My Adventure: To design and build a linear 13.8Vdc / 80Amps (seeking advice)

[sarepairman2]

i have used those before. a russian trick is to solder it upsidedown on support wires so if it overheats it will fall and break the circuit.

[Circlotron]

The capacitor input filter will present a power factor of about 0.6 to your 1700VA transformer so that means you will only be able to draw about 1020 watts continuous from the transformer. Have you considered using a choke input filter? At these power levels it makes a lot of sense. Spreads the conduction of the bridge rectifier diodes over a much greater portion of the AC waveform, as well as making ripple filtering easier. Near sinewave ripple so what is left is also less audible.

Also, there are a lot of ultracapacitor banks you can get nowadays that are meant to go across a car battery, or in some cases even replace the car battery. I have on on my car in fact. 60 odd Farads across the output would do wonders for transient current capacity. http://www.ebay.com.au/itm/Ultracapacitor-Module-Battery-Eliminator-Car-Audio-Starting-Remote-Power-Solar-/111699706891?hash=item1a01d2440b&pt=LH_DefaultDomain_0&nma=true&si=4Q7ZGG%252BUylk78QbwZxamhv5NHik%253D&orig_cvip=true&rt=nc&_trksid=p2047675.l2557

[AF6LJ]

Found a solution for the inrush resistor sitting in a parts bin here at the shack...

Here are the ones I used...

I didn't design the board W7RY did. His circuit is self powered and works well. He supplies the board and the BOM you have some latitude in the parts you use. I didn't want to use the sandbar resistors in the BOM so I picked up the Ohmite 200 series resistors.

[FJHookah]

Updated Design with new Schematics....

Hi gang, Op here.  So after receiving a load of excellent recomendations, I've updated the design and schematics with the following:

• Decoupled the Pass Transistors
• Reworked the schematic drawing in a different program (easier to use - DigiKey's Schemit)
• Added extensive input common-mode filtering prior to the power transformer
• Added MOVs for input line protection from surges
• Added a 100 Amp output Fuse for peace of mind and additional safety
• Added a 100 Amp current shunt and a nice LCD backlit Meter (will need to add beads for RFI protection to the new meter)
• And finally, added a larger current in-rush resistor and a 5-pole Relay (the Relay performs double duty; 1) RL1b shorts out the inrush resistor shortly after the Momentary 'On/Off' button is turned On. 2) It opens the circuit in theory when RL1b is activated when SCR1 fires due to over-voltage issue, rendering the PS offline and bleeds the huge filter caps.

I would like your thoughts on the design at this point; is the math good enough to proceed to building prototype PCB and supply in a wooden chassis?

My biggest design challenge, one that's kicking my hiney at the moment, is whether the RL1a/b design will function as designed by providing in-rush shorting of R3 before R3 melts, as well as shutting off the power supply at the relay when SCR1 fires. I'm lost on how to design this part. Thanks!

And finally, if I were to hire a professional to perform a deep-dive on this design to ensure it's accurate and safe prior to building a prototype PCB and building the supply for initial theory validation and safety checks, how much money would you estimate that service could be?

[Circlotron]

The bases of all the series pass transistors are driven through capacitors.
Not good.

[AF6LJ]

Just a couple of notes.
1. Your inrush limiter is much easier to impliment on the primary. You won't need a 100Amp relay.
Something nice and pedestrian like a relay with 25A contacts (make and break current rating) and a coil that is say 12 or so volts you can use with a dropping resistor just after the bridge rectifier.

2. When I suggested the 100A fuse I meant it to go after the filter cap and before the pass transistors...
That is a much better place.

3. sense your supply is linear all that input filtering is way overkill
You don't have to worry about RF coming down the AC line and getting past your power transformer and causing problems. There is so much winding to core capacitance in a 60HZ power transformer any RF will get sucked to the core and to ground
You could leave C1 and the MOVs that should be well enough filtering.....


 

[FJHookah]

Thanks for the helpful comments. This project is a bit over my head, but that's how we challenge ourselves. The AC Entry receptacle that I purchased from Mouser includes the common-mode filtering you see in the schematic, so I can't remove that unless I decide to use a non-filtered receptacle. I've made the edits to the series pass transistor bases, and moved the 100A fuse. I'll do some more research on how to incorporate the inrush relay. I like the idea of dealing that with in the AC entry stage; many viable relay options for that solution. I still haven't been able to wrap my head around the work-flow electrically for that piece of this puzzle.

It's times like this that I really wish there were a few mentors locally, that I could convince hanging out in my lab for a couple of days to check my math would be a rewarding and fun experience, haha.

