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

• My Adventure: To design and build a linear 13.8Vdc / 80Amps (seeking advice)
  Posted by FJHookah on 08 May, 2016 17:33
• Hello,
  
  I'm seeking your advice and expertise for a new project that I've recently started. My objective is to design and build a very high quality, very safe, very stable 13.8v Linear DC Power Supply that can sustain up to 80 Amps at a 50% to 75% duty cycle. At this stage, I've pieced together a design based on old schematics found while cruising the web that makes use of the LM723 voltage regulator. I've upgraded various components in support of the high current requirement of this power supply.  For example, I'll be using a single 2N5686G to drive 10 output stage 2N5686G Series Pass Transistors. Each output transistor is capable of 50 Amps and a breakdown voltage of 80Vdc.  Yes, this is perhaps a bit overkill, but it also means that driving the power supply at the full spec of 80 Amps is only pushing the 2N5686 output transistors at approximately 16% of their maximum which hopefully means it will be fairly easy to keep them properly cooled without the need for 75db of Fan noise.  The output current will be load-balanced using 0.047 Ohm 12.5W resistors (chassis mounted w/heat sink casings) to ensure each transistor is equally sharing the load. My hope is that by doing it this way, it will help improve the long-term reliability of the power supply.
  
  Let me share a little about the build objectives and key power supply features:
  

  • The primary use is for my HAM Radio hobby. Granted, I do not need an 80 AMP 13.8V supply for my hobby, as I'm typically only needing up to 50 AMPS, but I wanted to challenge myself and learn at the same time. To be honest, there is a little bit of a cool-factor in building such a beast of a linear supply the right way, including applicable safety features. I currently use an Astron VS-70M that struggles to drive a VHF amplifier that I use quite often. The Astron supply is supposedly capable of 56 AMPS full duty-cycle, but trust me, it would require liquid nitrogen to keep it cool at that output level. I want to build a power supply that wouldn't even break a sweat at 60 AMPS full duty-cycle.
  • Design criteria - 13.8V, 80 AMPS at up to 75% duty-cycle, soft starting, over voltage protection, current limiting, short circuit protection, remote voltage sensing and automatic passive and active cooling fan control (with removable and cleanable dust fan filters).
  • The power supply needs to look as GREAT as it performs. I'm strongly considering using a Chassis sold by diyAUDIO that has large external fin heat sinks which both sides of the chassis completely. It is in a 4U size chassis and 400mm deep. It's a large chassis, yes, but the size is required for the Transformer, filter caps, etc. The company will also perform ALL of the CNC machining if I submit a CAD file to them. This means that it will be professionally built and drilled / CNC'd, with black anodizing making it the best looking DC supply in Texas! (smiling).
  • I'm not as concerned with the cost to build this power supply as I am building it correctly, to spec, with rock-solid protection features as noted above. For example, the chassis alone will be approximately $550, the Transformer was $225 shipped, the pass transistors were $154, the balancing resistors were $65, the filtering caps were $206, and the bridge was $68. This will end up costing approximately $1400, which is completely ok with me.

  
  Here is a listing of critical components with their specs that I've selected and sourced already:
  

  • AC power entry:  TE Connectivity filtered module (with built-in rocker switch)
  • AC power entry fuse:  13 AMP push-button breaker (metal casing)
  • Transformer:  Antek 1500VA 17V (they are building this one for me, as it is not a common size)
  • Bridge Rectifier:  IXYS 160 AMP, 1200 V.  Max surge = 2800 Amps,  Vf = 1.43, power dissipation = 300W at 25C, screw mounting
  • Filter Caps:  Vishay 0.082F / 50V (quantity 2) for a total of 0.164F filtering
  • Pass Transistors:  2N5686G (quantity 11, with one used as the Driver).  50 AMP / 80V
  • Current-balancing resistors:  Vishay 12.5W 1% - 0.047 Ohms

  
  Questions and some of my current challenges that I'm struggling with:
  

  • Aftter the bridge rectifier, there is a momentary switch that will be placed on the front panel of the chassis. The idea is that the switch will open the gates, so to speak, for the inrush of current caused by having such a large transformer and huge filtering caps. I need a way to adequately 'slow-start' this beast. What are your thoughts?
  • I've considered going with a parallel bridged rectification design instead of using just one bridge, but am I right in thinking that there would not be much benefit from going with a parallel bridge setup? Should I stick with just the one 160 AMP bridge or consider two in parallel?
  • I've ordered the Transformer with 17 VAC on the secondaries, assuming that the post-rectification process will yield me approximately 24 to 26  'no-load' raw DC volts. With what I've read and researched thus far, the heavy loaded DC volts will sag to approximately 16 to 17 due to normal transformer load sag expectations. Is my assumption accurate; what are your thoughts?

