Pre filter and dissipation issues for an 8 amp regulator

Discussion in 'Power Electronics' started by coinmaster, Oct 13, 2016.

  1. coinmaster

    Thread Starter Member

    Dec 24, 2015
    350
    6
    I'm building a -30v series regulator which will be used to power 4 2 amp current sources but I have come across some practical issues.

    For starters I cannot figure out a way to pre filter the AC going into the regulator because I cannot find any lytic caps that can handle the ripple current. I either get large caps which sink huge ripple currents or I get small caps and use a big RC resistor which drops the voltage to nothing with amps of current. There are no cheap inductors that will work and even if they did I would have to use large caps anyway.

    I also have a question about heat dissipation. The entire setup will dissipate about 936W across 18 mosfets, 24 if I decide to go with separate regulators per current source.

    I currently have a 12 inch by 12 inch alluminum heatsink that has thick 1.3 inch deep fins throughout the surface area on one side. I purchased it here http://www.heatsinkusa.com/12-000-wide-extruded-aluminum-heatsink/

    How much heat can I expect this thing to dissipate?
     
  2. Papabravo

    Expert

    Feb 24, 2006
    10,152
    1,793
    What is the ripple current and why do you think that garden variety electrolytic capacitors cannot handle the job?
     
  3. coinmaster

    Thread Starter Member

    Dec 24, 2015
    350
    6
    Let's assume I went with a crc filter with 10,000uf caps and a 10 ohm resistor. The first cap would take in about 30 amps ripple current.
    Sadly That's about as far as it goes for capacitor size because any less and the resistor wattage and voltage drop will be too intense and higher capacitance will draw even more ripple current.. So you see my issue. Unless LTspice is lying to me that is.
     
    Last edited: Oct 13, 2016
  4. Papabravo

    Expert

    Feb 24, 2006
    10,152
    1,793
    No -- I don't see your issue at all. I guess you'll just have to explain in greater detail.
     
  5. coinmaster

    Thread Starter Member

    Dec 24, 2015
    350
    6
    Only film caps can get in the realm of 30 amp ripple current rating.
     
  6. Sensacell

    Well-Known Member

    Jun 19, 2012
    1,132
    267
    I think part of the problem is your concept of 'filtering' in a DC power supply, there is no need for any series resistors, don't use them.

    Look at any basic schematic for a linear power supply, there are never resistor in series with the filter caps, except perhaps themistors, to limit inrush current.

    The DC from the rectifiers directly feeds large capacitors, don't think of this as an RC filter, but as an energy reservoir.
    The job of the regulator is to compensate for the 100/120 Hz ripple on the reservoir caps.

    As for heat sinks, look for the C/W rating of the heatsink, that's the degrees C per watt.
    At 900 watts, you should consider forced air cooling, this method can increase your safe dissipation dramatically.
     
  7. coinmaster

    Thread Starter Member

    Dec 24, 2015
    350
    6
    I tried using an LC filter instead of a CL filter in LTspice and it cleaned up the excessive ripple current nicely. I guess I should have thought of that earlier. Now I just need to find a sufficient inductor that can handle more then 2 amps.

    What is a safe thermal derating for inductors?

    The rating on the heatsink I have says C/W/3": approximately .85

    So with a 12''/12'' heatsink that makes 144 square inches which should mean .016C per W right?
     
  8. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,292
    1,255
    A schematic of what you have so far might be in order.
     
  9. Sensacell

    Well-Known Member

    Jun 19, 2012
    1,132
    267
    I think it means .85 degrees / Watt per 3" slice.
    That suggests 0.212 Degree per watt total @ 12 inches.
    900 Watts X .212 = toasty 191 C temp rise. Gonna need some serious fans.

    A choke rated for 8 A is going to be a MONSTER and you don't need it.
    Fix up the current sink design to provide tighter regulation, it can remove the ripple as long as you have enough headroom. Is it really a current sink? I don't see a sense resistor?
     
  10. coinmaster

    Thread Starter Member

    Dec 24, 2015
    350
    6
    It's not 12 inches though it's 12 by 12 so 144'' / 3 = 48 .8 / 48 = .016c which should mean 14.96c increase, right?

    I didn't post the current sink, just the regulator. I don't know how to make a current sink that can reject nearly the entire AC power supplies worth of ripple and still provide tight current regulation. @ 2 amps capacitors are not going to regulate well and they will sink huge amounts of currents which leads back to the original problem.

    Why would I need an 8 amp inductor? If I'm going to derate it wouldn't I want to derate it by x2 at most?
     
    Last edited: Oct 14, 2016
  11. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,292
    1,255
  12. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,292
    1,255
    8 amps X 30 volts is 240 watts. Where is the other 696 watts?
     
  13. coinmaster

    Thread Starter Member

    Dec 24, 2015
    350
    6
    Nevermind I got it figured out. Ronv is right, a bunch of high value caps in parallel does the job perfectly.

    In the current sources and the B+ regulator.
     
  14. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,292
    1,255
    I don't know what you are doing, but for what it costs to burn 600 watts I would do either a switching regulator or at least a tracking switching pre-regulator for the B+.
     
  15. coinmaster

    Thread Starter Member

    Dec 24, 2015
    350
    6
    Never heard of a switching regulator. Is it suitable for audio?
     
  16. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,292
    1,255
    Sure.
     
  17. coinmaster

    Thread Starter Member

    Dec 24, 2015
    350
    6
    Oh you mean a smps. Transient response too slow for audio :(
     
  18. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,292
    1,255
    By comparison audio is slow. Is this another one of your tube bias schemes?
    Don't forget the other half of the sentence.
     
  19. coinmaster

    Thread Starter Member

    Dec 24, 2015
    350
    6
    You're right. I just found out smps is only considered slow in HF applications. This is both a really good thing and a really bad thing because now I need to learn how to design smps which is very difficult from what I hear :( This is gunna be lame.
     
    Last edited: Oct 16, 2016
Loading...