Thanks again... I appreciate all the feedback I can get.

Since the edits I've just made are minor, I won't republish a schematic revision just yet.

[AF6LJ]

Thanks for the helpful comments. This project is a bit over my head, but that's how we challenge ourselves. The AC Entry receptacle that I purchased from Mouser includes the common-mode filtering you see in the schematic, so I can't remove that unless I decide to use a non-filtered receptacle. I've made the edits to the series pass transistor bases, and moved the 100A fuse.
 I'll do some more research on how to incorporate the inrush relay. I like the idea of dealing that with in the AC entry stage; many viable relay options for that solution. I still haven't been able to wrap my head around the work-flow electrically for that piece of this puzzle.

I didn't know about the mains entry connector, that's cool.
Moving the fuse means that not only is the regulator and the load protected from overload but the small voltage drop the fuse will introduce into the circuit is ahead of the regulator and won't t regulation.

There are a couple of ways to operate the inrush current limiting.
Having the relay contacts on the primary side with your limiting resistor is the preferred way.
How you energize the relay coil can be done with an RC network, and rectified mains to provide voltage like the one I used in my amplifier, or you can drive the relay coil off the unregulated DC, using the inrush resistor on the primary side and the slowly building DC voltage across your caps to pull in the relay contacts after a second or two.

It's times like this that I really wish there were a few mentors locally, that I could convince hanging out in my lab for a couple of days to check my math would be a rewarding and fun experience, haha.

Thanks again... I appreciate all the feedback I can get.

Since the edits I've just made are minor, I won't republish a schematic revision just yet.

Take it a step at a time.
Build up the regulator, incorporate one transistor, use another DC supply to feed it, and put a small load on it. Then add the rest of the pass transistors and put it under the same small load. (a half dozen amps or so) and make sure they are sharing the load.

When you get your big transformer, caps, and diodes you will be ready to go knowing your regulator is functioning properly.

[oldway]

For transformer and rectification, I should use a triac preregulator to reduce power dissipation in series power transistors.

[mariush]

Another thought...

LM1084  (or LT1084) ... linear regulators capable of 5A max with 1.3v max drop ... there was also LT1083 with 7.5A max but it's discontinued nowadays

Take 5 of these, slap them on a nice heatsink. use a 0.01 ohm resistor between the outputs of each one and the final output to share the output between all regulators  and connect the feedback/adjust pins of all to a couple of resistors to configure the output to 13.8v - if the outputs are on a large copper/some other metal rail .. you can take the feedback directly from the output jack.

You now have a "module" that's capable out outputting a maximum of 25A, as long as the input voltage is minimum 13.8v + 1.3v let's say 15.5v minimum ... 
You can now make 4 identical "modules" and you can drive each module at 20A, and still have a reserve to go up to 100A .. each ldo will see maximum 4A on it and minimum about 8w dissipated on each to-220 device.  The 0.01 resistor of each LDO will dissipate a watt or so.
You can add fuses to disconnect a module if some ldo blows up / shorts and you won't lose the whole power supply, you'll only lose 20-25 amps.

I'm not sure how well this kind of design will scale up to so many ldo regulators, not sure about the balancing, how well 20 of them will balance the current between them.  There's going to be some imbalance as they heat up, the voltage out of them will slightly vary but that's taken care by the resistors on each regulator output ...

Anyway, it's about $2.5 per LDO so about a $50 investment for all and maybe a couple of dollars for the small resistors... and you can start with one module and test the 20-25A output with some 12v car lightbulbs / halogen lights.

[FJHookah]

great ideas; keep'em coming. Also, if there's a math guru out there that has the time, take a look at my schematics to see if there are any obvious gotcha's. I plan to validate the regulation circuit tomorrow as time permits.

Oh, thought that I might have hit a system design roadblock this morning. I ordered the pass transistors (2N5686G) and associated sockets last week. They arrived yesterday so last night I did a quick test-fit; NO-GO. It turns out that the 2N5686G, due to its higher current capability than the rest of the 2N line, has larger diameter emitter and base Pins. Unfortunately the data sheets do not always differentiate between the two Pin sizes. The majority of sockets available for the TO-3 package are made for the smaller Pin sized transistors (2N5302G for example).  I called Keystone and TE Connectivity (the two largest makers of the TO-3 sockets), TE makes them, whoohoo! They are 10-times more expensive than Keystones, but hey I'm not complaining.   

[AF6LJ]

I am not big on sockets, over time you can have high resistance connections and that can cause problems down the road. you might look into using Molex pins if you are sure you want to have solderless connections.