  
  I'm attaching rev1 schematics with hopes that this project may peak some interest such that you feel encouraged to offer advice, share your technical expertise, and perhaps even follow me along this exciting journey. There's no doubt I have a lot to learn, and I'm open to your harsh critique, although if you throw a little humor in once in a while that would be great too!   
  
  Thanks!
  
  Steve
  
  
  
  
  
  
• #1 Reply
  Posted by NiHaoMike on 08 May, 2016 17:47
• Why not a dual mode linear/switcher where the switching part is only active in transmit mode?
• #2 Reply
  Posted by nctnico on 08 May, 2016 18:07
• I agree. I'd get a 12V switching PSU from Ebay, turn up the output voltage a little, add some post filtering using common mode chokes and extra capacitors and be done with it.
• #3 Reply
  Posted by PA0PBZ on 08 May, 2016 18:59
• Quote from: FJHookah on 08 May, 2016 17:33

  it also means that driving the power supply at the full spec of 80 Amps is only pushing the 2N5686 output transistors at approximately 16% of their maximum which hopefully means it will be fairly easy to keep them properly cooled without the need for 75db of Fan noise.
  

  
  Just a minor point, but the specs of the transistors is not going to change what you are going to dissipate ((rectified voltage-13.8V)* Amps), so the cooling job will stay the same.
  
  I think you want S2 parallel to R1, not in series.
  
  Also, what is the SCR going to blow when it triggers except R3?
  
  Are you measuring the current for one transistor and hoping that it will be the same for the other 9?
  
• #4 Reply
  Posted by Towger on 08 May, 2016 19:06
• When money is no object. Where is the fun of hacking a old rack/blade server PSU up to 13.8 when you can build a 4U beauty which hardly makes a sound.[emoji3]
• #5 Reply
  Posted by German_EE on 08 May, 2016 19:18
• OK, some construction notes from someone who really has been there and done that, although mine is 120A and a two man lift.
  
  1) A separate transformer and rectifier driving the 723 is a good idea because it keeps it more stable and keeps load fluctuations out of the regulator circuit.
  
  2) Decouple everything, especially around the 723 and the bases of the power transistors. Ferrite beads on every base connection is a good idea as well.
  
  3) Slow start circuit, run a 50 ohm resistor in series with your mains supply and put a pair of relay contacts across it. Drive the relay coil with the voltage across your main bridge rectifier. This will give a soft start circuit that remains in circuit for a fraction of a second until the contacts close but those contacts and that resistor need to be BIG.
  
  4) If you can lift it on your own it isn't big enough.
  
  5) Dump that 10 ohm resistor after the bridge rectifier, at 80A it will (in theory) drop 800V and generate 64KW of heat
  
  Good Luck!!
• #6 Reply
  Posted by T3sl4co1l on 08 May, 2016 19:48
• If cost is no object, I suggest hiring a Consultant (nudge, nudge.. ) to fully review the design and ensure it's right the first time...
  
  Tim
• #7 Reply
  Posted by John Heath on 08 May, 2016 20:50
• I had a thought. An off the self lead acid car battery will act as both voltage regulator and filter condensers. All that is left is an off the self battery charger capable of a 40 to 50 amp charge. With the main job taken care of superficial window dressing should never be underestimated. First appearance is everything. I would suggest a flat black front face , it screams testosterone , with 200 AMPs in bold white print. Follow this with a PIC to display any irrelevant variable that comes to mind to dazzle and amaze the technical staff. Internal temp , outside temp , heat sink to ambient air difference temp , fan air differential pressure with estimated hours to air filter change. There is no such thing as too much information , more much like money is always better. I am being humorous about this but there is a grain of truth here. Sound engineering is not always enough. I little superficial razzle and dazzle can go a long way to sell a project.
• #8 Reply
  Posted by Seekonk on 08 May, 2016 21:32
• There are nice power supplies out there like Xantrex XTR7.5-80 and other series cheap on ebay, <$100 shipped.  Just place one of these in series with an unregulatd 7-9V transformer/capacitor supply.  Voltage sense on the outside of the circuit, manual shows how.
  
