How to calculate bias resistor values for TIP2955 as pass transistor on an LM217?

Discussion in 'General Electronics Chat' started by Evil Lurker, May 11, 2012.

  1. Evil Lurker

    Thread Starter Member

    Aug 25, 2011
    117
    23
    Ok so I'm working on a junk box parts linear voltage regulator. So, I got myself a LM217 (not a typo) and a TIP2955 which I intend to use as a pass transistor. Both IC's are mounted on a decent sized custom drilled by yours truly original pentium heatsink and isolated with standoff washers, sil pads, heatsink compound, and all that good stuff with a mini dustbuster fan as a cherry on top. Basically the circuit I had in mind, or at least the first part of it will look like this:

    [​IMG]

    So here is the catch... eventually I plan on driving it off a boat anchor of a linear transformer, but since I don't have anywhere near the obscene amount of capacitance needed for an appropriate sized input filter I'm stuck running a 4 amp 20v rated switcher (formerly a laptop charging brick). I figured I might try and double duty the regulator circuit as a battery charger but should I try and draw anything more than around 4 amps out of my brick the overcurrent protection kicks in and *click* it shuts down. Therefore I need current limiting. As I understand it, by changing the value of the resistors on the front end I can limit the amount of current the TIP2955 will pass in addition to the LM217's output, the idea being to mostly rely on the LM217's built in current limiting feature.

    Where can I find this magical mathmatical formula?

    Second question... I need a filter cap on the output but I'm fresh out of 1uf tannies. I do happen to have some very very nice SMD tantalum 10uf 35v special "high surge" rated caps with a ridiculously low ESR... like comparable to OS-CON polymer caps with 10x the capacitance. Think these would cause instability issues? I'm also debating on putting a 10uf 50v regular electrolytic in between the output and adjust pin as in the "improved" ripple rejection schematics floating about.

    Thoughts?
     
  2. #12

    Expert

    Nov 30, 2010
    16,257
    6,757
    Long question, long answer.
    Capacit the output to ground all you want. It won't hurt anything. However, capaciting from the output to the adjust pin is a backwards move. It will cause the chip to go the wrong way when a sudden load is attached.

    I'm not impressed with a dustbuster fan for a heatsink. I don't think they are as reliable as the computer fan that came with the heat sink.

    The sucky part is that trying to do a current limit with the input resistors is temperature dependent. The one ohm resistor in the drawing says, "When the 217 draws about .5 or .6 amps, start using the 2955 to supplement the current, but limit that with the 10 ohm resistor. You can adjust the 1 ohm resistor and/or the 10 ohm resistor to get what you want because the vbe of the 2955 will go up as the current increases but it is temperature dependent. The hoter the transistor gets, the better it will conduct. When you need it the most, it will walk toward higher amps.

    A better way to limit current is to put a series pass resistor in the ground side and have that turn on a small npn that dumps the adjust voltage at Pin 1 of the 217.

    Got it? or do you need a drawing?
     
  3. dataman19

    Member

    Dec 26, 2009
    136
    29
    Try these two references...
    The first is a page about the LM317 (yea I know it isn't using the 2955 bypass transistors..
    ...
    http://www.phoenixcomputerlabs.com/All-About-Mixers/LM317.html
    ...
    The 2955 Bypass transistor operation is explained better in the second link (just scroll fdown to the regulator with the six 2955 bypass transistors and you will see a second diagram with voltage references)..
    ...
    http://www.phoenixcomputerlabs.com/All-About-Mixers/7812-7912.html
    ..
    Hope this helps...
    ..
    Dave
     
  4. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    Forget it, the current limit on linear regs is very inaccurate (as much as 2:1 variations) also has temp dependency and input-output voltage differential.
     
Loading...