[donmr]

It looks like you have shorted out the common mode chokes on the input.  They won't do much that way.

[FJHookah]

It looks like you have shorted out the common mode chokes on the input.  They won't do much that way.

Are you referring to the ground icon just below the Mains entry AC Signal icon perhaps?  The wiring of the choke is all inside the power entry module, so I did my best to draw it schematically. I can see how the ground icon at the entry would make it look confusing. If you're referring to something else, please explain, thanks.

[PA0PBZ]

If you're referring to something else, please explain, thanks.

[poorchava]

Have I missed the post where somebody points out that linear PSU of that power is just a stupid idea unless you're looking to build a heater?

[AF6LJ]

Have I missed the post where somebody points out that linear PSU of that power is just a stupid idea unless you're looking to build a heater?

They are more forgiving and in the long run, more reliable.
Besides here in the US we have electricity to burn.

[AF6LJ]

Oh and I almost forgot.
You don't have to spend half your life trying to keep the damned thing from generating broadband noise, we amateur radio operators have enough of that to deal with.

[Kleinstein]

The low power factor will not allow so much power at the DC side. 80 at 13.5 V are nearly 1100 W. With some loss for the diodes you are at 1200 W minimum - no loss for the regulation included.  With just a rectifier and filter cap this would be something like 1800-2000 VA for the transformer and already to much for a 13 A mains fuse.  So no way around some power factor correction.

In a linear supply this would be a big choke. At this high power it may not be that bad, as chokes like transformers get better when getting larger - but expect the choke to be nearly as large and heavy as the transformer. At least it could also save on the filter caps.

The regulator should get it's own supply to minimize the dropout, so less power is wasted and less heat generated.

[FJHookah]

Got it, thanks

[FJHookah]

Have I missed the post where somebody points out that linear PSU of that power is just a stupid idea unless you're looking to build a heater?

Although sarcasm is seen as good humor, given in the appropriate context, did you have anything useful to add to the discussion? If you're bored and would like to make me smarter; would you be so kind to provide an example basic schematic of how I could implement a timing circuit with the associated components required to manage the horde of inrush current I'm assuming the combination of the 1500VA transformer and the 164kµF filter caps will draw at Power-On? That would be most appreciated and useful.  Based on some great feedback I've received thus far, I'd also like to add a separate transformer supply for the regulator. If you could draw that out in a simplistic fashion, I would be most appreciative. Perhaps you have a few favorite links on the web that you might offer, where I could learn how to do this myself. Otherwise, we'll simply clap, smile, and put quarters in your hat each time you grace the thread with your comedic outbursts.   

[FJHookah]

The low power factor will not allow so much power at the DC side. 80 at 13.5 V are nearly 1100 W. With some loss for the diodes you are at 1200 W minimum - no loss for the regulation included.  With just a rectifier and filter cap this would be something like 1800-2000 VA for the transformer and already to much for a 13 A mains fuse.  So no way around some power factor correction.

In a linear supply this would be a big choke. At this high power it may not be that bad, as chokes like transformers get better when getting larger - but expect the choke to be nearly as large and heavy as the transformer. At least it could also save on the filter caps.

The regulator should get it's own supply to minimize the dropout, so less power is wasted and less heat generated.

Hmmm, I did not account for that much loss. I'll start reading up on that straight away, thank you. Although I must say, if the only corrective action is to add a 32lb choke, this design will quickly find itself in trash can.

[AF6LJ]

...........

The regulator should get it's own supply to minimize the dropout, so less power is wasted and less heat generated.

This is a good point and companies like Astron who have been making power supplies for two way, and amateur radio for decades do this on their higher power supplies.

[AF6LJ]

The low power factor will not allow so much power at the DC side. 80 at 13.5 V are nearly 1100 W. With some loss for the diodes you are at 1200 W minimum - no loss for the regulation included.  With just a rectifier and filter cap this would be something like 1800-2000 VA for the transformer and already to much for a 13 A mains fuse.  So no way around some power factor correction.

In a linear supply this would be a big choke. At this high power it may not be that bad, as chokes like transformers get better when getting larger - but expect the choke to be nearly as large and heavy as the transformer. At least it could also save on the filter caps.

The regulator should get it's own supply to minimize the dropout, so less power is wasted and less heat generated.

Hmmm, I did not account for that much loss. I'll start reading up on that straight away, thank you. Although I must say, if the only corrective action is to add a 32lb choke, this design will quickly find itself in trash can.

Your design will work fine, it isn't that different from an Astron RS-75.