  I have an old 7.5V 150A linear supply I'd love to get rid of.  Looks like an easy conversion to full wave bridge the transformer, new caps and just some minor circuit changes.  Thought of doing it years ago but it is just so damn heavy.
• #9 Reply
  Posted by sarepairman2 on 08 May, 2016 22:33
• I bet you can design something really nice if you put the thought into it.
  
  Since you are going all out have you thought about water cooling? It sounds like you have a fairly serious installation and you can redirect the heat somewhere else this way. And have a more unique piece of equipment. Granted, then you need to deal with pump electronics etc, but keeping a comfortable work environment will be worth it potentially.
  
  I would add a inrush limiting circuit. For this class of supply I would do a 555 timer + relay to short out a current inrush resistor once the capacitor banks are saturated.
  
  You might also want to put in a current limiting circuit on the output that also shorts out, so the load that you connect to it will also not experience a surge. I believe capacitors are just as stressed discharging at charging. If you connect a low impedance load do it, do you want 80 amps to shoot right out of your supply ? This you could do with a "full current" interlock, i.e. monitor your meters and disengage the current limiting, you could do this in steps.
  
  You could also make a fast acting anti short circuit, which would short the supply internally if a fast current transient is detected.
  
  I would also put panel meters on the current and voltage, possibly dial indicators to keep it simple and low noise.
  I would also put in thermal protection, i.e. a thermal disconnect if the heat sink over heats.  I don't see an output fuse.
  
  you probobly also want a mains supply indicator (neon/LED).
  
  make sure your fuses are easy to access
  
  also, how are you routing the current? bus bars?
  
  also, i would add a big ass disconnect switch (read about EMO switches)
• #10 Reply
  Posted by FJHookah on 08 May, 2016 22:37
• Op here....
  
  I just now read through the responses thus far. Thank you for the thoughtful comments and suggestions. I knew posting this thread would prove valuable for the project. To the whiteboard I go, thanks guys!
  
  Steve
• #11 Reply
  Posted by sarepairman2 on 08 May, 2016 22:43
• also with this bad boy you probobly want to add a remote disconnect option.
  
  I.E. the ability to design equipment that you hook up to it to tell it to shut down with a simple fault signal. i would do an optically isolated TTL or whatever interface.
  
  It does not need to be complicated, it would just help de-enegerize things to limit the scale of electrical  fires that might come about with high power RF shit, so you don't need to design your other equipment with heavy duty relays and stuff. even if you do, you can have super visor circuits and redundancy (which you should have messing around with 80 amps!)
  
  also think about how to route ground in your rack.
  
  if you want to be fancy, you could add a fiber optic disconnect bus, so that on your other equipment faults can be indicated by simple LED's which would shine into your power supply, terminating the output.
  This is what you should probobly do if its a RF noise heavy environment, rather then TTL on the remote disconnect
  
  then other things could just have fault LED's routed to fiber optics. simple to design, reliable circuits.
• #12 Reply
  Posted by AF6LJ on 08 May, 2016 22:51
• Keep us posted on your project.
  Most everyone has already mentioned what I had in mind.
  The only other thing I would do is put a 100 amp fuse between the last filter cap and the pass transistors. If something goes wrong at these current levels it can get down right destructive. You have enough voltage overhead that you can afford to loose a couple hundred millivolts in a fuse / fuse holder. Better safe than sorry....
  
• #13 Reply
  Posted by Jebnor on 08 May, 2016 23:05
• With regards to your soft start option, take a look at Daves teardown of the   It shows one soft start option in there.
• #14 Reply
  Posted by xrunner on 08 May, 2016 23:11
• Quote from: FJHookah on 08 May, 2016 22:37

  Op here....
  
