correction.  

That relay is not a voltage converter. It appears to be a phase controller and would probably not like a non-linear load such as a transformer with a rectifier and huge capacitors on the output.

120V rectified will give 168V so the capacitor will need to be rated to 200V. A 200V 10F capacitor will need to store 0.5*10*200² = 200kJ. To put this into perspective a 10,000µF 50V capacitor stores 12.5J and has a diameter of 25mm and is 50mm high, your capacitor would be 16000 times that volume.

You need a switched mode power supply and to know what you're doing to do this properly/safely.

I will add PF correction to the list of issues.

I have 3 450V 980MFD caps so I have an idea of the space & weight.  

I will let you know how the controller handles diodes & caps on it's output.  I received a non-answer from Crydom on that one.

I have never seen a switch mode with a variable voltage output.  Do you have a link ?

I fully realize designing or building a switch mode PS is out of my league.  

This is kind of a very brute force approach to this don't you think?  kW class power supplies are almost always switching.  I am interested to see this evolve for you as 1.5 kW is not typically portable and light weight.

The design we use day in & day out are on wheels but you need a crane to lift them.  The 3-Phase variable auto-transformers alone probably weigh 100 to 150 lbs.  They feed a set of 1 to 1 transformers wired in open delta, they feed an ancient  3-Phase bridge, no smoothing caps on the output, just a voltrap.  The PS can produce a voltage greater than 300 VDC, but the meter pegs out at 300. It is truely brute force 20th century technology.

The continuous output current can be scaled back & just allow for a 6-10 second surge.

[SG-1]

I am building, hopefully, a power supply with the following primary specs. 

1. Powered from a 15 or 20 ampere receptacle.
2. 0 to 125 VDC output. 
3. 12 to 16 ampere output max.
4. 1 volt ripple @ max output.
5. Power supply is to be portable.  Meaning light weight as possible.
6. Must be able to withstand repeated short circuit current on the output. It will be used to energize control circuits for the first time and test them.  Many times short circuits are discovered due to either bad wiring or engineering, usually the wiring.

I don't see any line or load regulation specs here.

The non-portable PS have no line or load regulation, as described above.  A 125 VDC control circuit is designed to operate between 100 & 140 volts for closing a Medium Voltage circuit breaker,  70 VDC open circuit voltage to trip the breaker ( minimum trip voltage ). The output can sag & does sag when 2 breakers are charging their closing springs. 

In actual operation, when the gear is commissioned the voltage is supplied by a large bank of batteries, with lots & lots of cold cranking amps.  The batteries are maintained in float mode which is about 138 volts.

[SG-1]

Here are some pictures of the antique diode bridge that got this whole mess started.   
One picture shows a diode that has ejected it's anode. Why are there resistors in the assembly ?  They measure 220K ohms.   When I get time I will put some pictures of the voltrap up. 

http://s1204.photobucket.com/albums/bb408/Finduilas1/

Thanks, for all the advise so far. 

[Zero999]

I will add PF correction to the list of issues.

I have 3 450V 980MFD caps so I have an idea of the space & weight. 

Please use the modify button rather than creating loads of posts.

A 450V 980µF capacitor stores 100J, so that's 300J in total, the 10F 200V capacitor you require will store 16000 and will be 53 times the size of all your capacitors put together.

[SG-1]

That is one honking capacitor, that's not going to happen.  A picture of the voltrap is now up at the same link. 
http://s1204.photobucket.com/albums/bb408/Finduilas1/

[Simon]

if you up your frequency using switch mode technology you will greatly reduce the value of the capacitors and the size of the transformer. Also to stop the startup tripping the current limit you have a few options, have current regulation so that it will hold a max current of 12-146 amps rather than just cutting out (you could implement both methods, current hold and cutout). You could also put the sensing circuitry for the current regulation after the smoothing capacitors, if you do this your feedback line for any regulator must come from the output so that the regulator takes into account the drop on the sensing resitor.

[alm]

I agree that these requirements definitely scream for a SMPS, but designing one will be non-trivial. Not something you can just copy from a schematic. One example of an SMPS with adjustable output voltage/current is the Agilent 603xA (3MB PDF service manual), but they're designed for much lower power (1kW or so).

[Simon]

I believe that SMPS can be parralled ? might simplify matters even ther it sounds harder

[scrat]

alm preceded me, I was just writing to ask about your budget, since a power supply like you describe is usually quite expensive, but could cost you even more (time and risk) to build one yourself.
Another thing to take into account is precision and dynamics (= use for which they are intended). The 1V ripple you mention makes me think it is not for a laboratory use.

Agilent features also some power supplies actually built by TDK lambda. We have two of their (TDK) laboratory regulated power supplies 300V, 8A for rack mount, each occupies some rack modules (about 10cm high). I don't remember the price, something about 3000 € each, I suppose. Ah, and they have isolated output.
Just take a look, before starting a too large project.

I believe that SMPS can be parralled ? might simplify matters even ther it sounds harder

Yes, but you must tell the control to act like this, since they cannot regulate both output voltage, if paralleled, should be master and slave.

[SG-1]

Thanks for a ton of information.  I was thinking earlier that this would be sooo much easier if I could get up to 6 pulse rectification.  That is what the "non-portable" units use by the way of a 3-Phase input & no output filters. You see these are not lab units.  The ripple is about 6% (measured value ). That is the trouble with modern power supplies when you do not need the precision,$$$, built into them.  Maybe we do and just do not know it yet ?   

Switch mode technology is the way to get the frequency up.   Budget ?  As cheep as possible...  Right now I am exploring the cost. It may be that a fixed output supplys is the final solution.   The more complex it gets the less us old timers are going to trust it.  Right now it will take some convincing to allow more electronics than just diodes into the supply.     Just because that is what we have been doing for the last 30+ years.

Thanks to everyone for the advise.

[Zero999]

At that power rating, using a quasi resonant topology looks like a good idea.