  I just now read through the responses thus far. Thank you for the thoughtful comments and suggestions. I knew posting this thread would prove valuable for the project. To the whiteboard I go, thanks guys!
  
  Steve
  

  
  See, I told you you'd get some good suggestions! 
  
  (Steve is a ham that lives in my area and frequents a local repeater - I suggested that he post his design here for the EEVBloggers to review
• #15 Reply
  Posted by FJHookah on 08 May, 2016 23:46
• Quote from: PA0PBZ on 08 May, 2016 18:59

  
  I think you want S2 parallel to R1, not in series.
  
  Also, what is the SCR going to blow when it triggers except R3?
  
  Are you measuring the current for one transistor and hoping that it will be the same for the other 9?
  

  
  R1, perhaps not working as I'm thinking it should in series, is supposed to help slow the inrush of current when S2 is actuated. Thoughts? Another gentleman responded with a suggestion to bump up that resistor to 50 Ohms, but it would need to be fairly large in order to tame the inrush apparently. I think I'll make that edit in the schematic now and see what we think.
  
  Regarding the SCR; that's one of my design roadblocks.... I'm not quite sure how I want to effectively shutdown the PS once D5 conducts and fires the thyristor SCR1, but as it stands if the load voltage hits 15 volts (creating an over voltage scenario), the SCR short circuits C1 through R3. That's all fine and dandy right, but what I have to add to the circuit is a method to prevent C1 from recharging on the next subsequent rectifier cycle, which would then effectively shut down the PS as desired.  I've seen a relay used in a schematic like this one; I'll edit the schematic to get your thoughts on its practicality and function. I'd appreciate your feedback again if you don't mind.
  
  Regarding current measurement; when the current across R21 reaches 8 amps (which is an indicator that the power supply is generating close to 80 amps total), the voltage from R11 and R12 is approximately .9 which is tapped from R11 to bias a transistor in the LM723 regulator chip, thereby making use of its current limiting function. But yes, you're absolutely right to notice that I'm only taking measure from one of the 10 pass transistors. I'm not sure how to design the circuit to take individual measurements from all 10 and run that through a comparator of sorts which might then assume the duty of feeding the LM723 .9 volts to turn on its current limiting function? I'm wide open to suggestions on improving the functionality of current limiting for this design. Thoughts?
• #16 Reply
  Posted by FJHookah on 08 May, 2016 23:54
• Quote from: sarepairman2 on 08 May, 2016 22:33

  I bet you can design something really nice if you put the thought into it.
  
  Since you are going all out have you thought about water cooling? It sounds like you have a fairly serious installation and you can redirect the heat somewhere else this way. And have a more unique piece of equipment. Granted, then you need to deal with pump electronics etc, but keeping a comfortable work environment will be worth it potentially.
  
  I would add a inrush limiting circuit. For this class of supply I would do a 555 timer + relay to short out a current inrush resistor once the capacitor banks are saturated.
  
  You might also want to put in a current limiting circuit on the output that also shorts out, so the load that you connect to it will also not experience a surge. I believe capacitors are just as stressed discharging at charging. If you connect a low impedance load do it, do you want 80 amps to shoot right out of your supply ? This you could do with a "full current" interlock, i.e. monitor your meters and disengage the current limiting, you could do this in steps.
  
  You could also make a fast acting anti short circuit, which would short the supply internally if a fast current transient is detected.
  
  I would also put panel meters on the current and voltage, possibly dial indicators to keep it simple and low noise.
  I would also put in thermal protection, i.e. a thermal disconnect if the heat sink over heats.  I don't see an output fuse.
  
  you probobly also want a mains supply indicator (neon/LED).
  
  make sure your fuses are easy to access
  
  also, how are you routing the current? bus bars?
  
  also, i would add a big ass disconnect switch (read about EMO switches)
  

  
  Outstanding suggestions; you've just given me another few days worth of work and I love it, haha.
• #17 Reply
  Posted by sarepairman2 on 09 May, 2016 00:09
• yea finish a design then give yourself some time off to do something else for a while so you can think clearly about it again, your starting a complicated project, this class of design you could easily have an employer pay for your work if they had the need.
  
  my other suggestion is to put significant thought into the wiring of the thing, i.e. figure out what kind of crimps/connectors you are going to use, figure out how to route wiring, wire gauges, etc.
  
  no one thinks of this, but the "interconnects" design is actually a skill in itself... the design only begins on the circuit board.. then you have complicated systems engineering.
  
  i suggest looking at WAGO/weildmuller for connectors.
  
  and i suggest learning about DIN RAIL mounting, and spending some time looking at how to do cables right (i.e. strain reliefs, zip ties, shrink tubing, cable covers..
  
  i suggest big ring lugs on high current carrying wires... then you can use big ass bolts as connectors.
  
  think about how to guard things against short circuits (i.e. covers, enclosures, fully shrowded connectors).
  
  you will also have to use proper torque if you do this...
  
  this project may cost you alot more then you think if you want to do the systems engineering professionally... crimp tools for instance... maybe brazing bus bars even.. torque wrench or possibly turns measurement on ring lugs.. proper grounding..
  
  be sure to properly ground your chassis..
  
  putting together what i like to call an experiment and designing a piece of lab equipment are two completely different things.
  Once you have some know how you can make your experiments nicer, but making a nice piece of servicable test equipment is an art in itself, even if you have to do zero PCB level work.
  
  
  programmers  got it easy, man.
  
  scenario: you need to do work trouble shooting it.  how do you prevent it from blowing up if you accidentally drop a screw?
  
  you might also want to include some test points in shielded locations that you can safely measure something for trouble shooting.
  
  i have seen long time engineers argue for a LONG time about things that a hobbyist would just consider non issue.
  
  
  all this may seem like over kill, but if you ever want to actually design an industrial product this is what you should do. of course, there is no harm in making a piece of shit that is convenient, I do it all the time, but i would not necessarily leave it unattended or rest easy around it.
  
  also, do not design for compactness in mind. Compact things go through many many revisions etc. if you make it compact you will just be cursing yourself later when you find something that you want to change (and you will).
• #18 Reply
  Posted by sarepairman2 on 09 May, 2016 00:24
• However, on the plus side, you are the financier for your project. This is a HUGE plus.
  
  If you are willing to spend money then it is much easier to get a solution, plus you have much less frustration to deal with. In a company your beautiful design will be gutted by bean counters and you will get frustrated  when you design to this level, then something is denied, making half of your "robust decisions" irrelevant due to the decreased reliability/performance/etc as a consequence of the "financial adjustment"
  
  and you can't get into a philosophical design argument with other engineers (like i often do)... so long you don't let the eevblog forum get into your head... its just a hobby man
  
  
  also since its an RF supply, you might just wanna put a actual POWER meter, like watt meter on the thing, so you can calculate efficientcies or some shit easily
• #19 Reply
  Posted by FJHookah on 09 May, 2016 00:34
• Quote from: sarepairman2 on 09 May, 2016 00:24

  However, on the plus side, you are the financier for your project. This is a HUGE plus.
  
  If you are willing to spend money then it is much easier to get a solution, plus you have much less frustration to deal with. In a company your beautiful design will be gutted by bean counters and you will get severely angry when you design to this level, then something is denied, making half of your "robust decisions" irrelevant due to the decreased reliability/performance/etc as a consequence of the "financial adjustment"
  
  and you can't get into a philosophical design argument with other engineers (like i often do)... so long you don't let the eevblog forum get into your head... its just a hobby man
  
  
  also since its an RF supply, you might just wanna put a actual POWER meter, like watt meter on the thing, so you can calculate efficientcies or some shit easily
  

  
  Well stated. If only I had a neighbor like yourself willing to drop by and sit at my workbench with me daily until I crack this nut 
• #20 Reply
  Posted by FJHookah on 09 May, 2016 00:36
• Quote from: Jebnor on 08 May, 2016 23:05

  With regards to your soft start option, take a look at Daves teardown of the Xantrex 300v 4A power supply.  It shows one soft start option in there.
  

  
  Thanks for posting the video... I like the extensive input filtering and will incorporate some of that straight away. Also, ah ha! They use a big resistor in parallel with a relay to control in-rush. I'm updating the schematic now.
• #21 Reply
  Posted by mariush on 09 May, 2016 01:48
• Here's a suggestion ...
  
  Have a look at LT4320 - ideal bridge controller chip: http://www.linear.com/product/LT4320
  . You pair it with 4 high quality mosfets and it should be doable to have a bridge rectifier of up to 30-40 Amps.  I'm not sure you'd find mosfets good enough to build a 80 amp rectifier with it.
  
  The benefits would be
  1. minimal heat generated ... you have maybe a couple of watts lost on each mosfet (you need 4) so let's say 10w in total  for something like 30-40 amps of current , compared to 2-2.5v times current for the classic bridge rectifier
  2. less capacitance required - since there's a much smaller voltage drop on the mosfets you'd need much less capacitance to keep the voltage above some threshold required to regulate the power supply to 13.8v
  
  It was already tested with up to around 30A : https://www.eevblog.com/forum/beginners/a-niche-chip-for-mains-powered-ac-gt-dc-'linear'-based-psu/
  
  You could for example give up on your custom transformer and just buy a couple of these : http://www.antekinc.com/an-8412-800va-12v-transformer/ and use one such IC + mosfets for two windings of each transformer (out of 4) , so in total you'd have 4 bridge rectifiers. You'd have a DC output with a peak of about 17v at the end of each rectifier, so you just need to add capacitance to stay above around 15v to then regulate down to 13.8v
• #22 Reply
  Posted by AF6LJ on 09 May, 2016 02:06
• Quote from: FJHookah on 09 May, 2016 00:36

  Quote from: Jebnor on 08 May, 2016 23:05

  With regards to your soft start option, take a look at Daves teardown of the Xantrex 300v 4A power supply.  It shows one soft start option in there.
  

  
  Thanks for posting the video... I like the extensive input filtering and will incorporate some of that straight away. Also, ah ha! They use a big resistor in parallel with a relay to control in-rush. I'm updating the schematic now.
  

  That is the Go-To method for inrush current limiting when large currents are involved. Suggestion on the resistor (since you are running this on 110V I assume you will only need one)
  I used these   www.ohmite.com/cat/res_200.pdfIn an inrush current limiter in my legal limit HF amplifier.
  They work great, very overload tolerant, and a heat sink is not required.
  They do have to be elevated above the board a bit, if you are using a PC board for your inrush limiter.
• #23 Reply
  Posted by sarepairman2 on 09 May, 2016 02:35
• another idea for you:
  
  in case you want to use that high power for other voltages or w/e, it might make sense to make two enclosures on a rack, one for the unrectified DC and one as the step down power supply.
  
  that way you could do things like try to experiment with designing a step up supply, or have different voltages, or whatever.
  
  design would also be a bit less stressful this way, since you could have the arguably simpler rack done, with AC MAINS considerations in mind, then design the other rack with DC practice in mind.
  
  just have some bus bars go up to the smaller form factor regulator...
  
  the filter and transformer are really the mass bulk of the design.
  
  additionally not having these elements in the same chassis would allow easier airflow/thermodynamic design of the regulator section.
  
  If you want it very sleek you could probably make the regulator section slim, then have a single squirrel cage/enclosed style fan blowing over separately ducted transistor row(s) (arrange them like a army parade maybe?) .. put nice filters on the fan too.
  
  this would allow you to optimize airflow as you don't have to deal with big capacitors and shit getting in the way
  
  you don't have to build an iphone
  
  I find it very soothing psychologically to divide up a project into different usable components. I hate the idea of having some kind of monsterous workload before I can get something remotely usable... since if you build the unregulated DC box you could hook up something simpler/etc to it.  beastly inductive heater... spot welder.. electrolysis... furnace..  etc. unregulated supplies have tons of uses because you don't need to worry about doing anything more then blowing a fuse.. and if they go its a quick fix. You don't need to worry if a inductive kick weaked all your regulator transistors or such . doing any kind of electrical energy experiments with a regulated power supply makes me nervous.
• #24 Reply
  Posted by FJHookah on 09 May, 2016 03:19
• Found a solution for the inrush resistor sitting in a parts bin here at the shack...